Attribution is not about perfect truth. It is about reliable truth. If you cannot trust the direction of your data, every optimization is a gamble.
The problem is that attribution breaks in quiet ways. A tag stops firing. Consent settings change. A CRM field mapping breaks. A new landing page launches without UTMs. Everything still looks normal, but the story is wrong.
Sanity checks are how you stop trusting “pretty dashboards” and start trusting reality.
Sanity check 1: Do lead counts match within a reasonable range?
GA4 will never match the CRM perfectly. That is normal. What is not normal is a large gap that nobody can explain.
Common reasons for mismatch:
-
different definitions (GA4 counts form submits, CRM counts created leads)
-
duplicates or deduplication differences
-
consent and ad blockers
-
missing UTMs or broken mappings
-
offline steps that GA4 does not see
You are not chasing identical numbers. You are chasing explainable numbers.
Sanity check 2: Can you trace a lead end to end?
Pick five recent leads and trace the path:
-
click source
-
landing page
-
form submit or booking
-
CRM record created
-
owner assigned
-
outcome logged
If you cannot trace it quickly, your system is too fragile.
Sanity check 3: Are sources and channels standardized?
If the same traffic appears as facebook, fb, meta, and paid_social, you do not have attribution. You have labeling chaos.
Standardize:
-
utm_source values
-
utm_medium values
-
channel grouping rules
-
CRM source mapping
The goal is one language across tools.
Sanity check 4: Is deduplication defined and consistent?
If a lead submits twice, what happens?
If a webhook retries, what happens?
Deduplication decisions affect:
-
conversion rates
-
cost per lead
-
pipeline reporting
-
sales workload
Define the rule, implement it, and document it.
Sanity check 5: Are outcomes connected back to campaigns?
If you do not store UTMs in the CRM, outcomes cannot connect back. That makes revenue optimization impossible.
At minimum:
-
store UTMs at lead creation
-
log outcomes consistently
-
report qualified and won by campaign
If you want to go further, you can push offline conversions back to ad platforms where appropriate, but even without that, CRM reporting should be truthful.
Sanity check 6: Do you have a QA routine?
Tracking is not “set and forget.” It is “set and verify.”
A practical routine:
-
weekly quick check: event firing, lead creation, error logs
-
monthly deeper audit: mappings, consent impacts, attribution trends
-
after any website or CRM change: test the full path
Most attribution problems appear right after changes.
The most common attribution lies
If you see these patterns, do not optimize yet. Fix measurement first.
-
direct traffic spikes after a site update
-
paid conversions drop without a reason
-
CRM lead source is blank too often
-
one channel suddenly takes credit for everything
-
booked calls increase but CRM leads do not
Attribution is only useful when it is stable enough to guide decisions.
Why this is SEO, AEO, and GEO friendly
People search for “tracking audit,” “GA4 vs CRM mismatch,” and “attribution not accurate.” This post answers with concrete checks that can be implemented quickly, which makes it easy to reference.