Why your ad platform and your CRM disagree

Google Ads reports 84 conversions last month. GA4 shows 62. The CRM has 51 leads and sales say about 30 were worth talking to. Someone senior asks which number is right, and the honest answer is none of them, which is not an answer anyone wants in a budget meeting.

These systems will never agree exactly, and they are not supposed to. They count different things over different windows with different attribution. But a gap of two or three times is not a modelling difference, it is a defect, and it is usually one of a small number of mundane causes.

Start with double counting

The most common cause, and the easiest to check. Fire the form and watch the network requests.

The tag fires on page load of the thank-you page. Someone refreshes, or bookmarks it, or navigates back. Every one of those counts again.

Both a form submission trigger and a thank-you page view are configured. One enquiry, two conversions. This happens constantly when tracking has been added by more than one person over time.

The tag is present in two places. Hard-coded in the site template and also deployed through Tag Manager. Everything doubles, cleanly and invisibly.

Conversion counting is set to “every” rather than “one”. For a lead generation form, a customer who enquires three times is one lead, not three. Google Ads defaults are not always what you want here.

Then check what is being counted as a conversion

Look at the actual list of conversion actions in the ad account, and be honest about each one.

A newsletter signup is not a lead. A PDF download is not a lead. A click on a phone number link is not a call, it is an intent to call, and on desktop it is frequently a misclick. A thirty second page view is not anything.

If any of those are marked as primary conversion actions, they are inflating the number and, worse, the automated bidding is optimising to produce more of them. That is not a reporting problem, it is a budget allocation problem.

Then check the clicks that never became sessions

A gap between ad platform clicks and analytics sessions is normal, and a large one is not. Common causes:

• The landing page is slow enough that people leave before the tag fires. More common on mobile than anyone expects.
• Redirects strip the tracking parameters. An http to https redirect, or a www redirect, configured to drop the query string. The click happened, the session is unattributable.
• Consent is denied and nothing is modelled. Depending on configuration, denied consent can mean the session is simply absent rather than estimated.
• Ad blockers. A real and unfixable share, larger in technical audiences than in consumer ones.

Then look at the phone

If your business takes enquiries by phone, and you do not have call tracking, this is your largest gap and everything above is a rounding error next to it.

Without dynamic number insertion, the platform cannot know that the click produced a call. Ad platforms will report on calls from ad extensions, which covers a fraction of the real volume, and misses everyone who clicked through, browsed, and rang the number on the contact page. In several sectors that is the majority of enquiries.

Then look at the CRM

Once the tracking side is sound, the remaining gap is often on the receiving end.

Leads that arrive by phone and never get entered. Leads entered by a salesperson under a different source. Duplicate records merged, so three enquiries become one. Enquiries that arrive out of hours and get handled by text without anyone creating a record.

This is why the marketing number and the sales number diverge even when both systems are working. Neither is lying, they are counting different populations.

Attribution windows, last

After all of that, some difference remains and it is legitimate.

Google Ads credits a conversion to the date of the click, not the date of the conversion. GA4 credits the session. If your sales cycle is three weeks, a conversion in August may be attributed to a July click in one system and to August in the other. Comparing two months side by side will never resolve this.

Different attribution models split credit differently. Different lookback windows include different clicks. These are modelling choices, not errors, and the right response is to pick one system as the source of truth for each question rather than trying to reconcile them.

What good looks like

You do not need the numbers to match. You need to know why they differ and to have that explanation ready before someone asks.

A workable standard:

• Every conversion action has a written definition that sales agrees with.
• One enquiry produces exactly one conversion.
• Calls are tracked and scored before they count.
• Lead source is written into the CRM automatically.
• One dashboard is the source of truth for spend efficiency, and everyone knows which one it is.
• The known gaps are documented, with rough sizes.

That last point does more for credibility than any amount of precision. “Ads is about fifteen percent above the CRM because of blocked sessions and out-of-hours calls we handle by text” is an answer. “The systems just count differently” is not, and it is how marketing loses budget arguments.