First, check which count is higher and for which days
Start by making sure the two numbers describe the same thing. Campaign Manager's conversions here are website conversions recorded through the Insight Tag, defined, as the LinkedIn Ads hub explains, by a URL rule or an event-specific tag, each with a type, an optional value and attribution settings for clicks and impressions. If the numbers that disagree are Lead Gen Form leads rather than website conversions, the problem belongs to the leads not syncing diagnosis, which matches the two exports row by row. If Campaign Manager shows fewer conversions than the CRM, or none, the Insight Tag not tracking conversions diagnosis applies instead.
Then fix the outcome and the days. Name the single outcome both systems are meant to record, such as a demo request or a trial sign-up, and pull it from the CRM for a closed range of days, in one time zone, filtered to records with a LinkedIn source from UTM parameters or hidden fields. Pull the matching conversion from Campaign Manager for the same days. The two systems may place a conversion on different dates, and the sync diagnosis explains how date boundaries and time zones alone make honest counts disagree, so compare whole weeks or months rather than single days.
Finally, read the gap over time rather than once. A gap that has always been there and moves with spend is mostly counting rules. A gap that opened on a particular date points to a change: a new conversion, an edited window, a site release, a new form. The measurement change log is where that date should be found, and the five whys keeps the questioning going until the change is named.
Diagnosis table
Diagnosis
| Symptom | Check | Likely cause | Fix |
|---|---|---|---|
| The gap shrinks sharply when only click credit is counted | The conversion column split by click and view credit | View-through conversions counted beside clicks | Report view credit on its own line; keep it out of budget figures |
| Credited conversions arrive long after the last click; gap grows with time | The conversion's attribution settings and the date they were set; CRM time from first touch to outcome | An attribution window longer than the decision | Set the window to the decision time the CRM shows; record the change |
| Roughly double the CRM count from one date | Every conversion attached to the campaign and what each one records | Two conversions counting the same outcome | Keep one conversion per outcome attached; detach the other |
| More conversions than form records, including failed attempts | What fires the event tag; a failed test submission | An event that fires on the attempt, not the outcome | Fire the event on the confirmed outcome only |
| LinkedIn, other ad platforms and analytics each claim most of the same leads | Each platform's credited count against the CRM's total | Every platform crediting the same lead | Use the CRM for the outcome and a shared rule for channel credit |
| Gap made of internal names, junk entries and repeat submissions | CRM records removed or merged in the same days | Test, spam and duplicate submissions the CRM removes | Filter tests, block spam, dedupe before the tag fires where possible |
Fix, verify and prevent each cause
View-through conversions counted beside clicks
Each LinkedIn conversion carries attribution settings for clicks and for impressions, so the column can include members who saw an ad, never clicked, and converted later. A View-Through ConversionA conversion credited to an ad that was shown but not clicked, within a set view window. records that sequence; it does not show that the ad caused it, and many of those people would have converted anyway. The comparison of view-through and click-through conversions owns the weight each kind deserves.
Fix: split the column by click and view credit in Campaign Manager's reporting, report view-through conversions on a separate, labelled line, and base budget decisions on click credit read against the CRM.
Verify: click-credited conversions sit much closer to the CRM's LinkedIn-sourced records than the total did.
Prevent: decide once whether view credit is reported, where it appears and whether it counts for optimisation, and record that decision with its date.
An attribution window longer than the decision
The attribution settings decide how long after a click or an impression a conversion is still credited, within the options the help centre documents. A window much longer than the time buyers take from first contact to the outcome credits LinkedIn with conversions that followed long after its influence faded, often from people the CRM attributes to a later channel. Conversion LagThe time between an ad interaction and the conversion it leads to, which leaves the most recent reporting days incomplete. is the measure that matters here, and the LinkedIn Ads hub advises choosing windows before launch and keeping them stable.
Fix: read the CRM's time from first LinkedIn touch to the outcome for recent records, set the window to match that decision time rather than the longest available, and record the change.
Verify: after the change, the conversions credited to LinkedIn move closer to the CRM's count, and the trend line is broken at the date of the change.
Prevent: keep window changes rare, logged and reviewed with the person who reports the numbers, because a changed window changes the count without any buyer behaving differently.
Two conversions counting the same outcome
A campaign can carry more than one conversion, and it is easy to end up with two that record the same outcome: a URL rule on the thank-you page and an event-specific tag on the same submission, or a new conversion created during a site rebuild while the old one stayed attached. Each fires on the same visit, so Campaign Manager reports the outcome twice. The guide to conversion events across platforms owns the rule of one written definition per result and one source per result per platform.
Fix: list every Conversion ActionA Google Ads definition of one action to count, such as a purchase or lead, with its source, category, counting method, value and conversion window. attached to the campaign with what it records, keep one per outcome, and detach or archive the duplicate.
Verify: a single test submission now produces one conversion in Campaign Manager, and the column moves towards the CRM's count from the date of the change.
Prevent: keep a register of conversions, their definitions and the campaigns they are attached to, and check it whenever a conversion is created.
An event that fires on the attempt, not the outcome
An event-specific conversion counts whatever its trigger fires on. A trigger on the submit button counts every press, including submissions the form rejects for a missing field and double presses on a slow connection, while the CRM receives only the forms that succeeded. The Insight Tag not tracking conversions diagnosis owns the trigger fix and is not repeated here.
Fix: move the event to the confirmed outcome, usually a success event the site pushes when the submission is accepted, as the Insight Tag and conversions tutorial sets up.
Verify: a failed test submission fires nothing, a successful one fires once, and the column stops running ahead of the CRM by the number of failed attempts.
Prevent: treat the success event as part of the form's contract and test both branches whenever the form changes.
Every platform crediting the same lead
Campaign Manager sees only LinkedIn's own touches. A buyer who saw a LinkedIn ad, clicked a search ad and later came back from a newsletter may be credited in full by each platform that touched them, while the CRM records one lead. The guide to platform and analytics discrepancies explains why platform totals summed across channels exceed any single count of outcomes. LinkedIn's share of the journey may be real and still not be the whole credit; Assisted ConversionsConversions in which a channel appeared earlier on the path but was not the final interaction. and AttributionAssigning conversion credit to touchpoints. are the concepts at work.
Fix: stop summing platform conversions into a total, use the CRM's count as the outcome, and agree one shared rule for how credit is read across channels, supported by self-reported attribution where a form can ask how the buyer heard about the business.
Verify: the monthly report shows one total from the CRM and each platform's claim as context beside it, and nobody adds them up.
Prevent: put the rule in the reporting template, so it survives a change of analyst or agency.
Test, spam and duplicate submissions the CRM removes
The tag records whatever reaches the confirmation step: colleagues testing the form, an agency checking a launch, bots that complete forms, and people who submit twice. The CRM's spam filters and duplicate rules remove or merge many of these, which the leads not syncing diagnosis describes from the CRM side.
Fix: agree how internal tests are run and marked so they can be excluded, add spam protection to the form, and where the site can, show the confirmation page only for accepted, non-duplicate submissions.
Verify: the CRM's removed and merged records for the period fall, and the gap shrinks by about the same amount.
Prevent: keep a short list of who tests forms and when, and read removed records monthly.
Prevention: a written rule for which system answers which question
The gap between Campaign Manager and the CRM never closes completely, and it does not need to. What causes trouble is a team that reads whichever number suits the moment. Write a short rule and keep it with the monthly report. The CRM answers how many outcomes the business received and what they became: demos held, opportunities, revenue. Campaign Manager answers which LinkedIn campaigns, ads and audiences did better than one another, using click credit on fixed windows. View-through conversions appear on their own labelled line. Platform totals are never summed. Cost per outcome is calculated from spend and the CRM's count, which the CPA calculator handles, and the LinkedIn variant of CPA in the glossary explains why the platform's own cost per conversion counts a different thing.
Around the rule, keep three habits. A register of every conversion with its definition, attribution settings, the campaigns it is attached to and the date each was set. A change log in which any edit to a conversion, a window, a form or the site is recorded on the day it happens. And a monthly comparison of the two counts for the same closed range, read against the previous months, so a new gap is noticed when it opens rather than at the end of the quarter. The LinkedIn for B2B hub sets out how to measure pipeline rather than clicks, and the LinkedIn troubleshooting hub routes other measurement symptoms.
Worked example
Qeltrona Cloud Backup reconciles its demo requests
Qeltrona Cloud Backup, an invented cloud backup provider, sees Campaign Manager credit LinkedIn with 96 demo requests in March while its CRM holds 31 demo requests with a LinkedIn source for the same month. Splitting the column shows 58 click-credited and 38 view-credited conversions. The conversion list shows two conversions attached to every campaign: a URL rule on "/demo/thanks" created at launch and an event-specific conversion added during a site rebuild in January, both recording the same confirmation. A single test submission produces two conversions. The team detaches the URL rule, leaving 29 click-credited demo requests once the duplicate is removed, which matches the CRM's 31 after two internal test records are excluded. It moves the 19 remaining view-credited conversions to a separate line in the monthly report, and agrees that budget decisions use CRM demos joined to campaigns by UTM values. In April, Campaign Manager reports 34 click-credited demo requests against 33 in the CRM, and the cost per CRM demo, at £410, becomes the figure the team manages.
Common mistakes
Comparing Lead Gen Form leads with website conversions as if they were one count. Adding every platform's conversions into a single total. Leaving view-through conversions mixed into the figure used for budget decisions. Choosing the longest attribution window because it makes LinkedIn look better. Creating a new conversion in a rebuild without detaching the old one. Firing the event on the submit button. Counting colleagues' test submissions as demand. Changing a window or a conversion without recording the date. Treating any gap as proof that one system is wrong, rather than writing down which system answers which question.
