Your CRM records 80 enquiries while the advertising dashboard reports 96 conversions. That difference does not, by itself, show that the campaign is underperforming or that someone has installed the tracking incorrectly.
The systems may be counting different actions, using different dates or applying different attribution rules. A useful reconciliation identifies those differences before the business changes its campaigns.
Use the process below to establish a comparable reporting scope, explain the major gaps and decide which problems need a repair.
Define four numbers separately
Do not put every stage under a column called “leads.”
| Number | Working definition | What it does not establish |
|---|---|---|
| Conversion events | Recorded actions such as successful submissions or calls | The number of distinct prospective customers |
| Distinct enquiries | Enquiries after an agreed rule for duplicate records | Whether each enquiry is suitable |
| Qualified leads | Enquiries meeting documented service and customer criteria | Whether the business will win the work |
| Won customers | Sales confirmed under an agreed reporting rule | Which ad interaction deserves all the credit |
A person may submit two legitimate enquiries about different jobs. A duplicate delivery of the same form is a different situation. Decide how those cases are handled before calculating lead costs.
Also separate advertising-attributed outcomes from all enquiries. The CRM may contain referrals, organic visits, existing customers and direct contact. It is not automatically the denominator for an advertising cost-per-lead calculation.
Freeze a comparable reporting scope
Write a short scope statement at the top of the worksheet:
“Compare successful new-business enquiry submissions received during this period, for these services and locations, using this time zone. Exclude tests and known duplicate deliveries. Review qualification status as of this later date.”
Then document:
- The conversion actions included in the advertising export.
- Whether dates refer to ad interactions, completed conversions or CRM creation.
- The relevant attribution and conversion windows.
- Whether CRM records were created manually or imported later.
- The cut-off date for qualification and sales updates.
Google's offline-conversion guidance specifically advises comparing conversion-time reporting with successfully processed upload results. Submitted upload rows can include rejected records.1 Use the reporting view that fits the question, and label it clearly.
Build a worksheet around exceptions
Where a suitable identifier is available, match event and CRM records through a controlled process. Use access controls and approved handling for customer data. An internal review sheet does not need to expose full contact details to every reader.
This is a synthetic example, using invented IDs:
| Enquiry ID | Submitted | CRM record | Outcome | Reconciliation action |
|---|---|---|---|---|
| E-001 | 3 October | Created on 3 October | Qualified | No intake discrepancy; review attribution separately |
| E-002 | 3 October | Two deliveries of one submission | One valid enquiry | Apply the agreed duplicate rule and investigate repeated delivery |
| E-003 | 4 October | No record found | Unknown | Check notification, integration and submission logs |
| E-004 | 4 October | Created on 6 October | Outside service area | Correct the date interpretation and record the rejection reason |
| E-005 | 5 October | Present | Qualified, upload rejected | Resolve the import error before expecting it in the processed results |
An absent CRM record is an operational question before it is a campaign-quality judgement. A rejected offline upload is a measurement question before it is evidence of poor sales performance.
Explain differences in separate buckets
A variance register makes the mismatch useful:
| Difference | What to check | Likely next action |
|---|---|---|
| Timing | Time zones, interaction dates, conversion dates and delayed updates | Re-export on a comparable date basis |
| Action definition | Form success versus button click, calls versus qualified calls | Correct labels or tracking logic |
| Scope | All enquiries versus advertising-attributed enquiries | Restrict or separately report the comparison set |
| Repetition | Repeated tags, duplicate deliveries or repeated genuine enquiries | Fix the fault or apply a documented counting rule |
| Attribution | Multiple platforms claiming influence, different models or windows | Keep platform and business totals clearly labelled |
| Import processing | Accepted, rejected and pending uploads | Investigate results and repair eligible records |
Do not force every variance into a single cause. A campaign can have duplicate events and unsuitable enquiries at the same time.
Use IDs to solve the right problem
A consistent event or transaction identifier can help deduplicate supported conversion recording. Google explains that repeated submissions with the same transaction ID for the same conversion action are treated as duplicates; a static ID reused for unrelated transactions can undercount them.2
That mechanism does not establish a unique customer identity or decide whether two jobs from the same customer should count separately. Those are business and CRM rules.
Agree on what each identifier represents: an enquiry, a booking, a sale or an event. Keep it stable across the systems involved in that specific process. Check the requirements of each integration before assuming that an ID field works identically everywhere.
Accept the limits of individual matching
Some advertising results are modelled or aggregated. You cannot assume that every reported conversion has a corresponding row that can be matched to an identifiable person.
Keep two layers of reconciliation:
Operational reconciliation checks whether real submissions arrived, records were retained, duplicates were controlled and qualification stages were recorded.
Advertising reconciliation checks reporting definitions, attribution scope, processing results and plausible aggregate differences.
The aim is an explained, documented comparison. Exact numerical equality is only meaningful when the systems genuinely share the same scope and measurement basis.
Change campaigns after the diagnosis
If genuine enquiries fail to reach the team, repair the handoff. If conversion events repeat, repair the tracking. If correctly recorded enquiries consistently request the wrong service, investigate targeting, search intent and messaging.
For performance decisions, report at least distinct relevant enquiries, qualified leads and acquisition costs with their definitions. Add won-customer outcomes when the sales cycle supports a meaningful comparison.
A conversion-tracking review can help build this measurement map. For the commercial side of the decision, read lead volume and lead quality before judging success by a single conversion total.