B2B demand gen attribution. The guide that does not pretend it is solved.
Marketing attribution in B2B is a known-broken problem. The dashboards lie politely. The models flatter the wrong activity. The truth is messier. This page is the working version of how to think about it well enough to make decisions, without pretending you have solved it.
Why B2B attribution is harder than B2C
Three structural reasons.
B2B sales cycles are long. 90 days minimum, often 9-12 months. Most attribution models assume a 30-day window. The "click that closed the deal" was probably six months and twelve touches earlier than the model recognises.
B2B buying involves committees. 5-10 people influence the decision. Your dashboard tracks the one who clicked the form. The other 9 are invisible.
B2B buyers research in private. Slack DMs, podcasts, peer conversations, screenshots. Nothing trackable. Yet that is where most of the buying influence happens. The dashboard captures the breadcrumb at the end, not the journey.
What the major models actually tell you
Practical strengths and weaknesses.
Last-touch
Easy to set up, lies most. Tells you which channel got the click that converted. Ignores everything that built the buying intent in the first place. Useful for paid bottom-funnel, useless for everything else.
First-touch
Tells you which channel introduced the buyer. Useful for awareness budget allocation. Misses the 8-12 touches in the middle that closed the deal.
Linear / time-decay
Splits credit across all touches. More honest than first/last but still only sees what your tracking sees. Misses dark social entirely.
U-shaped / W-shaped
Weighted toward first and last with some middle credit. A reasonable compromise if you have to pick a single model. Still misses dark social.
Self-reported on demo form
A free-text "How did you hear about us?" field. Imperfect, but the only data point that captures dark-social channels. Run alongside another model for triangulation.
Multi-touch + 6-month windows
Best you can do without expensive tooling. Requires CRM hygiene most teams do not have. Worth the work for high-stakes channel decisions.
Practical recommendations
How to do this well enough to be useful.
First, run self-reported attribution on every demo and trial form. One question, free text. Costs nothing, captures dark social, builds a 90-day data set worth more than your dashboards.
Second, run two attribution models in parallel. A last-touch (or U-shaped) model for the channels you can measure cleanly, plus self-reported for the channels you cannot. Reconcile quarterly. Do not pick one and pretend it is the truth.
Third, expand your attribution window. Most B2B should run 90-day, 180-day, and 365-day attribution windows side by side. The 30-day default model will miss most of your demand gen impact and lead you to defund the activity that is actually working.
Fourth, do not believe any model that tells you a single channel is responsible. Demand gen is portfolio. The channels work together. Decisions like "should we cut content?" need broader signal than a single attribution model can provide.
Fifth, watch for the obvious patterns even when the model does not show them. Sales calls mention podcasts. Inbound DMs come from LinkedIn comments. Self-reported sources cite a community you forgot you posted in. The qualitative signal is real even when the dashboard cannot count it.
Common attribution failure modes
Worth being aware of.
Patterns that look broken
- Cutting podcast spend because it is not "tracked".
- Killing community presence because it is not "in the funnel".
- Believing your CRM's last-touch source as truth.
- Running 30-day windows on 9-month sales cycles.
- Attributing 100% of pipeline to your highest-volume tracked channel.
Smarter habits
- Self-reported attribution on every demo.
- Multiple model runs alongside each other.
- Long attribution windows (180-365 days).
- Quarterly channel-mix reviews, not monthly.
- Triangulate quantitative tracking with qualitative sales signal.
FAQ
Probably not unless you are above £25M ARR with a dedicated marketing ops function. The setup is heavy, the maintenance is heavier, the insight gain is marginal over self-reported + a single tracked model. Most companies should run cleaner data hygiene before adding more tooling.
For most B2B between £1M and £25M ARR, yes. With self-reported on demo forms layered on top. The bigger problem is not the tool, it is the attribution windows people set (usually too short).
Self-reported attribution. Show the percentage of pipeline that cites a source not tracked in the CRM. Show the trend. Most leaders accept the imprecision once the volume of "untracked" pipeline is visible.
Useful at scale (typically £20M+ marketing spend), expensive at smaller scale. Not the first thing to invest in for most B2B companies under £25M ARR.
Want to talk through your situation?
Thirty minutes with the team. We will pressure test your numbers, not pitch.
Book a strategy call →