“Have you ever considered turning Meta off?”

I had a client ask me this recently. They’re a newish client, and we’re pre-product channel fit. I’ve been asked variants of this question for a long time.

While it’s a slightly hackneyed solution (which we’ll get into in this post today), the cause for it as an idea is a good one.

As marketers, only some of our job is the actual doing work of making great creative that converts, it is equally our job to prove its value as well.

Meta can be tricky. Sometimes it over-attributes, other times it under-attributes.

Today, I want to share a few scenarios you might be in, to help identify you true, or “incremental”, CPA.

Subscribe now

What you can expect today

1. Why Meta CPA is rarely what Meta says it is

2. How to measure true Meta CPA when:

   1. Meta is your first channel and so far your only channel

   2. You have grown your business through various other means, and you are now testing Meta

   3. You have scaled your business effectively with Meta but now know customers come from multiple sources

   4. You are running retention/remarketing activity

3. Our average adjustments to CPAs

Why Meta CPA is rarely what Meta says it is

Lazy marketers believe that Meta is out to get you.

I don’t buy that. They are of course a business that needs to make money. But there’s an equilibrium where return to advertisers needs to be substantial enough to continue to invest. My experience to date is always that the outcomes and incentives feel aligned for advertisers.

So if it’s not just Meta lying, then what it is it?

Problem #1: Data shifts post iOS14.5

Considering this change happened half a decade ago, it’s odd that this is still a discussion. But for anyone with a memory of advertising in the 2010s, they’ll remember how much more data flowed to Meta. Privacy laws, and proactive changes from Apple mean that’s no longer the case.

The result is less data in platform, showing lower numbers of attributions.

Problem #2: Attribution doesn’t reflect how people buy

The bigger problem is attribution. During the 2010s, we all bought into the lie of attribution. Attribution is a form of measurement that prioritises assigning a value to a channel. Google Analytics does this. Google Ads does it, so does Meta.

Attribution is a flawed measurement protocol.

It doesn’t reflect how people buy. More often than not many purchases these days go:

• See an ad → discuss with my wife → she or I purchase later on on a different device

Google likely steals that attributed sale, Meta often doesn’t get credit.

Or even put another way. Let’s say its a long consideration product, I see an ad, I sign up to the newsletter to stay on top of things. Then a few weeks later I decide to buy. Email captures the sale, but Meta caused it.

Attribution is a broken part of the picture full-stop.

For more on this, I recommend reading my startup guide to measurement.

Problem #3: Advertising to existing customers usually spells attribution issues

Problems 1 and 2 affect all marketing, but predominantly acquisition.

Retention marketing – ads spent on your existing customers – face a different problem.

Meta is only modelling behaviour. And with existing customers where there are more drivers of that purchase behaviour, its often less accurate - especially if you’re running viewthrough conversions with an existing audience.

Think of this scenario:

• I am a customer and in the morning on the way to work I see a Facebook ad in my feed

• Then I get your newsletter which I read and I go and purchase off the back of

• Meta quite rightly sees that a conversion happened after a view event and logs it

• But even without the viewthrough issue, similar scenarios happen with click-attribution too

Problems summarised

At a high-level we can often assume the following:

So what can we do about it

Subscribe now

How to measure true Meta CPA

Measuring Meta CPA depends on where you are in your journey and there’s a few key specifics.

For the majority of the time, there can be some form of lightweight modelled behaviour and so most of these won’t cost you anything. That is until your business gets more complex and you need to use more advanced statistical methods.

Here’s how you can measure the true incremental CPA for Meta based on a few scenarios.

Meta is your first channel and so far your only channel

For starters, this means that your buyers don’t have any issue deciding where to buy.

Every decision a customer has to make reduces their chance of purchase. Decision fatigue is real.

So too are distractions. If someone googles you and sees you’re also on Amazon, and heads there, it’s easy to get distracted with other things and forget all about your brand.

And another problem is price. If you’re on sale via third party retailers who cut costs, that eats into your sales. I’ve seen this first hand with a few brands and the brands ultimately closing off a lot of their distribution strategy to prevent the Google Shopping competition.

Simplicity is good. Single sales channels are good. Not only do they convert better, but you also retain all that data. DTC is not just a great way of acquiring customers – you acquire customers who can have way higher LTV because you can get to know them.

But single marketing channels are particularly good.

Why?

Measurement is simple with single channel

If you’re spending £10k per month on Meta, and you have no other marketing activity, then you know that your blend CPA is effectively your Meta CPA.

We call something like this the “halo multiplier” which is the halo impact of Meta impact / attribution issues. Maybe some of these show up as ‘word of mouth’ or ‘organic’ or ‘direct’ – but at the end of the day, they’re all being driven by your Meta activity.

Your CPA here is £22.22. Keep it simple.

Recommended Meta attribution window: 7day click, 1 day view

Subscribe now

You have grown your business through various other means, and you are now testing Meta

Now let’s assume you’ve come in from a different route. You’ve built a marvellous retail business and are stocked in Whole Foods or Ocado.

Things are great. Retail’s likely your background, or maybe you’ve built a great organic brand via SEO in a high intent area. Either way something’s going well, but you now need to think about DTC.

In this example, Meta is likely to go the way of a heavy ‘retention’ campaign and accidentally capture orders that don’t belong to it.

The reason being is that sales are happening truly organically at this point. Whether it’s SEO sales or those created by customers who have seen you in their local Boots or Walgreens, Shopify is going to be registering sales throughout the day.

Now Meta is very good at finding people to serve ads to.

And if you have a Meta pixel on your site, and a user visits it, Meta will almost certainly retarget them. Or in an equally believable world, Meta has good data on who your customers are and is already showing ads to the people who are about to become customers.

Customer purchases – through no influence of the Meta ad – but Meta claims the credit. Or, as with the retention example, maybe the person was browsing around with high intent. Didn’t make it all the way and then the ad nudged them.

Now in these examples, that’s not Meta driving that sale, though it might legitimately, based on your setup, lay claim to it.

Here you need to try to calculate the true CPA by looking at the incremental uplift.

Imagine this scenario:

Now in this case, looking at your blend CPA doesn’t help you.

You may find that a £24 CPA is great and so you start scaling, only to see it shoot up the month after, and again the month after.

Here we might try to calculate the incremental CPA using a ‘before and after’ test.

Why before and after tests are highly flawed

In the intro I said that turning your ads off entirely was a hackneyed solution. It is. And for the same reason as measuring incremental CPA as I’m about to describe.

Before and after tests assume everything else stays the same.

We manage 20 account across £20m in ad spend, and I can tell you that there is so much variance day-to-day, and week-to-week, that before and after tests are at best a single datapoint. Any change to product or website has a dramatic effect. The news has a dramatic effect. In short, it’s impossible to use a before and after test to truly predict what would have happened anyway.

But, it is a useful as a single datapoint in a period of discovery.

Let’s go back to our example. And now use our no spend months as a proxy for what would have happened anyway.

We measure the ‘incremental uplift’ as the difference between what might have happened and what did happen.

This gives you the incremental CPA.

As you can see this changes the picture dramatically. In fact the FB campaigns got better every month here, even though blend and FB CPAs got worse.

Now this rarely works cleanly, because more often than not you have months that look like this:

And as a result, it’s very hard to determine what a good ‘baseline’ is.

But as I say it’s a useful data point.

In times like this, your best bet is to switch up attribution.

Recommended Meta attribution window: Incremental attribution

Subscribe now

You have scaled your business effectively with Meta but now know customers come from multiple sources

Let’s say you went DTC first, or you’ve grown DTC substantially.

You’re now spending £100k/month+ across Meta and Google, and your physical retail popups are driving lots of sales in-store as well as online based on your post-purchase checkout survey.

Measurement now steps up a gear in complexity.

At this point, there’s two new tools that come your way for measurement.

Measurement tool #1: the marketing mix model (MMM)

Modern marketing mix models use machine learning, though the earliest ones predate ML practices.

In their simplest form they evaluate input metrics like channel spend and impressions, and output ones such as sales and revenue.

The model then attempts to determine the impact of a channel.

The cost of an MMM is usually high as you’ll need to find a data scientist to build one, or buy from one of the existing providers – all of whom are relatively expensive.

The downside to MMM is its a historic view, and usually requires weeks or months of data to examine at once. It’s not a daily, reactive view in most forms of the measurement.

Sidebar: we’re looking to launch a self-serve MMM for ~£399/month soon, register interest in comments or email if you want to be first to join

MMMs produce lots of output but at their simplest, they too can produce a simple CPA chart like we’re used to seeing. This comes from our internal tooling, data anonymised.

Measurement tool #2: a ‘lift’ or ‘holdout’ or ‘incrementality’ test.

MMMs are correlative, rather than causal. The only true causal measurement technique is the lift test.

Whatever you call this technique, it’s where you run a hold out.

It’s similar to what we did with our before and after test in example 2 above, but statistically rigorous.

You split your audience in half:

• Half get served ads

• The other half don’t

You look at the difference between the two.

These tests are expensive to run. They require running hold out audiences which minimises your advertising’s impact. They require making very few changes to setup, so if you’re having a bad month, you need to stick it out.

And they’re costly in other ways.

But they will give you the best outcome of answering:

“what is the true meta CPA?”

Recommended Meta attribution window: no set path, test multiple with your chosen measurement tools of choice.

Subscribe now

You are running retention/remarketing activity

Finally, the question of retention activity.

Retention marketing can both be highly cost effective, and a total waste of money. We have clients at both ends of the spectrum, and it really needs to be tested with your own data.

We highlighted above why this problem can happen.

Think of this scenario:

• I am a customer and in the morning on the way to work I see a Facebook ad in my feed

• Then I get your newsletter which I read and I go and purchase off the back of

• Meta quite rightly sees that a conversion happened after a view event and logs it

• But even without the viewthrough issue, similar scenarios happen with click-attribution too

How to solve for it

Once you get to a reasonably high number of customers, this actually becomes one of the easiest places to get a real statistical read.

You test this with a holdout / lift / incrementality test, as you would on prospecting activity in example 3 above.

However, you don’t need to invest thousands in this.

Our approach is:

1. Split your list in half and assign everyone an even split of cohorts

2. Upload and sync one of those lists to Meta

3. Expose ads to that audience for 4-8 weeks depending on list size and conversion rate

4. Measure the incremental uplift of the two.

I wrote about this in depth a while ago: you’re measuring retention all wrong – how to measure retention marketing.

We’ve run six of these tests over the last half year.

• 2 clients found it to be highly incremental and cost effective

• 1 client fund it to be incremental but not at a cost effective CPA

• The other 3 saw no impact of the activity

Recommended Meta attribution window: again it’s something to test, but in general, we’re more likely to recommend incremental attribution → click-only → click + viewthrough

Subscribe now

Our average adjustments to CPAs

Measurement is in some ways a luxury. In the early days its often left behind, and then investor pressure to hit this quarters’ goals usually stops it from becoming a regular occurence.

And so we have our own proxies that we often use as a standard baseline. These are the universal numbers which we recommend when fresh into an account.

• Acquisition we always expect Meta to deliver about 10-25% more value than it says it does.

• Retention we expect it to deliver 50% less value than it says it does (esp on 7dc1dv)

It’s really important to find out these baselines and proxies for yourself. Some of the swings are fairly substantial and so in a world where a 5% difference can be make or break, knowing for sure is important.

I hope you’ve enjoyed this. Measurement’s one of those subjects that can often get really complicated, but it needn’t be.

We’re experimenting heavily with ways of helping clients with measurement at the moment. If you’ve got questions and you don’t know your true Meta CPA, then get in touch. If not, please leave a comment below. Thanks.

I’m building Ballpoint: the AI-native growth agency I always wanted to hire when I was a DTC founder and before that a head of growth. We have scale brands from £1m → £50m through digital advertising. If you’re looking for support, then you can email me on josh@weareballpoint.com.

In the meantime, if you enjoyed this please consider giving it a like, a comment or forwarding it to a friend.

Leave a comment

It’s everywhere. X, LinkedIn, Substack, even the NYT got in on it.

The truth is nobody really knows what’s coming. But the possibilities are starting to feel genuinely endless.

At the beginning of March, I published Becoming AI Native, this was a blueprint for what I thought it meant for businesses to be AI native. My thinking has evolved since that article.

In the article my starting block was: ‘we get AI to do a task instead of us doing it ourselves.’

Two weeks after I published it, I led an internal all-hands stating we’d be pivoting the business to be AI-native. We’re now a month in.

Today, my view of being AI native is broader. Yes, it’s getting AI to do work for you. But it’s also being a systems thinker, it’s being a software developer, it’s a total revolution in how your organisation collaborates, and it’s a total pivot (ultimately) in how you make money.

It’s working out where humans add extreme value and where machines can operate at infinitely more capacity than you. It’s actively developing critical thinking, taste (sorry Kyle Chaka), and interdisciplinary thinking.

Since my original posts, I have had a lot of people ask me about how we’re using AI – and for training and workshops. This is something we’re exploring, so do comment / DM if this is for you.

And so this is an update on where we are, focused on stuff that is high leverage and really helping us already.

I find this is also important to state. I feel like we’re 0.01% of the way to where we’re getting. This post is not about “we’ve replaced every human we ever need and can service £100m of business”, it’s an attempt at an honest reflection of where we are today. Warts and all.

What I’ll be covering today

• Key Claude foundations you need to build

• 15 Skills, processes, and software we’ve built

  • Setting up proper Claude md files and folder structures

  • Moving the agency to GitHub

  • Building daily podcast for the SLT with most important 5 mins from last 24hrs

  • Ad account discrepencies flagger

  • Conversational joining of Notion to Meta ads data

  • X Influencers weekly roundup

  • Chief of staff that is my new 'notes to self' organiser, prioritiser, delivery

  • Automated ads inspo database that collects from 4-5 different sources

  • CHJ: my management agent for staying on top of 121s, L&D, progress

  • Creator discovery and outreach agent

  • Self-serve Google Meridian

  • Finance project that connects Xero, creator costs, Toggl

  • Marketing Science education site for myself (Duolingo for marketing sci)

  • Pitch decks we edit in review meetings via voice

  • Cosgrove: our new account manager

• What we’ve got planned

I hope you enjoy

Josh

Subscribe now

Key Claude foundations you need to build

Before we get into what we’re building, a quick primer on the things you need to understand to make Claude Code actually work for your business. If any of these aren’t familiar, then use these as starting blocks.

1. CLAUDE.md

Every project in Claude Code can have a CLAUDE.md file. Claude reads it automatically every time it opens that project. This is how you give Claude persistent memory about who you are, what you do, and how to operate.

I’m going to go into a bit more depth on this below, but if you’re unfamiliar and not yet used them. Start with Anthropic’s guide on making useful ones.

2. Context engineering

Prompt engineering was the key thing of 2025, context engineering is the key thing this year. This is more important than writing a better prompt.

This is about how you structure information around your prompt: examples, constraints, formats, objectives…. context is the important thing here. Again Anthropic’s own guide is great.

Leave a comment

3. Skills

Skills are actually very simple: they’re just sets of instructions to be carried out in order.

In other words, they’re recipes. Reusable recipes or workflows that you might call upon again and again. This might sound simple enough, but there’s a real skill (no pub) in creating good Skills at the right level.

About 15 years ago I did Harvard’s CS50 and to this day remember the ‘making a peanut butter and jelly sandwich’ segment. If you’re stuck with Skills, watch this to remind yourself of how many steps you need to think stuff through.

4. MCP

MCP (Model Context Protocol) is how you connect Claude Code to your actual tools. Slack. Notion. Google Drive. Your CRM. Instead of copying and pasting context between systems, Claude can read from and write to the tools your business already runs on.

5. Claude Code

Whereas once I maybe had 20 or 30 chats started with ChatGPT everyday, now I barely ask one thing to Claude desktop’s native chat interface per day.

I’ve 99% migrated to Claude Code and in the command line. This isn’t true for the entire team yet, but it does feel like it’s everyone’s natural nth degree rather than just the choice for some people.

Leave a comment

15 Skills, processes, and software we’ve built that have revolutionised what we do

1. Claude.mds and folders to set up for success

This is one of those ones I skipped over for so long and is probably one of the single big unlocks.

Claude Code reads CLAUDE.md files from every parent folder it sits inside. So if you structure your folders properly, context travels between your folders by inherting info from the folder above.

At the top, my personal CLAUDE.md. I gave it my background, people in my life, my key projects, how I want Claude to work with me, and my skill level.

Below that, a Ballpoint folder with its own CLAUDE.md covering the agency itself: our positioning, tone of voice, how we work, the team, our clients.

We share the maintenance of this as a team via GitHub (more on that below).

Inside Ballpoint, a Clients folder with a CLAUDE.md that contains key client info: systems for how we structure accounts, Notion organisations and so on.

Then each individual client folder has its own CLAUDE.md with their context: brand, competitors, constraints, current projects.

What this means is when I open a project inside a client’s folder it’s pulling on:

• the client’s claude.md

• our clients’ claude.md

• the Ballpoint claude.md

• and my own claude.md

When I first read about people doing this stuff, I delayed it for ages becuase it never felt high-leverage. There’s a reason I’m putting it number one today.

2. Moving to GitHub

I’ve never liked Notion’s version control, or Google Docs’ version control. After three months on GitHub, I finally understand why engineers love it.

The first unlock was shared maintenance. We have long had spreadsheets that contain things like Meta account ids, and another spreadsheet for Notion client links, and another for copywriting best practices. This is now a universally shared set of documents – all via GitHub. It means if something changes it changes for everyone.

Not only that but our bigger projects now get worked on like software ones. One person builds, ships, another tests and feeds back, branches get merged via PR.

For agency work this feels like a revolution.

3. Daily podcast for the senior leadership team

Every morning at 5am, our senior leadership team gets an email with a 5-minute MP3 attached. It’s a podcast summarising everything that happened the day before — key client meetings, key team meetings. It pulls from Slack, Notion, Fathom (where our calls happen), and Granola, synthesises it, and sends us a daily update.

I listen to it on the way into work. As the company has scaled, there’s a lot I’d miss otherwise. This bite-sized brief is how I stay close to the detail without being in every room.

Stack:

• Claude

• Python

• Elevenlabs

• GitHub Actions

4. Automated ad accounts checks (Skill)

Checking ad accounts is an ongoing part of the job, and I don’t think handing it entirely to an agent is right. But headline checks now get done automatically as a starting point.

The Skill checks each account against reference CPAs from `clients.json` and builds a rolling memory of how things have changed day-to-day, flagging outliers (was this spend increase actually big for Tuesday?).

Perhaps most importantly, it doesn’t send anything if there’s nothing to report. Instantly, it’s infinitely more useful than a daily report that you instinctively ignore.

• Claude Skills

• BigQuery & Gcloud’s CLI tools

5. Joining Meta data with Notion links via the CLI

We’ve long used Weld, which has a great internal SQL AI tool. In the later days, I used Claude to help write the SQL for me.

Now we’re connecting to BigQuery directly from the command line, which lets us join data sources on the fly. Because Bigquery sits alongside the Notion MCP, it means joining data together is done in human-speak rather than infra + data eng speak. It means that with some simple setup anyone can chat about any client’s data and it provide proper understanding, without the need for a line of code.

• BigQuery & Gcloud

• Claude Skills

6. Staying on top of X without getting sucked in

I used to waste a lot of time on social media following people sharing hacks and tidbits. It’s very easy to get sucked into the DTC timeline.

Now I have a handful of key people I care about. Every Monday morning, a Skill summarises their posts from the week, highlights anything new, and shares the original post alongside. If there’s a spike in conversation volume around a specific topic, that gets flagged too. Apify does the scraping as part of the Skill.

• Claude Skills

• Apify

7. A Chief of Staff that is my memory & coordinator

The problem I kept hitting: I always have ideas. At any given moment I’m jotting something down, forwarding an email to myself, or voice-noting a thought. I never build the time back in to go through it.

So I built a chief of staff. It ingests from a few places — Telegram API, an email subfolder (`josh+x@weareballpoint.com` goes straight in), voice notes. Whatever I share, it reads: if it’s a YouTube link, it pulls the transcript; if it’s an article, it reads the article. It’s logged into The Economist, FT, and The New Yorker, so paywalled pieces work too.

Then it decides what needs my full attention versus what can just be summarised and filed as a data point. It also acts on things. “Get a meeting with so-and-so” → it books the meeting. This is one of the areas I’m developing most — it’s starting to connect to my other Skills so it can trigger them automatically.

• Claude Skills

• Telegram API

• Apify

• GitHub Actions

• Claude Schedule

• Firecrawl for YT transcriptions

8. Auto-tagged ads inspiration database

Tools like Foreplay exist for this. But the reality for us is that ad ideas get shared in fifteen different places: Instagram DMs, WhatsApp groups, Slack channels, email etc etc.

What we now have is a centralised inspiration database that scrapes all of these sources automatically. The actual ad creative (video or image) is pulled via Apify and stored. Everything is auto-tagged by creative type, so we can pull references back whenever we need them.

• Claude Skills

• Apify

• Notion

9. CHJ: my agent for managing

Named after Claire Hughes Johnson, who wrote Scaling People (one of my favourite books on management).

This is a management agent that holds my team’s progress, notes from 121s, the key projects different team members are working on, and non-work projects they’re collaborating on. It has our growth and progression frameworks built in, so we can stay on top of how people are progressing. It also tracks things like team energy which is one of our internal benchmarks.

• Claude Skills

• Fathom API

• Granola API

10. Creator discovery & outreach agent

Finding creators is human-intensive and time-intensive — a natural place for AI to help.

The Skill is connected to Notion, where our briefs live. It pulls client characteristics and key customer notes from the brief itself, then searches Instagram or TikTok (depending on the client) against a tuned set of criteria. It builds a database of candidates, and via the Gmail API, it can start the outreach process too.

• Apify

• Gmail

11. Internal self-serve Meridian MMM

We’ve worked with data scientists for a long time to build MMMs with Meridian. For clients with a reasonable degree of similarity, we can now self-serve the Meridian rollout via this Skill.

It connects to our existing data warehouse, which makes it simple to run. We’ve also rebuilt the frontend in our own branding — a mixture of Astro and TypeScript — so the output looks like us, not like a research paper.

• Claude Skills

• Google Meridian

• Astro

• Typescript

12. Billing & Finance agent

We track time internally with Toggl, invoice via Xero, and have various ongoing discussions in Slack and Notion. This agent sees all of that.

I can ask how long a project took, whether we’re over-serviced on a client, how a retainer is trending. The answers are much richer because it’s looking across all the facets — not just the numbers in Xero.

• Toggl

• Xero

• Claude Skills

• Slack

• Notion

• Gmail

13. Marketing Science Education Site

I built this for marketing science specifically, but it’s the kind of thing you could do for anything.

I find reading isn’t actually how I learn best. So I built a simple spaced-repetition learning app to teach myself marketing science properly. It has an education layer and a revision layer. Essentially Duolingo for marketing science.

• Javascript

• HTML & CSS

14. Pitch decks we update via voice in the review meetings

Nobody likes building pitch decks.

Ours are now built in HTML, CSS, and a little JavaScript. There’s a uniform design system across every deck. Core page formats are specified. Videos and images slot in as examples. It means zero misalignment in design principles, and everything follows a proper hierarchy.

The payoff: we edit decks conversationally. Wispr Flow open, walk through the deck almost live, record notes, hit enter. By the time I’ve finished talking about the current slide, the previous slides have already updated. Deck creation has gone from hours to minutes.

• HTML, CSS, and JS

• Claude Skills

• Wispr Flow

• Gmail for context on clients

• Analysis subagents

15. Cosgrove: our meeting note account manager

Transcription notes suck. And the reason is not because of AI but because of context.

It doesn’t matter whether you’re using Fathom (like we do) or Granola (ditto) or any of the others, they are still universal tools. Now, what they’re great for is transcribing and managing that process – that said we may well one day replace that too.

But the important thing is adding in context.

One of our most recent internal tool releases is Cosgrove (named after Ken Cosgrove, Mad Men’s affable account man). This is trained on a series of best practice human meeting actions Slack messages, these were perfected over three years to get the right degree of actions, insights and notes – linked to all the right places.

This was a fun one to build, and while we’re not 100% at output yet – I’d imagine by the time this goes to print, we will be.

Subscribe now

That’s what we’ve built, here’s what we’ve got coming.

What we’ve got planned

As well as all of the above, we’re also currently building four separate bits of software that are combinations of code + skills + research + self-improving loops.

For the time being, we’re keeping these under wraps. Though as I joked last week, the possibilities are starting to feel majorly exciting.

Here’s a few areas where we still believe humans do the best work:

• Starring in UGC ads. While simple statics will arguably all be replaced by AI creative in the near-term, UGC is still something that works because of its authenticity.

• Customer interviewing and analysis. The nuance of how someone answers a question, the tone, the subconscious meaning because of how they smiled, and the ability to push harder in the interview itself, all mean that this is a skill that will forever remain human-focused. Maybe we’ll have AI moderators for our focus groups, but it won’t be any time soon.

• Creative strategy. By this, we should be mindful of the differences between (a) the deep thinking work of creative strategy, and (b) the executional work of writing a brief. Joining the dots together requires significant outlier thinking: not just following a set of prescriptive best practices.

• Strategy in general. With AI we could make 10,000 ads for you tomorrow. It doesn’t mean that would be useful. There still needs to be cognitive decision-making in the process, and humans are the best at working out what to say No to.

• Client servicing and deep communication. Not all client servicing requires humans. There is lots of factual information that doesn’t need context, and soon I imagine our Ballpoint AI talking directly to the client AI to provide updates for immediate status reports. But the human time as we go through decisions, thinking, context, as we try to get people aligned and incentivised right – I can’t see this going away any time soon. Not just this, but the human-to-human sense that we’re in these trenches together as we are with our clients is still the most valuable work we can do.

For me a lot of this is resetting my first principles.

I’m asking questions based on:

• If we ignored everything we knew about growth up until now and started with a blank piece of paper and Ballpoint, how would we design that system?

• What is software, and what is a skill?

• How can we build genuine self-improvement loops into our AI work so we can start seeing exponential improvements in output.

If I had to make a bet, then I imagine the way this will manifest is this:

This is the core bet I’m making at the moment and I’m deeply excited as we embark on this journey to get there.

It usually happens like this: performance has gradually been heading in the wrong direction. First, you rightly ignore it as these things usually work themselves out. Then it starts to become a trend but you still allow it. And finally, it becomes something you need to pay attention to.

At this point, the usual run of experiments won’t cut it.

I re-read Ben Horowitz’ A Hard Thing About Hard Things last November and one of the best chapters remains Peacetime vs Wartime (I wrote about wartime before product-market fit three years ago).

What I’ve come to realise is growth fits into these two modes of peacetime and wartime as well.

Wartime growth is when you don’t have performance in the right place
Wartime growth is not having product channel fit 
Wartime growth is not having product market fit
Wartime growth is not finding marginal sales as you scale
Wartime growth is being unsure that your ads are actually delivering sales
Wartime growth is not breaking even in time for your cash flow – or ever
Wartime growth means your creative win rate is 5-10%
Wartime growth means brand guidelines must be ignored
Wartime growth you are holding spend flat while you figure things out
Wartime growth is trading off no statistical significance for big bets
Wartime growth is maniacal focus on getting CPA to a good place
Wartime growth is not having marketing calendars
Wartime growth is wholesale landing page swapping

Peacetime growth is when you are scaling
Peacetime growth is about relentless creative operations scaling
Peacetime growth is having a Job Story and angle that *works*
Peacetime growth is 80% finding new concepts to demonstrate that core JTBD
Peacetime growth is loosening efficiency so you can scale hard
Peacetime growth is aiming to triple spend in six months
Peacetime growth is measuring all creative by £ spent on it
Peacetime growth is allowing room for brand consistency and design
Peacetime growth is small, incremental CRO experiments
Peacetime growth is statistically sound data

• When you’re in peacetime, experimentation is how you grow.

• When you’re in wartime, experimentation is how you survive.

Every business starts in wartime, then if you’re lucky, you have some period of peacetime until inevitably you fall back into wartime again. And the cycle continues.

I’ve previously described these as ceilings and growth spurts in an S curve. But I’ve grown to realise a ceiling feels too gentle; these are cycles of peace and war.

Experimentation is always urgent because no matter what state you are in, it’s the way out. But during wartime, it requires that you go back to the drawing board and begin with first principles.

I write twice weekly here every week. This is the weekly Tuesday essay: a key learning from my 16 years experience growing consumer businesses. Every Friday is a weekend debrief: key stories, stats, and reads for the weekend.

Posts are (currently) free, but the archive is paid.

Subscribe as a paid member for full archive access now.

Get 50% off for 1 year

Diagnosing problems during wartime

I was auditing an account recently for a brand spending about £3m a year. During the early sales calls they mentioned who their target customer was the 50+ female. Hallelujah I thought.

The 50+ female is the holy grail for advertisers. The 50+ female still uses social, she over-indexes as a Facebook user (a platform still underserved compared to IG). She is both rich and part of a high earning household. And she’s the primary shopper.

But as we got into the account I pulled the breakdowns. (Sidebar, I can’t believe Meta still give us breakdowns, use them while you can).

2026, year to date, 35-44 is the biggest demo for spend, reach and purchases and not by a small amount either. Yes there’s a female skew, but it’s pretty marginal.

We extended the search back further and looking at the monthly trend saw their 55+ demo halve from 40% of spend to less than 20% over the course of 18 months.

We’re no longer in peacetime.

So what does experimenting in war look like?

Begin with my Hypothesis Matrix to analyse why something changed

A while back, I came up with a simple matrix for diagnosing big changes in performance.

It’s a two-by-two grid with Location along the x axis (External / Internal) and then Ownability on the y axis (Ownable / Out of our hands).

Key question above this grid – “Why has the audience shifted?”

In the external and out of our hands, I wrote down:

• Tax / budget concerns

• Less FB usage in category

• Meta algorithm shift

In external but ownable, I wrote:

• New competitors

• Run out of customers

• Shift in social media trends meaning creative no longer works

• Shift in social usage to different placements

In external and on the border I put:

• Culture shifts away from product category

In internal and out of our hands, I put:

• Price increases last year

In internal and ownable, I wrote:

• Creative

• Campaign setup

• Any brand changes, rebrand, redesign etc (?)

• Landing page / website changes (?)

• Messaging changes (?)

• Email onboarding flows (?)

In the early days of my career I was in a meeting one August when someone offered up “it’s summer” as an answer to why performance was down. Said colleague was quickly dressed down by the CEO for using the external factor.

That stayed with me ever since. The tone was obviously curt, but beneath it was an important message. The mindset of diagnosing problems should always start within.

And so whenever I fill in my Hypothesis Matrix, I always start with the top-right: internal and ownable. And while I do also write down client-side changes in that quadrant, I always begin by listing out stuff that we, Ballpoint, own first of all.

Under creative, we can start digging even deeper.

Is it because we’ve stopped producing a certain visual hook? Or maybe all of our videos have switched to 15 seconds whereas this audience likes 45? Have we changed the type of creators we use? And so on and so on.

Once I’ve got my initial draft done, I’ll go to the client and ask for what they think it could be. I’ll begin by talking through the Ballpoint-ownable hypotheses we have, before going into some examples of others we had.

Share

Data gathering and experimentation

Now, we switch into the next phase: data gathering and experimentation.

Another peacetime/wartime distinction:

In peacetime, getting insights and gathering data is usually about finding extensions of what you know. You might dig through Reddit to unearth a new framing of your problem. Or maybe a customer interview revealed a slightly different catalyst for your Job Story.

In wartime, you need to go deeper.

Every hypothesis on the matrix requires some digging.

At this point, you need to apply pressure and fast. You don’t have time to analyse everything. Fortunately with Claude, much of this data is quicker to get than it was a year ago. But still, there is such a thing as too much data.

Example data explorations

Back in Jan, we saw damp performance in some accounts. I blogged about some of those findings, in particular that health behaviours had changed significantly this year. Dry Jan isn’t a thing now, protein for women is on the rise, and we’re moving towards strength over skinniness as the ultimate in high status.

We might explore if competitors have started using big influencers or launched a TV campaign, or perhaps they’ve launched an offer that’s just dominating.

We dig deep enough to form mini-hypotheses at an experiment level. And then we begin creating experiments around those ideas.

Prioritising experiments with Impact-Probability (IP) and Expected Value (EV)

At this stage, I like to evaluate expected values using our Impact-Probability (IP) framework. This is a simplified version of the infamous ICE frameworks.

1. Impact is the degree that change could shift your problem

2. Probability is what % likelihood you have that result will happen.

Let’s say your core metric you optimise towards is contribution margin.

Impact you should define as what potential uplift that would have in CM next month.

Let’s imagine one experiment hypothesis means you could double your conversion rate meaning based on next month’s forecast you increase contribution margin by £250k.

But the probability of that happening sits at 20%. This puts the expected value of said experiment is £50k.

Expected value (EV) = Forecast uplift in key metric * Probability of event occurring

Maybe you’ve got another experiment you’re debating running which can only generate £75k of CM gains but you’ve got 80% probability of success.

• EXP01 EV = £50k

• EXP02 EV = £60k

Start with number 2.

The last resort: owning the unownable

In grave situations, you may well exhaust all of the lovely top-right quadrant of ownable and internal problems.

It’s at times like this that you usually need a total strategy reset. Perhaps your core customer has really radically changed. Maybe you’re selling wine to an age bracket that thanks to a really big influential movement has gone 15% teetotal.

Out of our hands ideas can become ownable with a strategy shift.

And you can turn external factors into internal ones with a revolution as well.

But these changes don’t come cheap.

I hope you’ve found this useful.

I love experimentation. It’s one of our founding philosophies at Ballpoint. But it’s taken three years to identify these two distinct states of business. There is wartime and peacetime growth. And the model for experimentation during those phases is entirely different.

If you want help pinpointing whether you’re in wartime or peacetime, then drop me a line. Or better yet, start a discussion in the comments below.

Leave a comment

I’m building Ballpoint: the growth agency I always wanted to hire when I was a DTC founder and before that a head of growth. We have scale brands from £1m → £50m through digital advertising. If you’re looking for support, then you can email me on josh@weareballpoint.com.

In the meantime, if you enjoyed this please consider giving it a like, a comment or forwarding it to a friend.

For a long time, if you said to someone you worked in marketing, the immediate image was of someone planning monthly events, launching campaigns, and so-called activations.

For some sectors and companies sizes, this is a great strategy. Apple events are a series of magical moments throughout the year. The Superbowl most recently on our minds, is a singul…

Read more

The metric most DTC brands optimise for is actively working against their growth. Here’s what to focus on instead — and a free calculator to get started.

In my first head of growth role, I reported purely on CPA. Along with ROAS, it can be one of the worst metrics in performance marketing.

I didn’t know any better. I genuinely believed we were heading towards a world where platforms would have perfect attribution: where every touchpoint in the customer journey was tracked, and some multi-touch model would assign a precise value to each part of your marketing. For most of the mid-2010s, it looked like that was exactly where things were going.

Then we started testing new channels, and the real problem with this approach became impossible to ignore.

At the time, I could hide behind the fact that measurement genuinely looked like it was heading in that direction. But I really wish I knew then what I know now.

Later, when I became a founder, I understood this problem instinctively, but never properly articulated it. And that’s the issue for most brands: the language we use to talk about performance keeps the entire business stuck in the wrong conversation.

Here’s why ROAS is the worst metric in performance marketing, and the metrics to focus on that will help you grow (and your CFO will love you for it too).

Subscribe now

ROAS teaches you the wrong lesson

In the classic advertising sense, ROAS makes you focus on efficiency rather than effectiveness. That sounds like a subtle distinction. It isn’t. It fundamentally shapes every decision you make.

Efficiency is about minimising waste. Effectiveness is about your ability to achieve an outcome. And the strategic outcome we should all be striving for is growth.

ROAS makes you believe the goal is the most profitable return per customer acquisition. Actually, the goal is to generate more net profit for the business, and to do that you need more contribution margin, not a higher ratio.

Meta is a scale channel, not an efficiency one.

ROAS tricks you into thinking you can maintain high profit per order forever, but eventually you exhaust the low-hanging fruit. The retargeting audiences, the branded search terms, the people who were going to buy anyway. These are your easiest conversions, but they’re also the ones with the lowest incrementality. I.e., if you turned off your campaigns, you probably would have had much of the same result anyway.

Byron Sharp’s research makes this point clearly: most of a brand’s growth comes from acquiring the long tail of light buyers. It comes from people who don’t know you yet and aren’t actively searching. Reaching these people costs more per acquisition, but they represent genuine incremental growth. Incrementality research from Haus and others consistently shows the same pattern: the campaigns with the best platform-reported ROAS often have the lowest true incrementality.

Here’s the uncomfortable bit: ROAS is also easily manipulatable. I can give you brilliant ROAS tomorrow by shifting all your spend to branded search. You’ll have a few weeks of beautiful dashboards. But it won’t actually do anything for the business.

CPA sits on the other side of the same coin. It tells you what you paid per customer, but nothing about what that customer was worth, or whether the spend was genuinely incremental.

One important caveat. I’m not arguing that ROAS and CPA are useless. There are minimum thresholds you absolutely need. You should be at least breaking even, either on the first purchase or through strong repeat purchase economics. But above those thresholds, the conversation needs to shift from “how do we improve our ROAS?” to “how much profitable scale can we add while staying above our floor?”

Quick plug

I write a 7-minute read for Substack every week. We manage £20m of ad spend per year, I’ve helped grow eight consumer businesses from £0 to £10m, and have built creative engines that scale for 9-figure consumer brands. Before I ran Ballpoint, I was a DTC founder and ex-operator.

If you’d like the entire archive of our learnings, consider a paid subscription. You get:

• Entire archive of this Substack

• Invite to the Ballpoint Summer Party

• Early access to all Ballpoint resources, reports and downloads

50% off Annual today

What to measure instead: contribution margin

If ROAS measures efficiency, contribution margin measures effectiveness. It tells you how much cash your marketing is putting into the business after the costs that actually matter.

There’s some jargon to unpick here, but it’s simpler than it sounds. Think of it as three layers.

CM1 is your product margin. Take your average order value, subtract discounts, subtract your cost of goods. This is what you’d earn on every order if nothing else existed. It’s the starting point.

CM2 is your variable margin per order. Take CM1 and subtract the other costs that come with each sale — shipping, payment processing fees, packaging. This is the real margin you earn every time someone buys, before any marketing cost.

CM3 is your margin after marketing. Take CM2 and subtract your cost per acquisition. This tells you whether your marketing is actually generating cash for the business, per customer acquired.

Two numbers matter here. CM3 as a percentage tells you what proportion of revenue survives all variable costs including acquisition. CM3 in cash terms tells you the total gross profit your marketing is generating. This second number — total contribution in pounds — deserves more of your attention. Because just as ROAS can mislead you on efficiency, a high CM3 percentage can disguise the fact that you’d make far more money at a lower percentage but higher volume.

Here’s a simple illustration.

Brand A spends £50k per month with a 4x ROAS and generates £15k in monthly CM3. Brand B spends £150k per month with a 2.5x ROAS and generates £40k in monthly CM3.

Brand B looks worse on every ratio. It’s making nearly three times the profit.

Now layer in repeat purchases. If a customer comes back within six months and buys again — without you paying a new acquisition cost — that repeat CM2 is pure margin added to the original customer. What looked like a break-even first purchase suddenly becomes very profitable over a six-month window.

Measure this on a blended basis first, across all your marketing. Then model it per channel, so you understand where your next marginal pound is working hardest.

Leave a comment

How to calculate it

The maths isn’t complicated. What’s harder is getting all the inputs into one place.

You need your:

• average order value,

• discount rate,

• COGS per order,

• shipping cost,

• payment processing fees,

• total marketing spend,

• CPA,

• LTV over various time periods

I’ve built a free calculator that takes all of these inputs and generates your CM1, CM2, and CM3, then models five different scaling scenarios automatically so you can see exactly what happens to total gross profit as you trade efficiency for volume. It also includes an incrementality worksheet, so you can model what your real CPA might look like once you account for non-incremental conversions. Link at the bottom.

For brands doing this properly, the gold standard is a lightweight data pipeline that combines your COGS data, sales data, and spend data into one unified view.

It doesn’t need to be complex, even a well-structured spreadsheet can do the job at smaller scale. But having these numbers in one place is what unlocks the ability to make real decisions rather than reacting to platform dashboards.

How to present this to leadership

Once you can see contribution margin clearly, the conversation changes completely.

Instead of reporting “we achieved a 3.5x ROAS this month,” which invites the inevitable follow-up “can we get it to 4x?”, you’re saying: “We generated £85k in gross profit from £200k in marketing spend, and our six-month customer value suggests that figure will reach £120k.”

That’s a conversation about investment and scale. Not about squeezing a ratio.

There’s one question every marketer should ask their finance team: what level of gross profit do we need to cover our operating expenses?

When you know that number, you can model precisely how much marketing spend gets you there and all the trade-offs between efficiency and volume. This is the conversation your CFO actually wants to have. They don’t care about ROAS. They care about whether marketing generates enough gross profit to cover its share of the business’s fixed costs and contribute to the bottom line.

Stop reporting ratios. Start reporting cash contribution.

Subscribe now

Thanks

I’ve introduced CM3 to over 30 brands now, and it has been one of the biggest strategic unlocks in thinking through performance marketing. Remember, Meta is a scale channel not an efficiency one.

I’d love to hear your experiences with CM3 in the comments or if you want to email, hit reply and let me know.

Josh

Leave a comment

The Contribution Margin calculator

I’ve put together a contribution margin calculator as a Google Sheet. Plug in your unit economics and it gives you CM1, CM2 and CM3, plus five scaling scenarios that show what happens to total gross profit as you increase spend at different efficiency levels. There’s also an incrementality tab to model what your real CPA might look like when you account for the fact that not every platform-reported conversion was truly incremental.

Read more

I used to hate the word ‘incremental.’ I remember when I first heard it – from someone at an agency when I was in-house.

It felt like marketing jargon, and once I learnt why they were saying it, I thought it was redundant. “I know when we increase spend conversions go up” was my gut response.

Fortunately in the 12 years since that moment, my thinking has…

Read more

A few weeks ago, we hosted a measurement roundtable at Ballpoint HQ.

We had three speakers:

• Rich Cousins – director growth at Thriva, for whom we’ve worked on an MMM and various lift study experiments

• Sameer Modha – ex-Google, now helping provide measurement chops to telly ads

• Franky Athill – startup veteran CMO, and founder of Attribution Lab.

It was a one…

Read more

Lifetime value debt

I’d like to thank Beth Carter & Daniela Nardelli for help and feedback with these drafts.

I’ve been speaking to lots of CMOs and founders of brands doing £5-20m in revenue. They all describe variations of the same problem: acquisition costs that worked at £50k/month don’t work at £250k/month. Meta is more expensive. Conversion rates dr…

Read more

Two years ago, I wrote about when the time is right for brand marketing. It’s been one of the historic most important read topics on the substack. A few weeks ago I had a conversation with one of our creative strategists that provoked a rethink of this.

Here's what most people get wrong about brand marketing: they think it's about values, emotions, or creative storytelling. Those are just tools.

Two types of advertising (and only two)

Strip away all the jargon, and there are only two types of advertising:

1. Advertising that wants you to do something today - Click this. Buy now. Sign up. Visit our store this weekend.

2. Advertising that wants you to do something eventually - When you're in market six months from now, remember us.

That’s why we do advertising. When someone says “we want to increase brand awareness”, I ask “for what?” Usually they respond “so eventually they go onto buy.” Increased brand awareness is just a proxy for the eventual action.

So these are the two reasons we advertise: to create an action now or in the future.

That’s performance vs brand.

Direct response vs mental availability.

Call it what you want, but this is the fundamental divide.

The confusion comes because people conflate the how with the why. A John Lewis Christmas ad isn’t a piece of brand marketing because it’s emotional and weaves a story together. We’ve got scrappy UGC that does that too. It’s a piece of brand marketing because the desired outcome is – when you’re ready to shop, you remember them. That could be this Christmas or June next year or maybe in five years.

The mental availability game

Brand marketing is about being there when that changes. It's about mental availability - ensuring that when someone finally is in market, you're front of mind.

But here's where things get interesting, and where my thinking has evolved since that 2023 post.

Subscribe now

The "what" problem

In a recent conversation with our creative strategy lead, we kept circling back to a critical issue: a lot of brands want to be known for everything except the one thing that matters.

Imagine you’re the founder of PRO T, a new protein crisp company. You’ve just launched to market.

One brand desire for that company might be “we want to be associated with people who push themselves for greatness” or “be for the people who join the Sunday morning run club” You might create some organic content that starts playing in this arena. Maybe you shoot some content of people on a trail run before sunrise.

This might make nice content. I’m also certain that it will look beautiful, you can shoot it well, you’ll love the outcome and you’ll feel proud if it.

The problem is that the public don’t care about you. They don’t care about your brand. And people won’t dig deeper than what is being shown them in an immediate moment.

If you’ve really made good content, then maybe someone watching your trail running video remembers the name Pro T.

But as my colleague says, “if the one thing they remember about you is trail running, then the next time they’re in the market for trail running shoes, they might go “Oh isn’t Pro T the brand that does that?””

People’s attention spans are famously short. But we’re also bombarded with thousands of messages a day. Meta serve an ad every 2 or 3 placements. If you spend ten minutes a day on their platform, you could see 100-200 ads. You probably remember five of them. Then you go to work. On buses, you see more. On the tube you see more. Maybe you listen to Capital FM and hear more. Marketing is everywhere.

And people also have a million things going on in their heads. The report that’s late for work. The friend they haven’t replied to on Whatsapp and it’s now been five days. The wedding invite you’ve still not RSVPed to. You’ve not slept because you’ve got a five month old at home. What’s going to happen in the last episode of The Girlfriend? What am I going to eat tonight? Did I leave my hair straighteners on? The fact that your favourite song just came on shuffle and now you almost hop through the street. The glimmer of sunshine that hits the road just as you get to work.

People aren’t waiting to discover new things. No-one wakes up in the morning wanting to find a new brand.

And so if you’re lucky – and I mean really lucky – and you manage to create something that actually lands. Then the single most important thing that can land with them is a single sentence descriptor of the ‘what it is you do.’

In my original post on this, I interviewed Tom Hillman about his prior time at Wise, and he said: "No matter what else people knew about them, 'send money abroad' was the one thing they wanted to be known for."

Subscribe now

Why this matters more than ever

The temptation to be everything is stronger than ever. Every brand wants to own a lifestyle, a movement, a cultural moment. But mental real estate is limited. You get one spot in someone's brain, maybe two if you're lucky.

Think about the brands you know well:

• Volvo? Safety.

• Nike? Athletic achievement.

• Patagonia? Outdoor gear with environmental consciousness.

Yes, these brands stand for more. But they earned the right to expand only after owning their core territory so completely that it became synonymous with their name.

The channel confusion

Another evolution in my thinking: channels don't determine intent.

My creative strategist made this point brilliantly: "You can do both buckets on all channels. Some are better suited to one than the other, but you can do both on all channels."

Your Facebook ads can build brand. Your TV spots can drive immediate action.

At the Goodwood Revival - a vintage racing event where every detail transports you to the 1960s - they make all the stalls – Tesco included operate and sell as if its the 1960s. That drives short term sales and long-term brand equity.

The question isn't "what channel?" but "what outcome?" And more importantly: "what's the ONE thing we want them to remember?"

A framework for clarity

So how do you approach this? Here's my updated framework:

1. Define your singular territory
   Not your values. Not your personality. The one functional or emotional job you do that people should remember. For protein crisps maybe that’s “delicious AND healthy crisps for gym goers who need more protein”
   

2. Assess your mental availability needs
   Are there enough people in-market today that you can capture profitably? If yes, focus there. You can spend millions on advertising until you need to worry about mental availability.
   

3. Make every touchpoint reinforce that one thing
   This hasn't changed from my previous post: "Every performance marketing ad you run impacts your brand. Every email you send out. Every delivery that gets sent to a customer." These should all reinforce the one thing you want to be known for.
   
   Play one of your ads to someone who doesn't know you. If they can't tell you the one thing you do, you've wasted that impression: whether it was meant to drive immediate action or build long-term memory.
   

4. Resist the lifestyle temptation
   Your brand can have personality. It can have values. But these should amplify your core message, not replace it. The protein crisps can be on sale at a trail running event, for sure, but only if the primary thing they’re known for is being delicious AND healthy crisps for gym goers who need more protein.

Brand marketing: the time is now

Brand marketing isn't about feelings or values or storytelling techniques. It's about lodging one clear memory that will be there when someone needs it.

Everything else - the emotions, the creativity, the cultural connections - these are tools to make that memory stick. Use them. But never forget what job they're doing.

Because if you don't know the one thing you want to be remembered for, neither will anyone else.

Leave a comment

🔗 When you’re ready, here’s how Ballpoint can help you

→ Profitably grow paid social spend from £20k/m → £300k/m
→ Create full funnel, jobs to be done-focused creative: Meta, TikTok, YouTube
→ Improve your conversion rate with landing pages and fully managed CRO
→ Maximise LTV through strategic retention and CRM - not just sending out your emails

Email me – or visit Ballpoint to find out more.

NB: We support brands spending above £20k/month.

❤️‍🔥 Subscribe to our Substack to learn how to grow yourself

… because agencies aren’t for everyone, but our mission is to help all exciting challenger brands succeed and so we give away learnings, advice, how-tos, and reflections on the industry every week here in Early Stage Growth.

Leave a comment

There’s one key metric that demonstrates the overall health of your marketing. It’s not CPA, it’s not ROAS, and it’s not even CM3 by itself. It’s contribution margin per session.

I.e. for every user who lands on your website, how much profit do they give you?

TL;DR
Most performance marketers obsess over the wrong numbers. They'll tell you their 3.5x ROAS …

Read more

Growth strategies and marketing strategies are almost always plans, not strategies.

They are staggered timelines of when to introduce campaigns and channels. Maybe they feature some goal setting.

But they’re not strategies.

A good growth strategy includes two core elements:

1. Analysis of where the curre…

Read more

“The final form of a brand is decided by the public” – at VISIONS Summit

TL;DR: Brand teams can no longer control their narrative and story. Gen Z see through it, and the way we communicate today means strict controls and guidelines are irrelevant. The smartest brand teams are starting to take note.

One of my favourite things about long bank holidays is …

Read more

Last week, I shared a preview of one of our forthcoming reports about upper funnel spend, and Fiona shared her current AI creative stack. That forthcoming report is going to be one of the first fully paywalled pieces we launch here. If you’re not yet a paid subscriber, you know what to do.

In Early Stage Growth this week, I’m looking into what really constitutes a win on Meta?

TL;DR

It’s difficult to know what to do when a new creative test ‘wins’. Do you double down? Does it reduce all CPA? How long do those ads last?

In our end of year report this year, we want to go into this in a lot more detail, but in the short term we want to share short term learnings to begin the discussion.

Things we’ll be covering:

1. What is a winner? How do we test?

2. Why do ad accounts need so many ads?

3. What are the types of winners – rocketships, maintainers, false starts?

4. What is a good rate of winners to losers?

5. How can I improve that rate?

Here we go.

How do you define a winner on paid social?

There are loads of schools of thoughts on how to run creative testing on Meta. There are those who throw everything into one mega ASC. There are those who test thousands of ads into heavily bidcapped ads and are content with most ads getting zero spend. There are those who us the Meta Experiments tool for each test. Some isolate every single ad in its own ad set.

There’s no perfect way. Everything in paid social is about tradeoffs.

So to understand how we make our decisions, we should first set out what we’re trying to do.

Why we test like we do at Ballpoint. The background thinking:

1. Ad volume is important to scale
   I’ll get into this in more detail in section two, but at a high-level each ad is a chance to find a new audience. Therefore to find new customers, you need new ads.
   

2. You can improve probability of success through strategy
   We are not a ‘throw everything at the wall’ agency: there are lots of agencies that find success with that method. But we aren’t, we think that strategy delivers better long term performance. It does that by increasing the probability of finding a winner.
   

3. Ability to learn improves strategy
   Like any good growth process, we run insight → experiment → learning. Learning feeds the next insights. And so in our testing approach we want to optimise for ability to test.
   

4. We know that despite strategy, chance still plays a huge role
   Therefore, we need to create an environment that allows us to factor chance in. We need to make sure we’re testing ads that don’t just follow what we expect to work.
   

5. Diversity AND volume are therefore needed
   With that in mind, it’s not just about following what we know. We have to test lots that doesn’t fit a mould as well.
   

6. Pace is vital in startups
   The most likely outcome for a business is that it will die. You will run out of money before you work all this stuff out. And so pace is vital.
   
   Say it takes 100 tests to find the first answer. If you do 1 a week you need 2 years of runway to survives. If you do 3 a week, you need 8 months.

As a result of all of these things, we by default run tests:

• Ads grouped by concept at an ad set level (variants within it, usually 1-5 depending on size)

• Run on ABO. This is an intervention away from how Meta usually works by forcing it out of a CBO/ASC mindset. We opt to do that to improve our ability to learn (point 3)

• Don’t use Experiments tool and don’t isolate ads too much (point 6 would mean that these ad too much cost and slow down overall learning).

All testing is about tradeoffs. This is how we think about our approach.

So what about these winners?

We compare ads for similar like for likes in the hero/scaling campaigns.

For EG, say

• Ad set 1 has £44 CPA

• Ad set 2 has £88

• Ad set 3 has £56

• Hero campaign is at £55

Ad set 1 is definite winner, promote the best variant(s).

Ad set 2 is fail of concept, but check if a variant won

Ad set 3 is likely a moderator as within a margin of error. Promote best variant(s)

Subscribe now

Why do ad accounts need so many ads?

Meta has become increasingly good at pairing individuals with ads that are relevant to them. We think of this as ad-audience fit. A lot of this started last year, but Andromeda in particular improved this.

Everyone responds slightly differently to things.

There are some universalities. There are some ways of communicating which reach broad audiences.

But there is so much micro-nuance in our day-to-day.

We could have the exact same needs and problems to solve, but I just happen to hate black and white in imagery and you love it. Or maybe I just really hate jazz and so the background music made me swipe past, whereas you don’t care about it and ignored it.

Then there’s the fact that life changes daily. Maybe you woke up grouchy or your favourite F1 team just lost or your baby is teething or it got a bit dark and rainy when you opened Instagram and it’s supposed to be summer. There is no uniform human experience.

And so ad accounts need ads for all eventualities if you want to scale.

Let’s be clear, there are advertisers out there spending tens of millions per month on paid social. That’s what success looks like. If you really have product-market fit and product-channel fit, then that’s what your aspiration should be.

And part of getting there is ads for all permutations of all problems solved, all jobs to be done, all people, all stages of the funnel, all whims of their mood.

Subscribe now

What are the types of winners – rocketships, moderates, small contributors?

We’re increasingly adding nuance to these categories, but in general we’ve got the following groupings:

1. 🚀 Rocketships – ads which (1) heavily improve CPA compared to the account CPA, by 20-90% and then (2) go on to become the dominant spenders in the hero campaigns.

2. 🥈 Moderates – ads which sit within a 10% bracket either side of account CPA. They then go on to remain in the account for some time but rarely become the biggest spenders.

3. 🎗️ Small contributors – sometimes an ad will do well in testing, you promote it and it ends up spending less. Maybe it gets the odd purchase or two at a very low rate, or maybe low levels of spend and no conversions. These aren’t fails necessarily, just maybe their audience size is small.

What does good look like? What ratio can we expect of these ads?

The million pound question. “How many ads should we expect to be winners?”

Here’s a few key stats from a small data set of ours (blend across account sizes):

• Concept win rate ranges from 22% – 46% with evenish distribution

• At an individual ad level 49% of ads have better CPAs than the account average that week – but it’s closer to 30-35% that receive enough spend to be considered winners

• 2.3% of winners are rocketships this rises to 3.5% amongst bigger accounts – indicating higher success at determining the very best ads the more we learn

• 17% of winners scale well – i.e. they’re spending more in month two than month 1

• 40% of winners maintain the account – spend stays pretty consistent, indicating either incremental reach

• 40% of winners fatigue within 4-6 weeks – indicating low longevity

“Maintain” might look like a dreadful word.

And if you’re pre-product-channel fit then what you obviously need is rocketships and good scalers to be able to change your trajectory.

But post-PCF, you’re fighting multiple fights: creative fatigue (which maintainers fix), new audiences, and ability to scale (scalers & rocketships fix).

Note we will be re-running this analysis in much more detail for our end of year report, where we’ll be running this across £15m of annual ad spend.

How can I improve that win rate?

When we talk about doing paid social today, this is really what we mean. Media buying is still important, yes, but ultimately how well you do creative is how you improve that win rate.

Improving paid social today is about building a creative growth engine.

1. Start with Jobs to Be Done
   Your best ads don't come from copying competitors or following trends. They come from deeply understanding the progress that customer is trying to make in their lives.
   

2. Test with intentional diversity
   Each JTBD insight should create multiple creative angles. Test different hooks, formats, and emotional territories. Get rid of your own bias. Nothing kills an ad account faster than making ads look like only ads you like. “I like it” is not a good sign of an ad.
   

3. Document relentlessly
   It’s vital. But without good documentation, you aren’t improving the learning rate, or creating future scalability.

4. Ensure at least 25% of ads are illogical
   Chance still factors in a lot. Lean into it. Throw in ideas that are totally counterintuitive. Whether it’s a messaging line you don’t think will work or a different style of creator or editing style. Break your own rules.
   

5. Speed compounds learning
   You don’t go to the gym once a week and except to get a sixpack. Consistent, high volume is vital. As a startup you’re almost definitely losing money. Velocity is what will save your runway.

Final thoughts

The most likely thing an ad will do is not reach anyone. But it’s through that process that you learn how to reach as many people as your product will allow.

Diverse ads buy you new audiences. Better ads improve performance within those audiences. Both of these are needed.

What are your rocketship win rates?

Leave a comment

🔗 When you’re ready, here’s how Ballpoint can help you

→ Profitably grow paid social spend from £30k/m → £300k/m

→ Create full funnel, jobs to be done-focused creative: Meta, TikTok, YouTube

→ Improve your conversion rate with landing pages and fully managed CRO

→ Maximise LTV through strategic retention and CRM - not just sending out your emails

Email me – or visit Ballpoint to find out more.

NB: We support brands spending above £20k/month.

❤️‍🔥 Subscribe to our Substack to learn how to grow yourself

… because agencies aren’t for everyone, but our mission is to help all exciting challenger brands succeed and so we give away learnings, advice, how-tos, and reflections on the industry every week here in Early Stage Growth.

Leave a comment

TL;DR

Alongside its click and view-based windows, Meta has introduced “incremental attribution” an attempt to help prove what ads are actually driving performance. Based on our own incrementality data, we’re not yet sold.

Incremental Attribution got everyone excited a month ago when it launched, but now a handful of experiments in, it’s still leaving som…

Read more

Contribution Margin 3 (CM3) tells you how much profit you’re marking after marketing costs. Unlike CPA, or AOV, or ROAS – it helps us understand the only question that matters: are we making real money? This post explains why it’s important and how to measure it.

The problem with metrics today

Thanks to every Shopify plugin in the world, we’ve got m…

Read more

CM3 is the profit leftover from a sale once you’ve taken into account COGS (inc. fulfilment), card fees, discounts, and marketing costs. It’s the only true measure of how your conversion rate experiments have taken place.

Today we’ll be exploring why it…

Read more

In startups, your desired outcome is usually survival. In more established businesses, it’s how much more money you can make.

Since I closed my startup, I’ve audited, assessed, or coached over 200 businesses looking at growth, an…

Read more

Growth meetings are the beating heart of the growth process. It’s the way we run our weekly meetings at Ballpoint, and how I used to run them in-house before then.

My back of fag packet estimations tell me I’ve run 874 of these to date (875 by the time this email goes out).

Here’s the…

Read more

“That’s great,” I replied “so 60% of your customers you acquire repurchase the following month?”

“Errr.”

There's a pause, then some questions, as we get to the crux of how that’s measured.

“60% of our revenue last month was from repeat customers” the founder eventually tells me.

“OK that’…

Read more

January at Ballpoint towers means two core things: (1) reflecting and reviewing on 2024 data now that the year is complete, and (2) putting together strategy for the year ahead.

One core part of how we think about strategy is our growth models. I’ve been building growth models for 11 years now and they’re absolutely core to other growth team I was in in…

Read more