Why Your Google Ads, GA4 and CRM Numbers Don't Match

An advertising report, a website analytics report and a customer relationship management system can each describe part of the same enquiry without producing identical totals. The useful starting point is to establish what each number represents.

A campaign report may attribute an action to advertising. A website report may record the action itself. A CRM may contain a contact that has been reviewed, merged with an existing record or moved into a sales process. Comparing the headline totals before understanding those definitions can turn a reasonable difference into a false alarm.

The opposite mistake is to assume every discrepancy is harmless. Missing events, duplicated submissions or incomplete field mappings can leave the business making decisions from faulty information. The aim is to distinguish expected differences from errors that need attention.

Google's data-discrepancy guidance identifies several possible causes, including counting, attribution, timing and tag setup. There is no single explanation that can be assigned to every account.

Begin with a specific discrepancy and a business question. Is the owner trying to understand how many enquiries arrived, which campaigns deserve credit or how much suitable work entered the pipeline? Those questions are related, but they require different evidence.

A useful reporting system makes the distinction explicit before presenting a combined view.

The word “conversion” can conceal several meanings. A business might use it for a submitted form, an advertising platform for a configured conversion action, and a sales team for a prospect that becomes a customer.

Current Google Analytics guidance on key events and conversions distinguishes important website or app events from conversions used for advertising measurement. It also describes shared conversion measurement across Analytics and Google Ads. The report and configuration being compared therefore matter; it is inaccurate to claim the platforms can never align.

Translate the platform terminology into the business process. What happened when the form was submitted? What qualifies the request as a useful enquiry? When does it become a sales opportunity? Which outcome represents completed business?

A hypothetical prospect might submit a form, follow up by telephone and later be added to an existing account in the CRM. Those interactions should not automatically be interpreted as separate new customers. The business needs a consistent way to relate them, within the information it is entitled and able to use.

Write a short definition for each important measure. Include what qualifies, what is excluded, where it is recorded and who owns the definition. “Qualified enquiry” is especially important because software cannot supply the commercial meaning on its own.

This work may reveal that the reports were never intended to match exactly. It may also expose a genuine inconsistency that has been hidden behind familiar labels.

Select one important action and trace how it is represented in the relevant systems. Record the settings in force for the period being reviewed, rather than assuming today's configuration explains older data.

The reconciliation record should capture:

Use Google's discrepancy documentation to investigate the relevant platform settings. Avoid relying on a remembered default, especially after an account change or a new integration.

Time zones and currencies also deserve an explicit place in a business report covering several markets. Decide how the organisation wants to compare periods and values, then preserve the original context needed to investigate a difference.

The purpose is to explain the gap, not to adjust the figures until they agree. Where settings can sensibly align, make a controlled change and record when it takes effect. Where the systems measure different things, label the distinction clearly.

A dashboard built before this work may simply make an unresolved disagreement easier to see.

A CRM becomes a stronger source of commercial evidence when its records reflect the actual handling of enquiries. That requires agreed fields and states, not merely a connection that copies website submissions into a contact list.

Salesforce's lead-management documentation illustrates the configuration involved: lead fields, assignment and mappings into later records. The general lesson is that a system connection still needs business decisions about what information should survive each transition.

Preserve the source information the business needs to evaluate acquisition, where it is available and appropriate. Also preserve the enquiry's progression: whether somebody reviewed it, whether it fit the service, what happened next and why it was closed.

Avoid making every missing source a forced attribution. Some customers will arrive through conversations or combinations of activity that the available records cannot resolve. “Unknown” is more informative than a confident label assigned without evidence.

Review duplicates deliberately. The same organisation may generate several legitimate opportunities, while several submissions may concern one request. A rule that merges everything by email address or company name can erase useful distinctions if the sales process has not been considered.

The aim is a record that supports an accountable commercial history. That helps the business examine lead quality and outcomes even when advertising attribution remains incomplete. It also gives marketing and sales a shared basis for discussing what happened after the initial contact.

Once the definitions and settings are understood, test the path from action to record. Submit an authorised test enquiry, inspect what is recorded and confirm where it arrives. Use appropriate test data and distinguish the test from genuine customer activity.

Check that the intended event occurs at the right point. A click on the submit button is not necessarily evidence that the request was successfully received. Examine repeated submissions, failed attempts and what happens when the person returns to a confirmation page.

Then inspect the integration. Are the relevant fields reaching the CRM? Does a retry create an unintended duplicate? Are records being rejected or left in a queue? Can the responsible team see failures and resolve them?

Allow for the processing behaviour of the systems involved, using current documentation and the account's actual setup. Do not invent a universal waiting period after which every report should agree. Equally, do not leave a persistent failure unexplained by calling it a delay.

Document what was tested, what was observed and what remains unresolved. A successful test confirms that a particular path worked under those conditions. It does not prove that historical tracking was complete or that every customer's journey is visible.

That level of precision makes a technical review useful to the owner who must decide whether the reporting is dependable enough for the next investment.

The final report should let the business answer its main questions without pretending that one number describes the whole journey. Acquisition measures can inform campaign management. Enquiry and opportunity records can inform sales capacity and qualification. Completed business needs its own verified commercial records.

Choose the appropriate measure for each decision and explain its limits. Do not add channel-attributed totals together as unique business outcomes unless the underlying overlap has been resolved. Do not present unqualified submissions as pipeline simply because both are available in the reporting tool.

Reconciliation also does not establish causality. Knowing which channel received attribution does not, by itself, prove how much additional business the activity created. Where that question matters, a suitable comparison or experiment may be needed. Google's experiment guidance describes comparing a treatment with an original campaign, while recognising that the evidence may remain inconclusive.

When scoping growth marketing and measurement work, ask which business decisions the report must support. That answer should determine the definitions and integrations, rather than the choice of dashboard dictating what the business measures.

The objective is a report the business can interpret and investigate. Expected differences should be explained, technical errors corrected and unresolved gaps visible enough that they do not quietly become the basis for the next budget decision.