SKILL.md · 24000
holds trigger triage, not verdicts.** Before comparing any two numbers, line up the conversion action, the date basis, the time zone, the attribution window, and the campaign scope. Plenty of healthy accounts breach the bands above for boring definitional reasons.
- Absence of a browser tag is not absence of tracking. Server-side sends, CAPI, native integrations, and offline imports leave nothing in the page source. Ask before concluding anything is missing.
- "This is working" is not a conclusion static analysis can reach. From a page and a published container, the most you can honestly say is that no fault is visible. The faults that leave every visible signal looking correct (a paused tag, an exception trigger, work that was built but never published) are exactly the ones that matter. So scope every verdict to what you actually checked.
Verification discipline
Verification is the audit's evidence standard applied at the end of a setup, and partial access is the normal situation, not the exception. You can usually observe the browser half yourself. You almost never have the user's ad account login, so the platform half usually runs through them.
- Verify what's in your reach, and record it as observed. The test entry, the expected dataLayer event appearing exactly once, the tag firing in Preview, the network request leaving the page.
- For everything out of reach, hand the user a named check with an expected result. Not "check that it's working", but "submit a test using this email, then open Meta Events Manager, go to Test Events, and you should see one Lead appear within a minute. Tell me what you see." Their answer then becomes evidence, recorded as reported.
- Use the fast feedback surfaces. GA4's DebugView and Meta's Test Events confirm within seconds. Google Ads conversion columns can lag by a day or more. Knowing which surface is instant is the difference between a five-minute verification and a two-day wait, and that wait is routinely misdiagnosed as breakage.
- Test with hygiene. Use an identifiable marker (a test email you can search for later) so the test can be found and cleaned out of the CRM. Fire it from a real browser on the real page, never enter real personal data, and know that a fabricated click ID only tests transport. It can never prove attribution, because it matches no real click, so don't present a passing synthetic test as proof attribution works.
- Close with a status, not a verdict. Built, installed, firing confirmed, arrival confirmed, or arrival still pending along with the one check that closes it. Never declare something done on your own authority when you didn't observe it.
Choosing tools
Every tool named in this skill's references appears alongside its real alternatives, with current prices, honest effort and lead-time costs, and the free or native path always stated. One entry in references/server-side.md is marked as a partner of this repository. That marker is disclosure, not endorsement, and the fit criteria are the same for every row. The standing test is simple. If a recommendation wouldn't survive swapping the named tool for its nearest competitor, it doesn't belong here.
Two practical rules when a tool from the spectrum comes up in your answer. If it's a partner-marked tool, carry the disclosure into the answer in the same breath, because the disclosure lives in the reference file and the user only ever reads your answer. That applies to any answer that names it, even in passing or as one example in a list ("managed tools like X or Y"); once per answer is enough. The wording is "a Verified Partner of the repository this skill comes from". Use it exactly as written, as a short phrase, and don't recite the reasoning behind it. And if you lean toward any one option, name its nearest alternative alongside it, so the user is choosing rather than accepting.
And one rule that prevents underselling the options from memory. Before telling a user that no server-side path exists for their moment, tool, or destination, check references/server-side.md. Between them, native integrations, direct platform APIs, server-side GTM, webhook and CRM sync tools, and managed tools cover most moments, including iframes, bookings, chats, and backend-confirmed events. If you can't confirm a specific tool is covered, say so and point the user to that vendor's own integration list, rather than implying it isn't covered.
Where everything is
Load a reference when the task reaches it. Nothing here is optional reading, it's just split out so this file stays navigable.
| Task | Load |
|---|---|
| How each form, booking, chat, and checkout tool signals success | references/conversion-moments.md |
| Pixels, tags, and GTM builds, the complete browser-side path | references/browser-side.md |
| Server-side, the options spectrum with prices and honest tradeoffs | references/server-side.md |
| Shopify, WooCommerce, and other store platforms, including the deprecation dates | references/ecommerce-platforms.md |
| Google Ads setup, statuses, the gclid lifecycle, Enhanced Conversions, and its failure catalog | references/google-ads.md |
| Meta setup, CAPI, deduplication, match quality, and its failure catalog | references/meta.md |
| LinkedIn, TikTok, Microsoft, and the rest | references/other-platforms.md |
| The audit's evidence procedure, scoring, and report template | references/audit-playbook.md |
| Why the numbers never match, loss magnitudes, and the refund asymmetry | references/discrepancies.md |
Assets: assets/snippets/ holds paste-in success detectors for 22 lead-gen tools, plus four universal patterns for the moments no form tool owns: phone call clicks (the most under-tracked conversion in local services), file downloads, thank-you page arrivals, and a generic AJAX success pattern. assets/gtm-recipes/ holds the importable GTM containers (detection and send, with event names in event-map.json), and scripts/build_recipe.py merges them with the user's IDs into one import file.
Boundaries
- analytics owns tracking plans, event naming, UTMs, GA4 property setup, and general GTM craft. This skill owns getting a conversion to a destination and proving it arrived.
- attribution owns models, measurement approaches, and reconciling credit across channels. This skill produces the accurate per-platform numbers that attribution then interprets.
- ads owns campaign structure, bidding, creative, and account audits beyond tracking. When a tracking audit finds the tracking healthy, the problem is usually theirs.
- cro owns improving the conversion rate. This skill only measures it.
- revops owns lead stages, scoring, and routing in the CRM. This skill owns the mechanics of sending those CRM outcomes back to the ad platforms (offline imports and CRM conversions APIs).