Your CEO asks which channels are generating pipeline. Your CMO sends a slide where 40% of pipeline is “Marketing” and 35% is “Direct” and 15% is “Other.” That slide is fiction. You know it. The board will ask the same question next quarter and you will be the one who answers.
I am George, founder of Leadpipe. Before Leadpipe I ran sales and partnerships for other companies, and I have seen the same pipeline-source slide fail in every boardroom for different reasons. Here is how a CRO runs an honest audit, what to fix, and how visitor identification makes the next audit faster.
Who this post is for
You are a CRO, VP Revenue, or head of GTM at a company between $5M and $75M ARR. You have a CRM that was set up by someone who left 2 years ago, at least one paid channel that is a black box, and a sales team that closes deals without logging touches. You own the pipeline number.
The answer up front: most pipeline-source reports are wrong in 3 specific ways. Self-reported attribution is noisy, last-touch attribution is biased toward whatever channel is closest to the form, and the anonymous research phase is invisible. A good audit starts with the CRM fields, moves to the attribution model, then adds person-level behavioral data from visitor identification to close the gap.
The audit in 4 phases
| Phase | Duration | Output |
|---|---|---|
| 1. CRM field audit | 1 week | Field map, bad-data list |
| 2. Attribution model teardown | 1 week | Honest channel-share report |
| 3. Anonymous pipeline overlay | 2 weeks | Influenced pipeline, not just sourced |
| 4. Playbook rewrite | 2 weeks | New routing, scoring, reporting rules |
6 weeks. Skip a phase and the audit is worthless. Don’t let RevOps or the agency running paid media run it solo. A CRO needs to sit in every working session.
Phase 1: CRM field audit
Pipeline source numbers are built on 3 or 4 CRM fields. If those fields are wrong, the whole report is wrong. Start here.
Fields to audit:
- Lead Source
- Campaign / Source Campaign
- First-touch Channel
- Last-touch Channel
- Original Referrer / UTM fields
For each field, pull a 12-month distribution and flag anything that looks suspicious.
| Field | Warning sign | Likely cause |
|---|---|---|
| Lead Source | >40% “Other” or “Unknown” | Form picker misconfigured |
| Campaign | Majority “null” | UTMs not being captured |
| First-touch | Majority “Direct” | Referrer strip or cookie loss |
| Last-touch | 60%+ one channel | Rule overrides real data |
| UTMs | Inconsistent naming | No governance |
If any single warning sign applies, stop using that field in your board slide. It is not measuring what you think. The post on Salesforce being full of bad data has concrete cleanup rituals.
Deliverable for phase 1: a one-page map of which fields you trust, which you don’t, and what cleanup is scheduled.
Phase 2: attribution model teardown
Most CRMs run on last-touch or first-touch. Both lie about different things.
Last-touch attribution lies when:
- A prospect did 3 months of research before filling out a demo form from a branded Google search.
- The channel closest to the form (paid brand, chatbot, SDR outreach) gets credit for demand generated elsewhere.
First-touch attribution lies when:
- A prospect clicked an ad 14 months ago, forgot, and came back through a podcast they heard last week.
- Early-funnel content gets credit for revenue it didn’t actually move.
Self-reported attribution lies when:
- The prospect picks “Google” because they don’t remember, even though they actually heard about you from a friend.
- The dropdown has 7 options and the 8th cause is missing.
None of these are malicious. They are structural. A CRO audit accepts the structure and compensates for it.
The teardown output:
Pick 20 closed-won deals from the last 2 quarters. For each, talk to the AE and reconstruct the actual buying journey. Write it as a timeline: first impression, second impression, first conversation, first serious evaluation, close. Now compare to what the CRM says the source was.
Pattern you will find: the CRM understates content, SEO, word-of-mouth, and the anonymous research phase. It overstates whatever channel is closest to the form. We wrote about this in Google Analytics is lying about your pipeline.
Phase 3: anonymous pipeline overlay
This is the phase that most CROs have never run, because the data didn’t exist 5 years ago.
Install visitor identification and overlay identified visits onto your CRM records. For every open opportunity and closed-won deal in the last 12 months, ask:
- How many identified website visits existed on this account before the deal entered the pipeline?
- Which pages were viewed?
- How many unique people from the account visited?
- Did the economic buyer visit before, during, or after the deal was in active conversation?
What you will find on an honest overlay:
| Segment | Typical finding |
|---|---|
| Closed-won enterprise | 8-15 identified visits from 3-5 people before first meeting |
| Closed-won mid-market | 3-7 identified visits from 1-3 people |
| Lost deals | Often 20+ visits, including heavy comparison-page activity |
| Pipeline “sourced by BDR” | 40%+ already had identified visits, BDR was the trigger not the source |
If you run this on 12 months of data, you will change how your team talks about pipeline. The honest story is rarely “BDRs sourced this” or “paid media sourced that.” It is “content and SEO surfaced the account, BDR converted the conversation, enterprise marketing kept them warm, AE closed.”
That story is inconvenient but true. A CRO who can tell it earns the board’s trust.
Phase 4: playbook rewrite
After the audit, 3 things should change.
Routing changes.
Any account with 5+ identified visits in the last 30 days should be flagged as in-market and routed to a named rep, not a round-robin pool. High-intent visits (pricing, comparison, demo) trigger same-day alerts. Low-intent visits trigger weekly nurture. See how to track when target accounts visit your site.
Scoring changes.
Rewrite lead scoring to include behavioral weight, not just firmographic. A perfect-ICP lead with 0 site visits is cold. An ICP-fit lead with 4 recent pricing-page visits is hot. Most models reverse this.
Reporting changes.
Your board slide now has 3 columns, not 1:
- Sourced pipeline (first identifiable touch).
- Influenced pipeline (any identified touch during the deal).
- Accelerated pipeline (identified touches in the 30 days before close).
The same deal can appear in all 3 columns with different channel credit. That is the honest picture.
Questions a CRO should ask in the audit
Running the audit is a series of specific questions. Here are the ones that surface the most value.
- What percentage of our pipeline comes from accounts that had identified website activity before the first meeting?
- Of our closed-lost deals, what did they visit on our site after losing? (Many come back.)
- What is the average time from first identified visit to first booked meeting?
- What is the lead-to-opportunity rate for accounts with 5+ visits vs 1-2 visits?
- Which pages are visited most by our highest-ACV customers in the 90 days before close?
- Which paid channels generate identified traffic vs noise? (Most LinkedIn ads pull ICP. Most display pulls noise.)
Each question is answerable if the data exists. If it doesn’t, your audit has uncovered the data gap.
What NOT to do as a CRO in this audit
- Don’t run it in parallel with a re-org. You need stable ownership to finish the 6 weeks cleanly.
- Don’t let the agency or a single vendor own the findings. Conflict of interest on every line.
- Don’t claim certification you don’t have. We track pipeline sources and compliance separately. For compliance, see GDPR-compliant visitor identification.
- Don’t kill the channel before you isolate the variable. If paid LinkedIn looks bad in the audit, check whether the UTMs were broken for 3 months before you cut the budget.
- Don’t present the audit as “marketing’s report card.” It is a diagnostic. Marketing, sales, and RevOps all share findings.
The CFO version of the audit
If the board or the CFO wants one number, give them this:
| Metric | Formula | What it tells you |
|---|---|---|
| True marketing influence | Pipeline with any identified visit / Total pipeline | Usually 60-80% once you look |
| Source honesty | Sourced pipeline / Influenced pipeline | Close to 1 means your model is believable |
| Win-rate lift | Win rate with 5+ visits / Win rate with 0-2 visits | Usually 2-3x |
These 3 numbers, tracked quarterly, replace the pie chart that lies.
What good looks like after the audit
The board meeting changes. Instead of defending the source column, you narrate the buying journey. ‘These 12 opportunities all had 6+ identified visits on the site before we picked up the phone. Marketing worked. Sales converted. Here’s the next 30 accounts we see researching.’ That’s a CRO in control of the data, not buried under it.
A CRO who has run this audit once runs it twice as fast the second time. The payoff compounds.
Leadpipe identifies 30-40%+ of your US B2B visitors with full contact data on the Pro plan at $147/mo. No credit card to start the 500-lead trial. Start identifying visitors →
