B2B Marketing Attribution: Models, Setup and How to Know Where Your Pipeline Comes From
Attribution arguments tend to start with the model. Should it be W-shaped or linear? Meanwhile three people are tagging LinkedIn three different ways, and some of the forms don’t save UTMs at all.
Marketing attribution is supposed to tell you which programs create pipeline. In the companies I’ve worked with, the tools were all there and running: GA4, HubSpot, Salesforce, the ad platforms. They just don’t agree. So budget follows whichever dashboard looked best last month, and the channel that was doing the work gets cut because nobody could prove it.
My view: attribution is a plumbing problem before it’s a model problem. It’s a chain. Tag the link, capture the source, keep it in the CRM, agree on a model, report it. Break one link and every report after it is wrong, no matter how nice the chart looks. This guide walks the chain in order, the way I’d check it on a paid engagement.
What is B2B marketing attribution? B2B marketing attribution is how you connect pipeline and revenue back to the marketing that created them. It assigns credit for each deal to the campaigns, channels and touchpoints that influenced it, so you can see which programs produce pipeline and which only produce activity.
1. It starts with how you tag links
A model is math applied to your data, and if the data going in is a mess, the model is garbage with extra steps. The problem I find first, almost every time, is UTMs: no convention, nobody in charge of it, and three people tagging LinkedIn three different ways. Fix the naming before you touch the model.
A working UTM setup has four parts:
- A locked naming convention. Every value lowercase, underscores instead of spaces, and a fixed list of allowed sources and mediums.
linkedin, neverLinkedInorlinkedin_ads. - One place links get built. A shared builder with dropdowns enforces the convention at the moment of creation, which beats cleaning it up later.
- Hidden fields on every form. UTMs in the URL do nothing for your CRM unless every form captures them into contact properties.
- Tags on everything, paid or not. Email, organic social, partner links, events and newsletters all need them. Untagged clicks lose their campaign detail and show up as referral, organic social or direct, depending on what the browser passes along.
One rule matters more than the rest: first-touch source never gets overwritten. Keep a separate set of original UTM fields that automation fills once, while they're empty, and keep them off your forms, so you always know what brought someone in.
Most first-touch data dies on a form that never saved the UTMs. Check that before anything else.
Read the full breakdown: The UTM Strategy Most B2B Teams Skip →
2. Then pick an attribution model that fits your stage
Once your data is clean, the model question is worth answering. An attribution model is the rule for splitting credit across touchpoints. First touch and last touch are single-touch models. Linear, time decay and W-shaped are multi-touch attribution models. Each tells a different story, and the right one depends on your sales cycle and deal size, not on what a vendor demo made look exciting. Start simple and add complexity when the data can carry it.
| Model | What it answers | Use it when |
|---|---|---|
| First touch | What brought them in? | You're judging acquisition channels |
| Last touch | What converted them? | Short cycles with few touchpoints |
| Linear | What touched them along the way? | You're early and want a fair, simple baseline |
| Time decay | What touched them closest to the deal? | Short buying windows where recent touches matter most |
| W-shaped | What found them, converted them, and moved them to sales? | Cycles over 60 days with a defined sales process and clean data |
Whichever you pick, write down why you chose it and what it can't see. Every model has blind spots. The trouble starts when someone treats the model as the whole truth in front of the CFO.
Read the full breakdown: Marketing Attribution Models Compared →
3. Make your two systems agree
If HubSpot says one number and Salesforce says another, attribution is dead on arrival, because nobody will trust either. The disagreements are predictable, and they're fixable once you know where to look.
- Lead source doesn't map. HubSpot's source values don't match Salesforce's picklist, so everything lands in a catch-all like "Web."
- Lifecycle stage and lead status drift apart. Marketing sees an MQL while sales has already marked the lead unqualified.
- Sync errors pile up. A required Salesforce field or a missing picklist value blocks records from syncing, and nobody checks the error log.
- Duplicates split the history. One person exists as two records, so engagement and form fills live in different places.
This is also why HubSpot attribution reports and Salesforce campaign influence don’t match. They're built on different fields, and the sync between them leaks.
The fastest check: pull 50 records that exist in both systems and compare lead source, lifecycle stage and company. The mismatch rate tells you how bad it is.
Read the full breakdown: Why Your HubSpot and Salesforce Don't Agree →
4. Close the tracking gaps you can't see
The attribution failures that cost the most are the ones you can’t see. A visitor moves from your site to an app, a scheduler or a landing page tool on another domain, or lands on a blog post and converts two pages later, and the source gets lost between the click and the form. You never see it in a report. You just see conversions credited to your own site, or to direct.
Three checks find most of it:
- Watch your direct share over time. There's no universal "too much." A jump after a launch, a new domain or a consent change means something upstream is dropping sources.
- Run the GA4 hostname report. Compare every domain sending data against every domain you run. Anything that gets traffic but isn't listed is probably tracking dark. Before you assume that, rule out a consent banner blocking it, a filter excluding it, or a tag sending to a different GA4 property.
- Check every subdomain for all three layers: the GA4 tag, the HubSpot tracking code, and the UTM hidden fields on its forms. App, landing page, blog and help subdomains are the usual gaps.
In the audits I’ve run, the fix is small: a missing tag, a domain list in GA4 for separate domains, or a root-domain cookie your forms can read.
Read the full breakdown: The Hidden Cost of Subdomain Tracking Gaps →
5. Then check whether it’s working
Clean attribution lets you ask the question that matters: is demand gen producing pipeline, or just activity? The numbers that answer it aren’t on most marketing reports.
- Pipeline by source. Leads are an input, and pipeline is what they turn into.
- Stage conversion rates, trended monthly. When one drops, something changed upstream.
- Time to pipeline. How long from first touch to open opportunity. It warns you before the pipeline number does.
- Sourced and influenced, side by side. Sourced shows acquisition, and influenced shows marketing's role across the whole cycle.
Then add one qualitative input: an open-text "How did you hear about us?" field on high-intent forms. When people keep naming a podcast your attribution says drove nothing, the podcast is probably working and the tracking is what’s broken.
Read those answers yourself once a month. It takes twenty minutes.
Read the full breakdown: How to Know If Your Demand Gen Is Working →
6. And report it so leadership believes you
The payoff for all this plumbing is a dashboard your CEO reads and believes. The dashboards I inherit tend to have 30 charts, and leadership reads five of them on a good day. So build the five.
Five numbers cover it: marketing-sourced pipeline, win rate on that pipeline, pipeline-to-spend ratio, MQL to SQL conversion, and sales cycle length. Write a definition for each, and keep sourced and influenced pipeline separate. Build the dashboard where sales already looks, which for most of the teams I’ve worked with is Salesforce, so both teams see the same numbers. Then add a half-page monthly note on what drove the results.
Read the full breakdown: Marketing KPIs and a Dashboard Your CEO Will Trust →
A ten-minute check
Before you buy anything or rebuild anything, answer these:
- Who builds campaign links, and is there one list of allowed sources and mediums?
- Do your forms save UTMs into the CRM? Pick one contact from last week and look.
- Is there a first-touch field that nothing ever overwrites?
- Do HubSpot and Salesforce agree on lead source for 50 random records?
- Does your funnel leave your domain anywhere, like a scheduler or app?
- Which five numbers does leadership look at, and does everyone define them the same way?
The first “no” is where you start. In my audits it’s one of the first two more often than not.
Where to start
Find out which link in your chain is broken.
The free B2B Attribution Audit walks your stack across tracking, lead flow, paid media, CRM data and reporting. It takes about five minutes, and you finish knowing the three things to fix first, plus a 30/60/90 plan.
Run the free audit →When you're ready to build it
The B2B Attribution System
Everything in this guide, packaged so your team can build it: the UTM system, GA4 and CRM setup, the model decision guide, and the dashboard. A weekend to two weeks of work. $399.
See what's inside →Common questions
What is marketing attribution?
Marketing attribution assigns credit for pipeline and revenue to the campaigns and channels that influenced each deal. In B2B it has to work across long sales cycles, several stakeholders, and two systems, usually a marketing platform and a CRM.
What is multi-touch attribution?
Multi-touch attribution splits credit across every tracked touchpoint before a deal, instead of giving it all to the first or last one. Linear, time decay, U-shaped and W-shaped are all multi-touch models. They only work when UTMs and CRM data are consistent.
What's the difference between first-touch and last-touch attribution?
First touch gives all the credit to what brought someone in. Last touch gives it all to what happened right before they converted. First touch is better for judging acquisition channels, last touch for short cycles with few touches.
What is the best attribution model for B2B?
There isn't one best model. Short sales cycles with few touchpoints do fine with first or last touch. Cycles over 60 days with multiple stakeholders usually get the most useful answers from W-shaped, as long as the data underneath is clean. Compare the models.
Why don't HubSpot and Salesforce show the same numbers?
Usually it's field mapping. Lead source values that don't match, lifecycle stages that don't line up with lead status, sync errors nobody checks, and duplicate records. Each one is fixable once you find it. See the common causes.
Why is so much of my traffic showing as direct?
Some of it is people typing your URL or using a bookmark. A share that jumps or keeps growing usually means sources are being dropped: a page or subdomain missing the GA4 tag, separate domains without cross-domain measurement, or a consent setup that blocks tags. Untagged links mostly lose campaign detail rather than landing in direct. Find your tracking gaps.
Do I need marketing attribution software?
Not to start. HubSpot, Salesforce and GA4 can produce trustworthy attribution once UTMs, hidden fields and the sync are right. Software helps at scale, but it can't fix inconsistent data underneath it.
How do I keep first-touch source from being overwritten in HubSpot?
Form submissions overwrite UTM fields every time. Keep a second set of original UTM properties that only fill when they're empty, and never put them on a form, since a form submission overwrites whatever's there. On HubSpot Professional a workflow can do it. On Starter, an automation tool like Make can. See the full UTM setup.
How long does it take to fix B2B marketing attribution?
The tagging, capture and tracking fixes take days. The longer part is waiting for clean data to build up, since your reports only become trustworthy once new contacts flow through the fixed system. Start with the free audit to see where yours breaks.