Let's TalkLet's TalkFacebook and Instagram advertising for online stores - creative volume, clean tracking and a testing cadence that finds winners before the budget is gone.
Creative is the entire game on Meta and everything else is administration. So we start there. Send us your product and your current best performer, and we will send back two genuinely different concepts - different angles, not two crops of the same image - so you can judge the thinking before committing to anything.
The failures we see repeatedly, and what each one actually costs.
Now that broad targeting outperforms interest stacks, the number of distinct angles you can test is what caps performance. Most brands still produce creative on request, which means one winner scales until frequency kills it and then acquisition cost climbs for a month while somebody commissions replacements.
Browser-based tracking loses a substantial share of conversion events, and the delivery system can only find more buyers like the ones it knows about. Accounts running on partial signal underperform for a reason that looks like a creative or bidding problem and is neither.
Warm audiences convert cheaply because they were already interested, and blending them into one return figure makes prospecting look better than it is. Brands regularly discover, when the split is finally reported, that cold acquisition has been unprofitable for months behind a healthy-looking blended number.
Feed errors, out-of-stock items still serving, missing product identifiers and mismatched pricing degrade shopping placements silently. There is no alert, the spend keeps flowing, and performance drifts down in a way that gets attributed to the market.
The real deliverables, not a list written to make a proposal look thicker.
Conversions API server-side, with the events verified against what the business actually records. Optimisation cannot outperform the data it receives.
Separate reporting so warm traffic never flatters cold acquisition. This is usually where the real performance becomes visible for the first time.
A fixed weekly cadence of new angles rather than requests, so fatigue is a scheduling matter rather than a monthly emergency.
Feed health monitored continuously rather than checked at setup, because shopping surfaces degrade without telling anyone.
Budget follows new-customer acquisition cost against contribution margin, not blended return on ad spend.
A skincare brand from no paid history to seven figures in nine months, built on the approach this page describes: broad targeting carried by creative volume rather than interest stacking, server-side conversions from day one, and prospecting reported separately so nothing was flattered by warm traffic. Acquisition cost fell 63% in 90 days at 8.4x blended - with the new-customer number healthy underneath, which is the part most accounts cannot say.
Read the full case studyBoth of the free concepts beat the ad we had run all year. There was not really a decision to make after that.
They set up the Conversions API properly and our reported ROAS dropped. It was the first honest number we had seen and it changed how we budgeted.
Creative volume was the unlock. We went from two new ads a month to eight a week and the account stopped fatiguing.
Because Ads Manager credits itself generously and, without server-side tracking, is working from partial data. We rebuild tracking first so that the number you optimise against is the number that pays wages.
More than most brands are producing. On a healthy account we are testing several new concepts a week. Accounts that fatigue are almost always accounts starved of new angles.
Where it works, yes. But product photography built for a website rarely stops a thumb. We usually mix it with UGC and motion, which is why creative production sits inside the same team.
Yes, if the margin supports paid acquisition. If it does not, we will say so - scaling ads on a product that cannot absorb the customer acquisition cost is an expensive way to find that out.
Rarely, and this is the biggest change in the platform in years. Broad targeting with strong creative now beats narrow interest stacks in most accounts, because the delivery system finds buyers better than a human guessing at interests. What that means practically is that creative volume has replaced audience research as the constraint - and most brands are still staffed for the old job.
Yes, and not as an optional extra. Browser-only tracking loses a meaningful share of events to privacy controls and ad blockers, which starves the optimisation of the signal it needs to find buyers. Server-side sending recovers most of it. We verify the events arriving match what the business records rather than trusting the setup - a Conversions API that is configured but sending the wrong event is worse than none, because it is confidently wrong.
Written for owners weighing this up, whether or not they work with us.
This is one part of a bigger service. Here is the whole of it, and the closest neighbours.
Tell us what you are running today and what is not working. We will tell you honestly whether meta ads for ecommerce is the right thing to fix first - and if it is not, what is.