Marketing attribution has one job in a growing B2B company: enough truth to reallocate budget every month. Not a courtroom, not a physics model — a decision aid. Most attribution projects fail by aiming higher than that and shipping nothing. Here is the build that survives board scrutiny precisely because it is honest about its limits.
The job attribution actually has
The board question is not “which click caused this deal?” It is “if we moved £5k a month from channel A to channel B, would pipeline improve?” Attribution exists to answer that — which means channel-level cost per opportunity, trended, with known blind spots labelled. Precision beyond decision-usefulness is expensive theatre.
Models, compared honestly
- Last click: free, built into ad platforms, systematically flatters capture channels. Fine as a signal, dangerous as the truth.
- First/CRM source + self-reported: the working choice at this size — one source per opportunity, plus the buyer’s own answer at the form. Cheap, explainable, durable.
- Multi-touch: credible only with high deal volume; below that the fractional credits are noise wearing decimals.
- Media mix modelling: the right answer at a scale most £1m–£20m companies have not reached; revisit past seven figures of annual spend.
The working build
- UTM discipline on everything outbound — our UTM link builder exists for exactly this.
- Source on the opportunity, set at creation, locked thereafter. Contact-level source is trivia; opportunity-level source is budget truth.
- Self-reported attribution — a free-text “how did you hear about us?” on demo and contact forms. Ten answers a month re-weight a quarter of click data.
- Offline conversion upload to ad platforms (e.g. Google Ads offline conversion imports) so bidding optimises to qualified pipeline, not form fills — standard practice in our Google Ads engagements.
- One monthly view: spend, opportunities, cost per opportunity, win rate — by channel, trended.
Honest limits
Dark social, word of mouth, and demand creation will always be under-counted by click models; long cycles smear credit across quarters; and small samples make monthly noise look like trend. The credible move is labelling these limits on the dashboard itself. What destroys trust is not imprecision — it is imprecision presented as certainty.
Reporting to the board
Same four columns every month: pipeline created, cost per opportunity, win rate by source, spend vs plan — plus one annotation line for what changed. Attribution feeds the CAC arithmetic the board actually steers by, which is covered in CAC payback period. Consistency of definition is the entire credibility game: the fastest way to lose a board is to re-define “source” twice in a year.
FAQ
What is marketing attribution?
Marketing attribution is the practice of assigning credit for pipeline and revenue to the marketing touches that contributed to it. Its real job is smaller than the vendor pitch: give a growing company enough truth to reallocate budget monthly. Any model that achieves that is working; any model that does not is decoration.
Which attribution model should a B2B company use?
For most £1m–£20m B2B companies: CRM source on every opportunity (first meaningful touch), plus self-reported attribution at the form, reviewed as cost per opportunity by channel. Multi-touch models add little at this deal volume — the sample sizes are too small for the extra precision to be real.
Why do sales and marketing report different numbers?
Usually because there is no single agreed definition of an opportunity source, so each team counts what flatters it. Fix the definition, tag at the opportunity (not the contact), and make one dashboard canonical. The argument is almost never about maths; it is about missing agreements.
How should attribution handle demand creation like content and brand?
Accept that click-based models under-report it, and compensate deliberately: self-reported attribution (“how did you hear about us?”), branded search volume as a proxy, and patience measured in quarters. Cutting creation because last-click cannot see it is the most common measurement-driven mistake in B2B.
Key takeaways
- Attribution’s job: enough truth to reallocate budget monthly
- At this size: opportunity-level CRM source + self-reported beats multi-touch
- Upload offline conversions so ad platforms optimise to pipeline
- Label the blind spots on the dashboard — certainty theatre kills trust
- Keep definitions stable; the metric feeds CAC payback governance
