Ads, analytics, and CRM each show a different conversion total
Nobody is lying on purpose. The event is not the same object. We write one name and map the rest as diagnostics or kills.
Analytics and Measurement
Three tools each reporting a different conversion count is not a stack problem first. It is a missing event definition. We name what fires. We name the source of truth. We name duplicates. We write what leadership may quote. We do not sell a dashboard as the product. We do not reprint the mix chooser. We do not run CRO experiments as this job.
Position
Analytics and measurement on this site is the source of truth for events and channels: what fires, what is duplicate, and what leadership may quote. It is not a Looker dashboard as a product. It is not the digital marketing mix chooser. It is not a GA4 cookbook. It is not landing page CRO. It is not PPC conversion-action ops as the whole job.
Digital marketing is the mix chooser. Open that hub when the unknown is sequence. Open this page when the unknown is the event and the quote rule.
Open digital marketingLead generation defines the record and the route. This page maps how that record fires in tools. A form without a map still produces three counts.
Open lead generationPPC owns actions inside the ad account. This page insists they map to the shared event or sit as diagnostics. Waste lists stay on PPC.
Open PPCTechnical SEO owns whether the thank-you and the offer URL can be fetched. Search Console as access diagnostic lives there. Commercial quoting lives here.
Open technical SEOIf two people cannot agree what a conversion is, channel debates are noise. We start with one event leadership will still recognize in six months: a conversation, a demo, a sale, or an import you can audit in CRM or commerce. Search intent on a thank-you page still has to match that event. Keyword targeting a vanity micro-conversion is how boards fill with noise.
Source of truth is the second artifact. Analytics, ads, CRM, and Search Console answer different questions. Search Console stays a diagnostic for queries and pages. Ad platforms own their conversion actions inside PPC. CRM owns whether a human took the record. We write which tool may answer which question. We refuse a fourth tool that only exists to average the other three.
Dedupe is the third artifact: double thank-yous, two tags on one button, imported fills that already fired on the page, and chat events counted as demos. Double fires look like growth. Quote is the fourth: a sentence leadership may say, plus a list of numbers they may not add together. Channels may keep labeled footnotes. Leadership quotes the primary event or they quote nothing commercial from this program.
Lead generation supplies the sales meaning of the record. This page supplies the firing map. If either is missing, we do not decorate a dashboard. We will not claim incrementality percentages we cannot audit. A pretty board built on duplicates is worse than a spreadsheet with one honest column.
When this fits
This page fits when tools disagree and leadership still quotes a number. If you needed a mix sequence, a form definition, a landing experiment, a crawl pass, or a dashboard SKU, start there instead.
Nobody is lying on purpose. The event is not the same object. We write one name and map the rest as diagnostics or kills.
Growth is a ghost. Dedupe before anyone celebrates a campaign.
Queries are a diagnostic. We put Console back in its job and pick a commercial source of truth.
The quote rule died in a sprint. We restore a change control for events.
Unlike units were added. We split the footnote. AI SEO can keep sampling. It does not get to be the sale.
A board without a taxonomy reprints the argument in nicer type. We write the map first.
Process
Four measurement jobs in this order. Each job ends with a note finance and operators can keep. We do not start a Looker build in the last job. The point is one event leadership may quote.
We list every current conversion in analytics, ads, and CRM. We pick one commercial event. Search intent on the thank-you has to match it. Keyword targeting micro-conversions become diagnostics or kills. Semantic keywords in event names do not make two fires one object. Coverage of a campaign is not a reason to keep a junk event.
Outcome A dated event list: primary, diagnostics, and kills.
We write which tool answers which question. CRM or commerce for the primary when a human or an order exists. Ads for cost and platform-reported actions, labeled. Analytics for on-site pathing. Search Console for queries. Entity optimization here is the same event name across tools, not five synonyms. LLM visibility stays a footnote tool, not a source of pipeline.
Outcome A source map: question, tool, and what that tool may not answer.
We find double tags, double thank-yous, imported duplicates, and chat events counted as demos. We fire a test. We watch counts. Topical relevance of a webinar is not a reason to keep two success events. If technical access breaks the thank-you, we hand that fetch to technical SEO and still record the gap here. Layout experiments stay on landing page CRO.
Outcome A duplicate list with owners to kill or relabel each fire.
We write the sentence leadership may say, and the additions they may not make. Search-position reports, blended return across unlike channels, and model mention counts stay off the commercial sentence. Recheck monthly unless spend is leaking daily. Screenshot culture is not measurement.
Outcome A quote rule dated for the next leadership meeting.
Deliverables
You leave with an event list, a source map, a duplicate list, and a quote rule. You do not leave with a Looker SKU, an incrementality percentage, or a blended return floor.
Primary, diagnostics, and kills. Dated. Last year's GA view is not this list.
Which tool answers which question. A tool that answers all questions is a finding.
Fires to kill or relabel, with owners. Untested assumptions are named.
What leadership may say. Unlike units stay unadded.
Who may add an event, and how a campaign gets a diagnostic without becoming the primary.
Benefits
Channel notes remain notes. Arguments get specific.
Campaigns stop being rewarded for a tag that ran twice.
Queries inform pages. They do not pretend to be revenue.
Change control sits with the map. The quote rule can last more than a sprint.
Methodology
This works when someone owns the primary event and someone can reject a new fire. If both are missing, we are collecting screenshots. We are not measuring.
We start with the quoted number, the conversion lists, and the person who can edit tags. If any of those is missing, we name it first.
We fire a test conversion. We do not trust a vendor demo. Does this thank-you match the CRM stage?
You may rename the roles below. You may not leave the primary event implied. An implied event is how three totals share one slide.
First working session
The first hour is the quoted figure plus the three tools that might have produced it. It is not a tour of Looker galleries.
We write the number. We ask for the export. If three exports could match, that is the job. If the real question is which engine to open, that is the mix hub. If the real question is sales meaning, that is lead generation. If the real question is fetch, that is technical SEO. If the real question is a landing layout, that is landing page CRO. If the real question is model sampling, that is AI SEO or LLM SEO.
Then we trigger a test event. We watch analytics, ads, and CRM. Extra rows are usually the job. A dashboard mock is not evidence.
We leave with a first map or a stop. A map names the primary event, the source, and one duplicate to kill. A stop names what is missing: tag access, a CRM owner, or a fight over newsletter-as-revenue. Both are outcomes. A BI rebuild is not.
If the hour reveals that PPC conversion actions are the only honest fire, we still write the source map so leadership stops adding analytics pageviews to those actions. If the hour reveals that SEO wants search-position numbers in the same sentence, we refuse the sentence. Those numbers are not this event.
AI search
Honest events can keep leadership from quoting fiction. This page cannot sell AI Overviews, a chat mention, or LLM visibility as a score.
People ask whether better measurement will land the brand in Google AI Overviews (AIO). If the site already publishes three conversion stories, generative search has a mess to lift. One event name is ours as taxonomy. Measuring invented answers sits on an AI search audit. A program across search UIs is AI SEO. Mention sampling is not this scoreboard.
AI search and analytics both need a fetchable path. A virtual page that never loads for a crawler is a finding. We do not sell Overview presence.
We check that the primary event uses the product or conversation name you still stand behind. That is entity optimization as a measurement input. It is not a Knowledge Graph project from this page.
LLM visibility needs a sentence a model can quote, then measurement on LLM SEO. A quote rule can footnote sampling. It cannot replace pipeline.
Schema
JSON-LD can describe an offer the page already states. This work does not add conversion markup as a search-position product.
If schema implies a sale the event is not, extractors will repeat the lie. Keep types matched to the primary event.
We refuse FAQ blocks that exist only to host dashboard numbers. Real questions can live on the measurement page.
Markup will not force generative search to repeat your KPI. Eligibility is not a purchased mention.
Who we work with
Someone has to refuse a new conversion. We work with that person. We do not replace them with a second BI retainer.
Owns the quoted sentence. Holds the line when a channel wants its own primary.
Ships dedupe. Rechecks after a template or GTM publish.
Can keep footnotes. If they try to run this as an account rebuild or a crawl program, two jobs collide.
Questions
The event leadership may quote, and one count of that event across tools. We name what fires, which tool is the source of truth, what is duplicate, and what may be said in a meeting. We do not sell a dashboard as the product.
The mix hub chooses which engine opens this quarter. This page names the event and the quote rule. A chooser is not a measurement map. A dashboard product is not a chooser.
No. CRO tests whether a paid destination finishes the query you bought. This page does not run experiments on layout or proof. Duplicate fires can look like a CRO win. That is a map problem, not a headline test.
No. Implementation follows the map. A cookbook without an event definition creates more duplicate conversions. We will name what fires, what is duplicate, and what leadership may quote.
Lead generation defines the record sales will take and the capture path. This page insists that the same event has the same meaning in every tool, and that duplicates are named. A definition without a map still produces three conversion counts.
One event, one count
Show the three conflicting counts, the number leadership quoted last week, and a screenshot of each conversion list. We will say which event you actually have, which fires are duplicates, which tool may speak, and what must stop being quoted. Then we name the map or the stop.
Name, home, kill, speak. Not four vanity dashboards.