Tracking & attribution engineeringClient work
Tracking recovery for purchases that stopped recording after Shopify's 2026-08-26 checkout upgrade, or after a store changed hands.
Sounds familiar?
How I solve it
I build the purchase leg back on the supported path that survives checkout changes — Shopify Customer Events and web pixels — and prove it with a live order rather than a screenshot of a test event. Where a store changed hands, I first separate what is inherited from what is broken, because treating an old tag as intentional is how a duplicate gets a second life.
The two entry points are the same recovery. One is the 2026-08-26 Shopify checkout upgrade, which forced Thank-you and Order-status extensibility on non-Plus stores and stopped code in Additional scripts from executing. The other is a store acquired from a previous owner, where the entire stack arrived undocumented.
The process
6 steps. Scope and a fixed price are agreed before the first one.
What fires today, from which owner, with what identifier, and what moved or died. On an acquired store this is the ownership map: every tag attributed to a source or marked unknown.
Additional-scripts code becomes a Customer Events / web pixel, so nothing depends on the deprecated checkout hook.
The click id and the client id are written onto the order, so the purchase can be joined back to the click that produced it.
Two owners with the same label is the usual cause of double counting; one owner and one event per order is the target.
A test order, then a real order: one purchase event, the correct value, the click id attached.
What changed, where, and how to roll it back.
Handover
Built with
Proof
Client work Done for real clients. Client details are anonymised.
Before → after
On a headless store rebuilt ahead of the same deadline, the share of sales traceable to any source went from 6% to 76%, and duplicate purchase reporting went from +100% to 0% — read from the live GA4 and Google Ads APIs in the August 2026 handover.
Headless Shopify automotive-repair retailer, US market
6% to 76%
sales traceable to source 6% → 76%; duplicate purchase reporting +100% → 0%; Google-attributed sales per day 0.1 → 10.5.
Read the case: 16 Days Before Shopify's Deadline: Rebuilding Measurement for a Headless StoreFixel Pixel
the delivery engine this recovery hands off to; one release to 61 store apps on 2026-09-26.
[He] took a Shopify and Google Analytics job that a couple of previous freelancers called themselves experts could not finish. [He] delivered great.
What you can look at
A redrawn timeline of the 2026-08-26 checkout upgrade, plus a before/after purchase-debug excerpt rebuilt on a synthetic store (redacted identifiers). The real reads behind it are in the numbers below.
Client details anonymised. Figures read from live platform APIs in August 2026.
Price and timeline
A bounded recovery is priced as a Focused Fix before it starts, from USD 450. If it is not yet clear what broke, the honest first step is a Working Session (USD 195) on one question, or a read-only Tracking Health Check (USD 395) with a written verdict in two working days after kickoff. The deadline that matters is already behind us: 2026-08-26, when the checkout upgrade stopped Additional-scripts code on non-Plus stores. Anything built after that has to use the supported path.
QuoteFixed before work starts
A note from Daniilbefore you decide
If you are on Shopify Plus and never used Additional scripts, and your purchase conversions still agree with Shopify to within a percent or two, this is not your problem. And if all you need is a straight answer about whether the break is real, a 60-minute Working Session is cheaper than a rebuild — I’ll tell you that rather than sell the larger job.
— Daniil
Questions
No. The upgrade stops the old code path; it does not stop a web pixel. The fix is to move the purchase leg to the supported path, which is what this recovery does.
Not necessarily. The work is to stop depending on Additional-scripts code and duplicated owners; whether the channel app stays depends on what your read-only map finds.
A test order first to confirm the shape, then a real order to confirm the purchase lands once with the right value and click id.
No theme rewrite. The pixel path is the supported one, and every change ships with a rollback record.
I can map what exists now and label it by evidence. I can't invent intent from a previous owner's setup, and I won't present an inherited tag as if it were mine.
The first answer is free, within one working day. Or write directly: next@taskfordaniel.com
