Blog Inquiry counts

Inquiry counts

Why Do GA4, Your Forms and Your CRM Show Different Inquiry Counts?

Three systems, three numbers, and nobody sure which one is right. A gap can come from counting different things at different moments. The job is to explain every gap, not to force the numbers to match.

Sunlit marble desk with three frosted glass plaques on orange, blue and purple glowing bases, labeled GA4 events, Form tool and CRM contacts, joined by light lines to a magnifying glass, with a monstera plant, a mug and a tablet, beside the title Why Do GA4, Your Forms and Your CRM Show Different Inquiry Counts?

We read the official Google Analytics and Google for Developers documentation on form events, key event counting, data freshness, reporting surfaces, consent mode and the Measurement Protocol, plus HubSpot's documentation on duplicate contacts. The common thread: each system counts a different unit, at a different moment, with its own rules for repeats. What we offer is a set of definitions, ten causes to check, a one-inquiry trace and a twelve-step reconciliation method.

5 or 1

key events counted when one visitor triggers a key event five times in a session, depending on the GA4 counting method (Google Analytics Help)

12 days

how long attribution credit for GA4 key events can keep changing after the event is recorded (Google Analytics Help)

24 to 48 hours

how long Google says GA4 processing can take, with report data changing in that window (Google Analytics Help)

The thesis

Do not ask which system is right. Write down what each one counts, trace real inquiries from the form to the CRM, and give every difference a cause, evidence and an owner.

The scene

Three dashboards, three numbers

A teaching example.

Your monthly review starts with a question nobody can answer. GA4 shows one number of form submits. The form tool shows a different number of submissions. The CRM shows fewer new contacts than either. Someone says the website is losing leads. Someone else says the systems always disagree.

Both may be wrong. Before you change the ads or rebuild the integration, find out what each system is actually counting.

How we looked

What the official documentation says

Google and HubSpot documentation, checked October 29, 2026.

We read Google Analytics Help pages on enhanced measurement, key event counting, data freshness, reporting surfaces, personal data and custom dimensions; Google for Developers pages on consent mode, recommended lead events and Measurement Protocol validation; and HubSpot's page on deduplicating records. HubSpot is one documented example of CRM behaviour, not a recommendation.

The method below is built from that documentation.

The units

Each system counts something different

Same word, "lead", different units.

What each number means

Where you lookWhat it counts
GA4 eventsActions sent to analytics, such as form_submit
Form toolSubmissions the form or server accepted
CRM contactsPeople, one of whom can send many inquiries
CRM inquiriesSeparate requests, if your CRM keeps them apart
Qualified inquiriesRequests judged against agreed criteria
OpportunitiesPotential projects with a next step, not sales

A Teamulate working method, based on the documentation listed below.

A returning customer shows why this matters. They send a new inquiry. The form tool adds a submission and GA4 records an event. HubSpot documents that a form submission with an existing contact's email adds the information to that contact, so the count of new contacts does not move. Nothing is lost. The units are simply different.

Google's recommended lead events (generate_lead, qualify_lead, working_lead) help name the stages, but they do not set your business rules. You decide when each one is sent.

The evidence

Why GA4 numbers move and do not match records

What Google documents about counting and timing.

GA4 enhanced measurement sends form_start the first time a visitor interacts with a form in a session, and form_submit when the form is submitted. A submit is not the same as an accepted inquiry: the click can happen even if the form tool rejects the request, and the request can be accepted while no event is sent.

A key event can be counted once per event or once per session. In Google's own example, someone who triggers a key event five times in a session counts as 5 or as 1. Changing the setting applies only to future data. The once-per-session setting is not a way to remove duplicates: it can also merge two real inquiries from one visit.

Timing matters too. Google says processing can take 24 to 48 hours, report data may change in that window, and attribution credit for key events can change for up to 12 days. Reports include key event modeling and data-driven attribution; the BigQuery export does not. Neither is a list of individual inquiries you can match one by one.

Consent settings change what is sent at all. In basic consent mode, Google tags are blocked until the visitor interacts with the banner; in advanced mode, they send measurements without cookies while consent is denied. A missing event is not proof of a missing inquiry, and the fix is never to work around a visitor's choice.

Symptoms

Ten causes to check before you call it a leak

Each one points to a different fix.

What you see, and what to check

What you seeLikely cause to check
More submissions than new contactsExisting contacts were updated, not created
More events than submissionsThe event fires on the click, not on acceptance
Fewer events than submissionsConsent, blocked tracking or a broken tag
Counts change between exportsReport data still processing
Month end does not matchDifferent dates and time zones
One person counted twiceDouble click, repeated webhook, two tags
Two real inquiries counted onceKey event counted once per session
CRM has more than the formCalls, emails, referrals and imports
Channels disagreeDifferent attribution scopes and models
Integration says successA technical reply taken as business success

A Teamulate working method. These are hypotheses to check, not verdicts.

The last one deserves a note. Google documents that the Measurement Protocol does not return HTTP error codes even when an event is malformed, and that events sent to its validation server do not show up in reports. "The API answered" is one step, not proof that the record reached the place it should.

The trace

Trace one real inquiry, end to end

Six checks, one record at a time.

  • Accepted: the submission ID and accepted time in the form tool or server.
  • Processed: the same ID handled once by the integration.
  • Recorded: an inquiry record in the CRM, linked to a new or existing contact.
  • Assigned: the right person or queue got the task.
  • Answered: the time of the first real human reply, separate from the automatic confirmation.
  • Observed: an analytics event that can be linked, where that is permitted.

Mark each check Yes, No or Unknown. Unknown is an honest answer, not a failure. Keep the trace in your own systems: Google's policies forbid sending personally identifiable information such as email addresses or phone numbers to Analytics, including in URLs and custom dimensions.

The method

Reconcile one form and one period

Volume first, channels second.

Fix the scope: one form or process, one service, one period and one time zone. Agree the definitions before you look at the numbers. Record the current settings, counting methods, filters and integration rules, so you know what you measured.

Pull permitted records from the form tool, the integration log, the CRM and GA4, and note when you pulled them. Classify submissions by rules set in advance; doubtful cases stay Unknown. Match to the CRM using exact IDs where they exist. A likely match without an exact key is a candidate for human review, not a proven match.

Only after the volume is reconciled, look at channels. Keep the first known source, the source seen on this inquiry and the self-reported answer to "how did you hear about us?" as separate fields. Do not turn unknown into direct.

Log every difference with a category, evidence, owner and next check. Fix only what you have reproduced, record the date of the fix, and reconcile again to make sure it worked.

Measuring

Measure coverage, not agreement

A missing result is unknown, not a pass.

Useful measures: CRM coverage (valid inquiries with a confirmed CRM record, out of all valid inquiries in the period), observed event coverage, the count of unexplained differences with its denominator, time from accepted to recorded, and time to the first human reply. A zero denominator is N/A.

There is no universal acceptable gap. One unassigned urgent inquiry can matter more than many explained differences in analytics. And do not create a GA4 custom dimension for every unique ID: Google warns that unnecessary high-cardinality dimensions, more than 500 unique values in a day, can push data into an "(other)" row.

Support

What a marketing team with AI support can keep running, and what stays with you

Start with one form and one month.

Start small: one form, one period, a handful of traced inquiries and a written list of what each system counts. If a shared spreadsheet answers the question, that is a good answer too.

What stays with you: deciding what counts as a valid and a qualified inquiry, and who owns each difference.

For B2B teams, checking the path from a form to a reply this way is part of the marketing operations our AI marketing team runs alongside you, with every step visible. You can see how our workflows run. To talk it through, request a demo.

Checklist

Checklist: before you trust the numbers

Run it on one form.

  • One form, one period and one time zone are fixed.
  • Each system's unit is written down.
  • The key event counting method has been checked in GA4.
  • Each accepted submission has an ID that retries reuse.
  • Contacts and inquiries are separate in the CRM.
  • A test submission was traced from the form to the owner.
  • Every difference has a cause, evidence, an owner and a next check.

Key terms and facts

The short version

Analytics event
A record of an action sent to GA4, such as a form submit. It is not proof that the business received a usable inquiry.
Accepted submission
A record the form tool or server accepted and stored. It can still be a test, spam or a repeat.
Valid inquiry
A real request for a conversation about a service, after removing tests, spam, technical repeats and other agreed categories.
Reconciliation
Matching records across systems and writing down why each difference exists, with evidence and an owner.

Key facts

  • TEAMULATE REVIEWGoogle Analytics Help, Google for Developers and the HubSpot Knowledge Base, checked October 29, 2026.
  • DOCUMENTEDGA4 enhanced measurement sends form_start on the first form interaction in a session and form_submit when a form is submitted.
  • DOCUMENTEDa GA4 key event can be counted once per event or once per session; a change applies to future data only.
  • DOCUMENTEDGA4 processing can take 24 to 48 hours, and attribution credit for key events can change for up to 12 days.
  • DOCUMENTEDin HubSpot, a form submission with the email of an existing contact updates that contact instead of creating a new one.
  • TEAMULATE METHODa reconciliation for one form and one period.

Takeaways

What to do with this

Different units

Events, submissions, contacts and opportunities are not the same thing.

Trace, do not guess

Follow real inquiries one by one before you blame the integration.

Unknown is allowed

Without evidence, the honest category is Unknown, not "lost to consent".

Volume before channels

Settle how many inquiries there were before arguing about where they came from.

Questions and answers

Common questions

Why does GA4 show more form submits than my form tool?

Two causes to check first: the event fires on the click rather than after the submission is accepted, or two tracking setups send the same event. Test one submission and compare the event, the form record and the CRM record.

Why does my CRM show fewer new contacts than form submissions?

In HubSpot, a form submission with an existing contact's email updates that contact instead of creating a new one. Check your own CRM's rule, and count inquiries from existing contacts separately.

Should I set key events to once per session to remove duplicates?

No. Google recommends once per event, and once per session can merge two real inquiries from one visit. Remove technical duplicates with a submission ID instead.

How long should I wait before trusting a GA4 report?

Google says processing can take 24 to 48 hours and attribution credit for key events can change for up to 12 days. Mark recent reports as preliminary, but do not wait to check whether a real inquiry reached its owner.

Can I send the email address to GA4 to match records?

No. Google's policies forbid sending personally identifiable information to Analytics. Keep the link between systems in your own records.

Sources

Named sources

Teamulate review of official documentation, checked October 29, 2026: