Project attribution-paid-media-measurement-repair · revised 11 October 2026
Project
Make one client site's paid-traffic measurement trustworthy from click to conversion
Campaign tagging, ad conversion tags, consent order and any site-to-checkout hand-off on one site are checked together and proven with test visits.
This asks for a proposal by email. Nothing is charged, and nothing starts, until you have agreed the list, the price and the terms in writing.
The result you are buying
Disagreeing numbers usually come from several small faults at once: inconsistent campaign links, conversion tags that fire twice or without value, a consent order that blocks or leaks signals, and a hand-off to another domain that starts a new session. Fixing one leaves the others.
Who it’s for: An agency or in-house marketer responsible for paid traffic to one site and unable to reconcile what Google Ads, Meta and GA4 each report.
Usually starts when: Platforms disagree about conversions, paid traffic is shown as Unassigned or Direct, and a client is asking which numbers to believe.
The result: You buy the finished chain, not the parts. A test visit arriving from a tagged paid link is attributed in GA4 as agreed, reaches a test conversion that GA4 records once, fires each ad platform's tag once with the agreed value and currency, and, where the site hands buyers to a checkout domain, stays in one session across the hand-off. The evidence is a test book, with one row for each test visit, and a reconciliation note.
How the work fits together
The project is complete when each required part has passed its named test, the end-to-end test visits are attributed as the test book describes and you have accepted the test book and the reconciliation note. You accept the finished set, not each internal step.
Fix campaign link tagging so paid, email and social land in the right GA4 channel Job Included
Your live campaign links use one naming rule, and each tested link pattern appears in GA4 with the agreed source, medium, campaign and channel.
Links are corrected first so the test visits can be traced. In this project the component covers up to ten link patterns.
Prove your Google Ads and Meta conversion tags each fire once on a test conversion Job Included
One test conversion fires your Google Ads and Meta conversion tags once each with the agreed value and currency, read in Tag Assistant and Meta Test Events; the Ads status is recorded as it stands.
After: Campaign tagging and channels
Make your cookie banner's choices reach your Google tags in the right order Job Included
Consent defaults load before any tag, a banner choice updates them, and Tag Assistant shows the same states you agreed with your own policy.
Technical wiring from your own policy; not legal advice.
Keep one visitor as one visitor when your site hands them to a separate checkout Job Optional
A test journey from your site to your checkout domain stays in one GA4 session, with the checkout not listed as the traffic source.
Only if a separate checkout or booking domain is used and the hand-off check finds it needs repair; the check itself is in the from price.
Make each order send one GA4 purchase event, not two or three Job Optional
A test order creates exactly one purchase event with its own transaction ID, and reloading the confirmation page does not create another.
For shops where purchases are doubled.
Get one GA4 key event recording again, proven with a test event Job Optional
One named action on your site produces one GA4 event with the agreed name, visible in DebugView and then counted as a key event.
Only if the test conversion does not already record in GA4 as a key event; the project's end-to-end test needs it to record.
How an engagement works
The price covers the agreed link patterns and platforms only. Extra platforms or patterns are quoted separately.
How it starts
You send the site address, the platforms and a description of which numbers disagree.
We review the links, tags and hand-off and propose the parts that apply, with a fixed price.
You agree the parts, price and terms in writing. Nothing starts before then.
We repair each part in a workspace, test it and send the evidence for your review.
We run the end-to-end test visits, write the reconciliation note and hand everything over.
Who decides what
You approve link and tag changes and publish. Account roles stay with you.
Handover
You receive the naming rule, corrected links, the container version, the test book and the note. Anything unfinished is handed back with notes.
Sharing your product safely. Send descriptions and links with personal data removed. Before work starts we agree with you in writing which identity receives the scoped roles.
What is included, and what is not
- A naming rule, corrected links and a custom channel group if one is needed
- Repaired conversion tags and consent wiring, prepared in a workspace for your approval
- An end-to-end test book: each test visit from click to conversion with evidence from each system
- The reconciliation note
Included
- One site, one GA4 property, one Google Ads account and one Meta account, with up to ten campaign link patterns (a pattern is one distinct combination of source, medium and campaign-naming style)
- Campaign link naming and channel assignment, ad conversion tags and consent order, each checked and repaired
- A hand-off check to one other domain, where the site uses one and, where the check finds the hand-off needs repair, the cross-domain repair priced as the optional component
- An end-to-end test book: each test visit from tagged link to test conversion, with a unique test campaign value and evidence from GA4 and from each ad platform's test tool
- A reconciliation note for one test week, in a table, that lists for Google Ads, Meta and GA4 what each counts, the days and conversion definition used, and which differences cannot be reconciled
Not included
- Campaign management, budgets, bids, audiences or creative
- Server-side events or the Meta Conversions API unless added to the list
- Legal advice on consent or privacy
- Deciding which platform's figure is right for billing or reporting
- Any promise about ad performance, revenue or rankings
How we know it’s done
Agreed with you before work starts. Each check produces evidence you keep.
Each tagged test link pattern from the agreed list, opened with a unique test campaign value in a new browser profile, carries the agreed source, medium and campaign, and after Google has processed the data appears in Traffic acquisition under that campaign value and the agreed channel, not under Unassigned. Paid-search patterns that rely on Google Ads auto-tagging are marked as read from the link, not visit-tested.
Evidence: Address and Traffic acquisition screenshots per test link; Realtime may be added as a quick sighting but is not evidence.
One test conversion fires each ad conversion tag exactly once: the Google Ads tag in Tag Assistant with the agreed value and currency, and the Meta event in Meta Ads Data Advisor (formerly Meta Pixel Helper) and Meta Test Events with the agreed name, value and currency. Reloading the confirmation page adds neither.
Evidence: Tag Assistant, Data Advisor and Test Events screenshots, including the reload.
On first visit, consent defaults are set before any tag fires, and after accept and after reject the states match the written default, accept and reject table you supplied.
Evidence: Tag Assistant consent screenshots for each path.
One test visit from a tagged test link with a unique test campaign value, followed by one test conversion, appears in GA4 after Google has processed the data as a key event under that campaign value, with the agreed session source / medium and channel.
Evidence: Exploration screenshot filtered to the test campaign value, and the test book row for the visit, with its time and test value, beside the Tag Assistant and Test Events screenshots from the matching tag-tool run.
Where the site hands buyers to another domain and the cross-domain repair is in the proposal, a test journey from a tagged link through the hand-off carries the _gl parameter and, where the checkout side can be put into debug mode, shows the same ga_session_id on the page views on both domains in DebugView (otherwise marked not run in the test book), and a separate marker journey made without debug mode is not attributed to the checkout domain as a referral in Traffic acquisition. Where the repair is not bought, the same check is run once and its result is recorded in the test book as found, without a repair. Where the site has no hand-off, this check is recorded as not applicable.
Evidence: Destination address, DebugView screenshots of both page views where run and, from the separate marker journey, the Traffic acquisition rows for the test value.
The reconciliation note covers one named test week and lists, for each of Google Ads, Meta and GA4, the figures read from that system, the days and conversion definition used, and the differences that cannot be reconciled.
Evidence: The reconciliation note as a table, with the system and date range each figure was read from.
Sign-off. You review the test book and the reconciliation note, then accept the finished set in writing.
If it fails. A part that cannot pass its test is named with the reason and what it would take, and is removed from the list with a matching change to the price. Nothing is billed as delivered that you have not accepted.
When it fits, and when we stop
It fits when
- You hold the accounts or have written authority from the account holders, and can grant scoped roles
- A test conversion route exists without real payments or real customer contact
- The site uses Tag Manager or an agreed route for editing tags
- A test browser can accept analytics cookies on the site
- The test conversion is already recorded in GA4 as a key event on this property. If it is not, the one-off key-event repair (GBP 195) is added to the proposal first at its listed price, because the project's end-to-end test needs the key event to record
- A consent banner or consent tool is already installed, and you supply in writing the default, accept and reject states your own policy requires for each Google consent type; we do not choose the states
- No Active data filter on the property drops the test visits. Google says a developer-traffic filter removes activity from debug mode and that excluded data is never processed; if one is Active, the test visits whose rows are read in GA4 are made without debug mode, separately from any DebugView or Tag Assistant run, agreed in writing before work starts
We stop and tell you if
- Account ownership is unclear or disputed
- More than ten campaign link patterns or more than two ad platforms are in scope
- Ad click information cannot reach the conversion page because of a redirect you will not change
- The test conversion does not record in GA4 and the key-event repair is not added to the proposal
- No one can state the default, accept and reject consent states you want; we do not choose them for you
What could go wrong
Link changes are listed old and new side by side and tag changes are a named Tag Manager version, so each part can be reverted on its own. If the project stops, accepted parts stay accepted and the rest is handed back with notes.
Scroll the table sideways to read it all.
| Risk | How we handle it |
|---|---|
| Fixing the tags makes platform numbers move and a client reads that as a performance change. | The handover records the change date and the reconciliation note says what changed in collection. |
| Test conversions are counted as real activity. | Tests use a marker and are logged in the test book so they can be excluded or noted. |
Each part is reviewed separately from the work that produced it, and you accept the finished set. No human supervisor is included unless your proposal names one. At launch the work is largely automated, and we say so.
Questions
Will the numbers across platforms match afterwards?
Not necessarily. Platforms count differently. The reconciliation note explains what can and cannot be compared.
Do you run my campaigns?
No. We repair collection and prove it; we do not change campaigns.
What if my test conversion does not show in GA4 as a key event?
Then the one-off key-event repair (GBP 195) is added to the proposal first at its listed price, because the project's end-to-end test needs the conversion to record. Without it the project does not start.
Can I buy the parts separately?
Yes. Campaign tagging is GBP 245, ad conversion tags GBP 295 and consent wiring GBP 395 as separate fixed-price jobs. The project adds the end-to-end test book, the reconciliation note and one coordinated acceptance of the finished set.
Send an enquiry
Send us
- The site address, the ad platforms in use and a description of which numbers disagree
- Five example campaign links, with personal data removed
- Whether the test conversion already shows in GA4 as a key event, and which consent tool is installed
- No logins and no account IDs.
Later, once you agree
- Scoped roles on GA4 (Editor), Tag Manager (Approve, so container versions can be saved for you to publish), Google Ads and Meta Events Manager, granted to the identity we agree with you before work starts
- The campaign link list and a safe test conversion route
- A written table of the default, accept and reject consent states for each Google consent type, from your own policy; the consent check is made against this table
- Your named person to publish and to approve link changes
Your site, accounts, containers and data stay yours. We work in a Tag Manager workspace and, where the plan needs it, on a theme or staging copy, with the lowest role each step needs, granted to an identity we agree with you in writing before work starts. You review and publish. We never ask for passwords.
Email fallback: open your mail app
If website submission is unavailable, review and send the fallback email yourself. An email fallback is not a website receipt. Or write to hello@syntheticindustry.ai with “attribution-paid-media-measurement-repair” as the subject.