Advantage Shopping Campaigns in 2025
How much we're spending across our clients, comparative CPAs, and the differences in net new visitors
Back in May 2023, I wrote a piece on how we’d been adopting ASC at Ballpoint.
We’d been testing it for the year leading up to it, and the headline conclusions were positive.
Where we were running ASC it was delivering on average a 33% reduction in CPA.
We’d run these tests in a variety of ways including isolated A/B tests using Meta’s inbuilt Experiment tools, as well as looking at various like for like campaigns in aggregate.
For a long period, we’ve been highly bullish on ASC. With new accounts where it’s missing, it’s often one of the first things we introduce.
In early 2024, well over half of our overall spend was going into ASC with some clients close to 100% in dominance. It had been consistently proving out against Manual structures.
But then performance for ASC began to drop.
It happened slowly, and then it became impossible to ignore.
ASC weight across 10 key accounts
This is our % of spend in ASC across 10 key accounts.
The headline here is that the average now sits at less than 15%. We were above 50% this time last year.
This date period looks at the last 120 days of spend. For accounts 3, 6, and 7, we turned off ASC due to poor performance historically.
CPA differences for ASC vs Manual
If you ask why we’re not running that much ASC its because of this: only two of our accounts have a positive CPA compared to Manual.
And it should be pointed out those positives are minor: 4% and 9% are significantly lower than the 33% heady days of 2023.
The far bigger picture is one of negativity.
3 accounts had over double the CPA in ASC compared to their equivalent Manual setups.
Our Manual setup by default is:
18+
All placements
Excluding existing customer lists
No regional targeting
No LAL or Interest targeting
No gender targeting
While for ASC we always opt for 0% existing customers.
ASC performance by account size
Those 3 accounts where it’s more than 10% of spend are a bracket of monthly spend above the other brands.
In those instances, even when we’ve not got the performance in ASC, the reason we opt to push the campaign is to help diversify our ad mix. But that said, we’re frequently exploring experimentation where we go ‘all in’ on Manual or ASC again.
In our smaller accounts, more often than not, ASC is now heavily underperforming and so quickly becomes a media technique we’re either dropping or not prioritising as a test.
No, ASC doesn’t over-index on engaged users
One of the criticisms of ASC has always been that it drives a much higher volume of already engaged users, even if they aren’t yet customers.
As well as using the engaged users audience in Meta, one other indication of this is measuring net new visitors by campaign type.
Net new visitors is a measure of:
New visits / Total visits = New new visits
This is a useful metric because it reveals how well you are bringing in new potential customers to your website.
A low % shows that Meta is not finding brand new users for you.
The data shows no uniform pattern.
2 accounts see a minor (2.9% and 2.8%) improvement
2 accounts show negligible decreases (-0.6% and -0.9%)
1 account shows a bigger negative swing of -6.4%
If ASC was overindexing on engaged users, then we’d see the ASC %s much, much lower than the Manuals. Only one account shows a reasonable swing here, but there’s no clear pattern.
Why has the quality of ASC seemingly gone down?
ASC was heralded for many reasons.
It was introduced as the most AI-forward package in Meta’s suite of advertising tools when first launched.
It heavily simplified the media buying process. And as a result, people like me suggested that ‘the cost of media buying is going to zero.’
And it heavily reduced the amount of control we had over targeting customers.
While there were strong early performances with it, those gains do seem to have petered out now.
Why that is we’ll never know, a couple of my hypotheses are:
As adoption of ASC has increased, more below average advertisers have fed it with poor data
It’s encouraged more media buyers to approach Manual in “Broad” and “Consolidated” modes, meaning the data in Manual campaigns has improved
Whatever human signals ASC was originally learnt from were reflective of the covid-era boom for home shopping, and newer consumer mentality doesn’t respond in the same way
How Ballpoint is approaching ASC in 2025
We still test ASC in every account at some point. But our confidence that it can be a solid contributor now is much lower.
Of our biggest accounts, it remains a core part of that structure. Here we view it as a way to reach a different subset of an audience – recognising that it likely reaches a different part of that audience at a different time.
We also aim to view accounts holistically as a general principle meaning overall account CPA is more important than individual campaign CPAs. Just as overall ad set CPA is more important than individual ad CPAs. People do not see individual ads in isolation: every part of your ad account can contribute to someone’s user journey.
But if you have been default-ASC without checking it, maybe it’s worth a brand new retest. Exact-for-exact, brand new campaigns, isolate the test and run it for two weeks and report back on where you get to.