The numbers disagree because each system answers a different question, credits touchpoints differently, and looks back over a different window. Here is how each one counts, and how to reconcile them.
Google Ads, Meta, LinkedIn, GA4, and your CRM are five independent systems. None of them sees the full picture, and none of them coordinates with the others. When a lead arrives, each ad platform asks a simple question: did someone we showed or clicked an ad for convert within our window? If yes, it claims the credit.
GA4 asks a different question: which session or channel should get credit for this key event? Your CRM asks a third: is this a real, unique person who became a lead? Three questions produce three totals.
Before comparing numbers, check that you are comparing the same unit.
| System | What it counts | Typical inflation or loss |
|---|---|---|
| Ad platform | Conversion events it can tie to an ad click or view, such as form submissions, calls, or imported actions | Duplicates, spam, overlapping credit, view-through conversions, modeled conversions |
| GA4 | Key events that occurred in a measured session, assigned to one channel by its attribution settings | Loses users who decline consent, block scripts, or switch devices |
| CRM | Unique contact or deal records that passed your own rules | Loses leads that never reached the CRM; source data is often missing or overwritten |
A form that is submitted twice, a spam bot that fires your thank-you page, and a returning customer filling in a second inquiry can each be one conversion in an ad platform and zero or one in a CRM. The first step in any audit is to confirm that the tracked event is a real success signal. See website conversion tracking for how this is verified.
An attribution model is the rule for dividing credit when a customer touches several ads or channels before converting. The model changes who gets the credit and, in some tools, how many conversions are counted at all.
| Model | How credit is assigned | Tends to favor |
|---|---|---|
| Last click | All credit to the final click before conversion | Bottom-funnel channels such as branded search |
| First click | All credit to the first known touch | Discovery channels such as paid social |
| Linear | Equal credit to every touch | Long, multi-channel journeys |
| Position-based | More credit to the first and last touches, less to the middle | Journeys where introduction and closing both matter |
| Time decay | More credit to touches nearer the conversion | Short sales cycles |
| Data-driven | Credit estimated from patterns in conversion paths | Depends on the platform's data and its own modeling |
The bigger issue is not which model is used, but whose data it runs on. Google Ads only sees Google touchpoints, Meta only sees Meta touchpoints, and LinkedIn only sees LinkedIn touchpoints. Each applies its model inside its own walled garden and has no knowledge of the others. The result is that every platform can legitimately claim the same conversion. LinkedIn, for example, uses a last-touch approach per platform, so a lead that also clicked a Google ad still shows up in both places.
GA4 is the only one of these tools that sees every channel together, which is why it usually credits a conversion once, to one channel.
A conversion window, or lookback window, is how long after an ad click or view a conversion can still be credited to that ad. Longer windows mean more conversions are claimed, and those claimed later are less likely to have been caused by the ad.
| Platform | Click window | View window |
|---|---|---|
| Google Ads | Default 30 days; configurable from 1 to 90 days for Search and Display | View-through default 1 day, up to 30. Engaged-view (video) default 3 days |
| Default 30 days; adjustable to 1, 7, 30, or 90 | Default 7 days; adjustable to 1, 7, 30, or 90 | |
| Meta | Default 7-day click; 1-day click is also available | 1-day view. Meta removed its 7-day and 28-day view windows in January 2026 |
| GA4 | Set by your property's attribution settings and reporting model; it only credits conversions tied to a measured session | |
Meta has moved in the opposite direction from Google and LinkedIn, so the longest windows on the market today belong to Google and LinkedIn, where clicks can be credited up to 90 days later. If your sales cycle is long and the window is set to the maximum, a lead that clicked an ad three months ago and later found you through organic search can still be claimed by the ad platform.
Google Ads reports conversions by default on the date of the ad interaction, not the date the conversion happened. A lead that arrives on day 40 of a 90-day window is back-dated to the click. This is why last month's Google Ads numbers keep changing after the month ends, and why comparing them to a CRM report filtered by lead creation date rarely matches. Use the by-conversion-time columns when comparing to your CRM.
A view-through conversion is credited when someone saw an ad, did not click, and converted later by some other route. These are a legitimate measure of influence, but they are also the easiest way for a platform to claim credit for a conversion it did not cause.
Here is a single person's journey with default settings on each platform.
| Day | Touchpoint |
|---|---|
| Day 20 | Sees a LinkedIn ad and does not click |
| Day 22 | Clicks a Meta ad and browses the site |
| Day 25 | Clicks a Google Search ad |
| Day 26 | Submits the contact form |
| System | Counts | Why |
|---|---|---|
| 1 conversion | View-through, within its 7-day default | |
| Meta | 1 conversion | Click within its 7-day window |
| Google Ads | 1 conversion | Click within its 30-day window |
| GA4 | 1 conversion | Assigned to one channel by its model, often the last click |
| CRM | 1 lead | One unique contact, source may be recorded as just one channel |
That is three platform conversions for one lead. Add up the ad platform dashboards and you have tripled the real number without any tracking being wrong. This is why ad platform totals should never be added together or treated as a company-wide lead count.
| Symptom | Likely cause |
|---|---|
| Platform conversions far exceed CRM leads | Duplicate firing, spam submissions, form-click tracking instead of true success, or overlapping platform credit |
| GA4 lower than Google Ads | Consent and script blocking in GA4, view-through or modeled conversions in Google Ads, differing attribution windows |
| Platform conversions higher than GA4 for social | View-through credit that GA4 cannot see |
| CRM leads higher than platform conversions | Phone, email, and walk-in leads that never touched the tracked form |
| CRM shows "direct" or blank source | Click IDs and UTMs not captured in hidden form fields |
| Numbers change after the reporting period closes | Conversions reported by click date and late-arriving conversions inside a long window |
Platforms also add modeled or enhanced conversions to recover activity that cookies and consent choices hide. These are estimates, so they increase platform totals without an equivalent record in the CRM. The same effect appears when a platform recovers conversions with hashed first-party data, discussed in why conversion tracking matters for paid campaigns.
If you are not sure which of these applies to your setup, an analytics audit compares GA4, your tags, your ad platforms, and your CRM side by side.
Google Ads credits conversions to ad clicks and views within its own conversion window, often 30 days, and reports them on the date of the ad interaction. GA4 uses its own attribution settings, only credits conversions tied to a session, and can assign the same conversion to a different channel.
Meta counts events such as form submissions that it can match to an ad click or view, including duplicates and spam, and does not know whether another channel also influenced the lead. A CRM counts unique records that passed its own rules.
There is no universal number. Platform totals are usually higher because of overlapping attribution, view-through credit, and duplicates. Track the ratio between platform conversions and CRM leads over time and investigate when it shifts.
Match the window to how long your customers actually take to convert. Long windows suit long sales cycles but inflate reported results when they credit clicks or views from weeks earlier. Keep views short and compare against CRM data before changing bidding.
Use your CRM for how many real leads and deals you have, GA4 for how visitors behave across channels, and ad platforms for in-platform optimization and trends. Each answers a different question.
Platform settings change often. Confirm current windows and defaults in each ad manager before changing a setup.
Start with an analytics audit to find where the numbers diverge, rebuild reliable events with website conversion tracking, or connect campaigns to closed revenue with marketing attribution.