You Don't Own Your Attribution Data. Here's What That Costs You.

Quick answer. Ecommerce operators have internalised one lesson well: do not build your business on rented land. Own the store, own the email list, own the customer relationship, and treat marketplaces as discovery you rent. That lesson stops one step short. Your measurement is still rented — the numbers that decide where next quarter's budget goes live inside ad accounts you do not control, under rules the platform can change without telling you, on a history that disappears with the account. The owned version is a file you already have: your GA4 export, and a model that runs on it.

The lesson sellers just learned in public

In September 2026 a video from a 370-subscriber channel did roughly 37,000 views and 242 comments — a breakout of about 494x that channel's normal performance. It was not a tactics video. It was a 14-year Etsy jewellery seller describing how her shop was permanently suspended days after she came home from hospital, and how nothing she had built on that platform came back with her.

Her conclusion — own the foundation, rent the discovery — is not new advice. What is notable is the scale of the reaction. An audience of sellers recognised the shape of the risk instantly, because most of them are one policy decision away from the same story.

The three assets she named as owned: your own website, your email list, and the people who actually follow you. That list is correct, and it is incomplete.

The part the lesson stops short of

Ask an operator who owns their customer list and they will answer immediately. Ask who owns their attribution history and the question usually does not parse.

Here is what it means concretely. Your record of which channels drove revenue over the last two years lives in Meta Ads Manager, Google Ads, and a TikTok dashboard. Each of those:

• Is scored by the party being graded. Meta decides what counts as a Meta conversion. Google decides what counts as a Google conversion. Every platform grades its own homework, which is why the claims routinely sum to more than the revenue that actually happened. That is the mechanic behind last-click attribution and every platform-reported number built on it.
• Can be re-based without your involvement. Attribution windows, modelled conversions and consent handling change on the platform's schedule. When they do, your historical numbers change meaning underneath you. Nothing in your dashboard tells you the series broke.
• Leaves with the account. A disabled ad account, a closed business manager, an agency relationship that ends badly — the reporting history goes with it. You keep the spend. You lose the record of what it bought.

None of that requires bad faith from anyone. It is the ordinary condition of data you do not hold.

What owning your measurement actually means

The pattern is the same one the video argued for. The difference is that most operators have applied it to the storefront and not to the number that allocates the budget.

The GA4 export is the owned artefact

This is the part that is easy to miss because it is unglamorous. A GA4 export is a CSV. You can download it today, keep it, version it, and hand it to anyone. It does not depend on a pixel staying installed, an SDK staying current, an ad account staying open, or a vendor staying in business.

What it needs is a method that turns it into an answer. That is a different job from storing it:

• Platform-reported numbers tell you which channels were present when a sale happened, as judged by the channel.
• Causal attribution estimates the counterfactual — how much of that revenue would have happened anyway without the channel. It is the question a budget decision actually turns on.

Run the second one on a file you own and the whole dependency chain collapses to something you control. The export is yours. The period is whichever one you care about. Re-run it in a year and the same input returns the same answer, regardless of what any ad platform did in the meantime.

The cost of not doing this compounds quietly, quarter after quarter, which is the thing we call marketing debt.

FAQ

Doesn't GA4 have the same problem? Google owns it too. Partly, which is exactly why the export matters. While the property is live you can export any historical window as a file. Once exported, that file is yours and does not change. The dependency is on getting the data out, not on Google's continued goodwill about how to interpret it.

My ad account has never been disabled. Why should I care? The account-loss case is the dramatic one. The common one is quieter: attribution windows and modelling change, your history re-bases, and you compare this quarter against a number that no longer means what it meant. You do not need a catastrophe for the record to stop being reliable.

Is this an argument against running ads on the platforms? No. Rent the discovery — that is what the platforms are genuinely good at. The argument is about where the scorecard lives, not where the spend goes.

How far back can I go? As far back as your GA4 property holds. Most brands have more history than they realise, and it is measurable today rather than after a 90-day learning period.

The Causality Engine alternative - concretely

You already have the data. It is in your GA4 property right now.

1. Open GA4. Export the period you care about as a CSV. Any window — last month, last quarter, the launch you are still arguing about.
2. Upload it at /start.
3. Pay €99. Once. No subscription, no annual contract, no setup call.
4. Within 5 to 10 minutes you get a per-channel causal view: which channels actually drove incremental revenue, with confidence intervals, not which channel happened to be tagged on the last click.

No pixel, no SDK, no engineering ticket. It runs on aggregated first-party data, which is why it does not break on the next browser or consent change. If the first read does not move a decision, you get your money back.

Want continuous attribution instead of a one-off — automated GA4 ingestion, a chatbot over your own history, a developer API — that is Pro at €299/month, cancellable any time.

Before you commit

• Watch it run on a real store first: the interactive demo, no signup
• See pricing: /pricing
• How it works, step by step: /product/how-it-works

Own the foundation. Rent the discovery. The scorecard belongs on the owned side of that line.