Most marketing attribution content starts with tools: GA4, HubSpot, Rockerbox, Northbeam. The tool conversation is premature. Before you configure anything, you need to be clear on what question you're trying to answer — because different questions require completely different attribution logic.
The founders who get attribution wrong don't have a tools problem. They have a clarity problem. They're measuring activity when they need to understand influence.
The Question Attribution Actually Has to Answer
There are two fundamentally different questions a B2B company might want to answer with attribution:
Question one: what introduced the buyer to us? This is the first-touch question. It's useful for understanding how people find you and which awareness channels are worth investing in. But on its own, it's incomplete — especially in B2B, where the time between first awareness and purchase can be six to eighteen months and involves multiple touchpoints.
Question two: what caused the buyer to act? This is the last-touch question. It tells you what pushed someone from considering to deciding. It's useful for conversion optimization, but it overstates the value of bottom-of-funnel channels and completely ignores everything that built trust before the decision.
The trap most companies fall into: they pick one question and pretend it answers both. They run last-touch attribution because it's what HubSpot defaults to, then they defund their content program because content "doesn't show up in attribution." Three quarters later, their pipeline is thin because the top-of-funnel work that was seeding consideration has dried up. They blame the market.
The right question for a $5M ARR B2B company with limited resources: which activities are driving deals I would not have won otherwise? That's a different question from either of the above, and it requires a model that captures both the starting point and the deciding factor.
The Model That Actually Works at This Stage
The model I recommend to companies in the $3M–$15M ARR range: last meaningful touch with first-touch assists.
Here's how it works in practice. For every closed deal, you record two things: the first marketing touchpoint (how they found you) and the last meaningful marketing touchpoint before they entered active sales conversations. "Meaningful" means they engaged with something — clicked a link, downloaded content, attended a webinar — not just a page visit.
You attribute primary credit to the last meaningful touch. You record first touch as an assist. Over time, you run two analyses: which channels are producing first touches on deals that eventually close (your awareness health metric), and which channels are the last meaningful touch before the sales conversation starts (your conversion influence metric). These two analyses together give you a functional picture of your pipeline.
This model is not perfect. No attribution model is perfect. But it's the right level of sophistication for a company that doesn't have a data team, needs to make channel budget decisions quarterly, and is working with six to eighteen month sales cycles where multi-touch data is incomplete.
Why Single-Touch Models Fail in B2B
Single-touch attribution — first or last — consistently produces the same distortion in B2B: it makes you underinvest in the channels that build consideration and overinvest in the channels that capture it.
Organic content is the canonical example. A buyer reads three of your blog posts over four months, downloads a guide, subscribes to your email list, and then books a demo after clicking a paid retargeting ad. Under last-touch attribution, the retargeting ad gets credit. Under first-touch attribution, the first blog post gets credit. Under neither model do you know that the content sequence is what made the retargeting ad work.
Your content marketing pipeline is almost certainly doing more work than your attribution model is giving it credit for. This is not an argument to defund paid — it's an argument to understand what's actually doing setup work before you optimize toward the channel that shows up at the end.
The practical test: pull your last ten closed deals. Interview the AE and, if possible, the customer. Ask them what they remember reading, watching, or engaging with before they first talked to sales. Then compare that to what your CRM shows. The gap between those two answers is how broken your current attribution is.
Setting This Up Without a Data Team
The common objection: "we don't have the infrastructure for this." You don't need custom infrastructure. You need three things working correctly.
UTM discipline. Every link your company puts into the world — paid ads, emails, social, partnerships — needs a consistent UTM structure. First touch can only be tracked if the channel is labeled when someone first arrives. This is not a technical problem; it's a process problem. One person needs to own the taxonomy and enforce it.
CRM fields that capture both touchpoints. In HubSpot, Salesforce, or whatever CRM you're using: you need a field for first touch source and a field for last meaningful touch source. These should be populated automatically where possible (UTM data) and manually completed by the AE at opportunity creation for the rest. This takes one afternoon to set up and two weeks of habit to enforce.
A quarterly attribution review. Once a quarter, someone pulls the closed-won deals, maps them to first touch and last meaningful touch, and produces a simple table: by channel, how many deals did we close, what was the average deal size, and what was the first touch vs. last meaningful touch breakdown. This is a spreadsheet exercise, not a BI dashboard project.
If your B2B SaaS marketing strategy is running on gut feel and recency bias rather than this kind of structured review, your channel allocation is probably wrong — but you can't know by how much.
The Number One Attribution Mistake at $5M ARR
The most common attribution failure at this stage is not a technical mistake. It's a political one.
Attribution data shows that certain channels are doing more work than their budget reflects. Often that means content is undervalued and paid is overvalued. The paid budget is visible and has a clear owner. The content program is diffuse and difficult to defend in a budget meeting. So when the numbers come back showing paid conversion looks strong (because it's capturing demand that content built), the natural conclusion is: put more into paid.
This works until it stops working — usually when the content program has been cut to the point where it's no longer seeding the consideration that makes paid retargeting effective. Then CPAs spike and nobody can explain why.
Good attribution doesn't just tell you what's working. It tells you why it's working — and that often means understanding the upstream dependencies between channels, not just the last thing that touched a deal before it closed.
If you want to map out the attribution gaps in your current setup, the Strategy Diagnostic is a good place to start. It typically surfaces two or three specific measurement problems that are distorting your channel decisions.