Ecommerce Attribution After Cookies: A Practical Model for Shopify Brands

Add up the conversions Meta, Google and TikTok each claim for last month. Now compare that total with the orders in your Shopify admin. For most mid-size stores the platforms are claiming well over 100% of the sales that actually happened. Everyone knows this. Budgets still get set on those numbers, because nobody has offered a workable ecommerce attribution model to replace them.

This article is that model. It is built for a Shopify brand spending between roughly £20k and £500k a month on marketing, run by a founder or a small team, without a data science hire. Four layers of measurement, stacked in order of how much you should trust them, and a weekly report you can actually keep up.

Why platform attribution broke, and why cookies are only part of it

Safari has blocked third-party cookies by default since 2020, and Firefox followed. Chrome kept threatening to do the same, then in April 2025 Google dropped the plan and left third-party cookies switched on. So the cookie apocalypse half happened. A large share of your UK mobile traffic is invisible to cross-site tracking, and the rest is gated by consent banners under UK GDPR and PECR.

The deeper problem is not technical. Each ad platform marks its own homework, with its own window, its own view-through rules and no view of the others. Server-side tracking recovers some lost signal, and we covered the setup in our guide to GA4, Meta CAPI and Consent Mode v2. It does not fix the incentive. A platform that reports fewer conversions gets less budget, so no platform will ever report fewer conversions.

Layer one: blended measurement and MER

Start with the number that cannot lie to you. Total revenue divided by total marketing spend is your marketing efficiency ratio (MER), sometimes called blended ROAS. Spend £60k across every channel in September, record £300k in Shopify sales, and your MER is 5.0. No pixel or consent banner can distort it, because both inputs come from systems you control.

Two variants make it sharper. New-customer MER divides new-customer revenue by total spend, which matters because many platform-counted conversions are repeat buyers who would have come back through email anyway. Contribution margin MER swaps revenue for gross profit after product cost, shipping and payment fees, and tells you whether growth is actually making money.

The trade-off: MER tells you the whole machine is working or not. It does not tell you which lever to pull.

Layer two: post-purchase surveys

A single question on the order confirmation page, "How did you first hear about us?", is the cheapest attribution data you will ever collect. Fairing, one of the larger survey apps, reports average response rates of around 54-58% for stores with more than a thousand survey views.

Wording matters. Ask about first discovery, not last touch, because customers remember finding you far better than the retargeting ad they ignored. Keep the list to eight or nine options plus "other", randomise the order, and include the channels platforms cannot see: podcasts, word of mouth, a creator, a physical shop. Put it on the thank-you page before the upsell widgets, and make it one tap.

Expect bias. Customers under-report paid social because a scroll-past ad does not feel like "hearing about" you, and over-report Google because search is where they went once they had already decided. Surveys are directional evidence for where demand is created, not a precise split.

Layer three: geo holdouts and simple incrementality tests

Incrementality asks the only question that matters: what would have happened if we had not run this? The cleanest answer without a data team is a geo holdout. Pick a set of regions, switch a channel off there, keep it running everywhere else, and compare the change in Shopify revenue between the two groups. The honest requirements:

Time. Plan for 4-8 weeks with no budget changes, no promotions and no new creative. Two weeks is usually noise.Holdout size. A 10-20% holdout gives enough signal without starving the business. Holding out Scotland against the rest of the UK is a common shape.Volume. As a rule of thumb, below about a thousand orders a month a geo test can only detect a channel that is close to useless.Patience. The paused channel will look worse in its own dashboard. That is the point.

Meta launched an Incremental Attribution setting in April 2025, and its own lift studies claim a 46% improvement when campaigns optimise for incremental conversions. Independent checks are more sober: Seer Interactive found that around a third of the conversions Meta reported would have happened without the ads. Use platform tools as a second opinion, not the verdict.

Layer four: first-party data and Shopify's own reports

Shopify's channel performance report credits orders to the last non-direct click within a 30-day window by default, and on Grow plans and above you can switch to first-click, linear or any-click. It relies on UTM parameters and referrer data, so it sees what a tagged link brought in and nothing more: no view-through, no cross-device, no offline influence. Shopify's help centre is straightforward about that.

It is good for three things: comparing channels under one consistent rule, catching UTM failures (a spike in "direct" usually means a broken link, not a surge in brand love), and seeing new versus returning mix per channel. Pair it with consistent UTMs on every email and every ad, and the retention view in our metrics guide. It is your cheap, always-on baseline. It is not the truth.

Be honest about the ceiling

Media mix modelling (MMM) is the grown-up answer, and it is out of reach for most brands reading this. It needs two or three years of weekly spend and revenue data, real variation in that spend, and someone to maintain the model. Below around £250k a month in media, the outputs are usually as uncertain as the judgement calls they replace, and the tooling costs more than the decisions it improves.

A brand at £50k a month should run MER weekly, a post-purchase survey always, Shopify's report as a sanity check, and one geo holdout a quarter on its biggest channel. A brand at £500k a month can add a rolling incrementality calendar, a lightweight MMM through a vendor or an open-source tool like Meta's Robyn or Google's Meridian, and a monthly reconciliation of platform claims against Shopify orders. Neither should pay for a tool that promises to "see the full journey". Nobody can see the full journey any more.

If the sum of your platform conversions is higher than your Shopify orders, your attribution is not a data problem. It is a decision-making problem dressed up as one.

A weekly report you can actually maintain

One page, one Monday meeting. Each number has a job, and a job it must not be given.

MER and new-customer MER (weekly and trailing four weeks). Drives the total budget: spend more, hold, or pull back. Not for picking winners between channels.Contribution margin after marketing. Drives whether growth is profitable. Not to be judged week to week during a launch or a sale.Survey mix (trailing 30 days). Drives where the next 10% of budget goes and what creative you test. Not a precise channel split.Shopify channel report, new versus returning. Drives UTM hygiene and retention spend. Not for arguing with a platform about its ROAS.Active test status. One line on the current holdout: what is being tested, when it ends, what decision it informs. Not to be read early.Conversion rate and AOV, by device. Drives the on-site work. This is the line most brands skip, and the one with the biggest return.

That last line deserves a word. In twelve years and over a hundred Shopify projects, the pattern we see most often is a brand fighting over attribution while its mobile conversion rate sits near 1%. Most Shopify brands do not have a traffic problem, they have a conversion problem, and a 20% lift in conversion rate improves every MER, every ROAS and every test result at once. That is why our Conversion Growth Retainer starts with measurement and then spends most of its time on conversion rate optimisation rather than on media.

The takeaway

Treat Shopify orders as the ground truth. Use MER to set the total, surveys to point the money, geo holdouts to check your biggest bets, and Shopify's report as your baseline. Ignore any tool that claims to remove the uncertainty, and spend what you save on the conversion and retention work that makes every channel look better. Email is the cheapest channel of all, and we set out how to run it in our guide to Shopify email marketing.

If you would like a senior second opinion on your measurement and your conversion funnel, from a Glasgow team working with brands across the UK and beyond, get in touch. We will tell you plainly what is worth measuring and what is not.