How to Audit Marketing Attribution for B2B Sales Cycles

What an attribution audit is for

Most B2B organizations already have attribution of some kind. The ad platforms report conversions. GA4 assigns key events to channels. The CRM has a lead source field. Someone builds a slide each quarter that says which channels produced pipeline.

An attribution audit asks a harder question: how much of that slide is true? Not whether the tools are installed, but whether the credit they assign reflects how your buyers actually decide, and whether leadership is making budget calls on evidence or on artefacts of how the software counts.

The audit matters most where sales cycles are long, deals involve several people and much of the persuasion happens offline. That describes most professional services, engineering, manufacturing, software and public-sector suppliers in Atlantic Canada, where a contract may follow a referral, a conference conversation, two webinars, a site visit and an RFP. If your reporting cannot yet connect leads to outcomes at all, start with connecting marketing reporting to sales outcomes. The audit assumes that connection exists and tests how far to trust it.

Step one: map how buyers actually decide

Before opening any software, take a sample of recent won and lost opportunities, perhaps a dozen of each, and reconstruct how they happened. Talk to the sales people who worked them. Where possible, ask the client.

For each opportunity, note:

• Who was involved on the buyer’s side, and which of them ever touched your website or ads.
• The first contact anyone at your firm remembers, and how it happened.
• The touches in between: content read, events attended, calls, proposals, referrals, reference checks.
• How long it took from first known contact to decision.
• What the client says made the difference, if they will tell you.

Now compare that picture with what the CRM and analytics say about the same deals. The differences are the audit’s first finding. Typical ones: the person who filled in the form was a junior researcher, not the decision-maker; the “source” recorded was the last click before the form, while the relationship began at an industry dinner two years earlier; the deal came through an RFP that no tracking will ever attribute to anything.

None of this means digital attribution is worthless. It means it sees a slice of the journey, and the audit needs to know how big the slice is.

Step two: test the data chain, link by link

Attribution is only as good as the weakest link between first visit and closed deal. Check each one with real records, not with the settings screen.

Source capture at entry

• Do all forms populate hidden fields for source, medium, campaign and click ID? Test each form with a tagged link.
• Is first-touch source preserved, or overwritten when the visitor returns through a different channel?
• What share of recent leads have an empty or generic source such as “website”? Check by reading a sample of records rather than estimating.
• Are phone, email and event leads given a source by a consistent rule, or by whoever enters them?

CRM structure and hygiene

• Are contacts linked to accounts, and opportunities linked to the contacts involved? Without that, a deal cannot be traced back to the marketing touches of the people behind it.
• Are opportunity stages, close dates and values maintained, or updated in a burst before quarter-end?
• Are lost deals recorded with reasons, or simply abandoned?
• When records are merged, is the original source kept?

Feedback to the ad platforms

• If you import qualified leads or closed deals into Google Ads, LinkedIn or Meta, check that imports are actually running, matching and arriving within the platform’s time limits relative to the original click.
• Confirm that what is imported matches the definitions your sales team uses, and respects the consent the visitor gave.

Our guide to measuring lead quality from paid campaigns covers how this chain should be built. The audit checks whether it still works.

Step three: check the models against your sales cycle

Every system that reports attribution makes assumptions. The audit writes them down and compares them with the journey from step one.

• Attribution windows. Ad platforms only credit conversions that happen within a set window after a click or view, and those windows are capped. If your typical cycle from first visit to qualified opportunity is longer than the window, the platform cannot see most of what it contributed, and will under-credit channels that start relationships while over-crediting those that close them.
• Which conversion is counted. If the ad platforms optimize toward form fills, they are judged on the earliest and least meaningful step. A channel that produces many cheap, poor inquiries will look better than one that produces few strong ones.
• Model choice. First-touch, last-touch and data-driven models answer different questions. GA4’s data-driven model is better informed than a simple rule, but it sees mainly web sessions from consenting visitors and nothing that happens after the form.
• The CRM source field. Usually a single value, which is first-touch or last-touch by accident rather than by decision. Find out which.
• Self-credit. Each ad platform credits itself for any conversion it touched. Add their reported conversions together and you will often exceed the number of leads that exist.

The finding here is often uncomfortable: the channels that look best in reports may simply be the ones closest to the form. Branded search, retargeting and direct traffic tend to collect credit for demand that other activity created. LinkedIn, events and content tend to be under-credited, because their effect is early and spread across people who never click.

Step four: account for what tracking cannot see

A B2B journey contains touches no tag will capture: a partner’s recommendation, a podcast, a post someone read without clicking, a conference panel, a reference call, a procurement portal. Three methods recover some of it.

• Self-reported source. A required, open-text “How did you hear about us?” field on inquiry forms, and the same question asked by sales on the first call. Free text is messier than a drop-down and much more honest. Compare its answers with the tracked source; where they disagree, the self-report is often closer to the truth about where the relationship started.
• Sales notes, structured lightly. A short field on the opportunity for “influences mentioned by the client”. Review it quarterly.
• Account-level engagement. Whether people from a target account have visited, attended or engaged before an opportunity opened, rather than whether one individual clicked one ad.

In small Atlantic Canadian markets, where a meaningful share of business comes from people who already know someone at your firm, this layer often matters more than anything the ad platforms report.

Step five: grade the uncertainty instead of hiding it

The audit should end with a rating for each channel’s reported contribution, not a corrected ROI figure. A simple three-level scale works:

• Well supported. Tracked source, self-reported source and sales recollection broadly agree, and the data chain was tested.
• Directional. The tracking is sound but sees only part of the journey, or the sample is small. Use it to compare periods, not to set budgets precisely.
• Weak. The credit depends on a broken link, a short window or a model assumption the journey review contradicts.

Where a budget decision depends on a weakly supported number, test instead of arguing. Pause or reduce a channel in one region or one quarter, hold others steady, and watch qualified opportunities rather than form fills. With long cycles the test needs patience, and small volumes make results noisy, so agree in advance what you will look at and for how long. A test that is directionally clear is worth more than a model that is precisely wrong.

What the audit should produce

A useful attribution audit fits in a short document with four parts:

1. The journey picture: how buyers actually decide, from the opportunity review.
2. The broken links: specific fixes to forms, CRM fields, imports and windows, each with an owner.
3. The confidence grades: each channel’s reported contribution, rated, with the reason.
4. The reporting change: how pipeline reporting will present attribution from now on, including the caveats that go on the page.

That last part usually lands on the executive marketing dashboard, and often prompts a parallel conversation between marketing and sales about shared definitions, covered in aligning sales and marketing in B2B. If the audit turns up large gaps between the CRM and GA4 along the way, why GA4 and your CRM report different lead numbers explains how to reconcile them.

Attribution audits are part of our analytics and measurement work, and the most valuable finding is rarely a technical one. It is the moment leadership sees that the channel they were about to cut was opening the conversations the channel they were about to fund was closing.