Last month, I was reviewing attribution data for a B2B SaaS client. Their dashboard told a clear story: Google Ads drove 67% of pipeline. LinkedIn was underperforming. Podcasts were invisible — barely a rounding error. The recommendation from their previous agency was obvious: double down on Google, cut podcast sponsorships, reduce LinkedIn spend.
Then they did something unusual. They asked their customers directly.
Over 40% of new customers mentioned a specific podcast episode as the reason they first looked into the company. A quarter cited peer recommendations in private Slack communities. Google Ads? When pressed, most buyers admitted they simply Googled the company name after hearing about it elsewhere. Google wasn't driving pipeline. Google was capturing demand that podcasts and word-of-mouth had created. The dashboard had the story exactly backwards.
The Last-Click Lie
Attribution tools are built to measure the final touchpoint before conversion. Someone hears about you on a podcast, thinks about it for three weeks, mentions you to a colleague, reads two blog posts on their phone during a commute, and then — three weeks later — Googles your name and fills in a contact form. The dashboard credits Google with 100% of that conversion.
This is last-click attribution, and it is the default in almost every analytics platform. It doesn't just undercount certain channels — it systematically overcredits the final, most visible touchpoint and makes invisible the channels that actually created the demand.
Multi-touch attribution models try to solve this problem by distributing credit across multiple touchpoints. But they can only credit what they can see. A podcast mention, a Slack conversation, a recommendation from a friend at a conference — none of these generate a trackable click. They don't appear in the model. They don't get credit. They might as well not exist.
Multi-touch attribution doesn't fix the fundamental problem. It just spreads the same incomplete data across more touchpoints. You move from confidently wrong to precisely wrong.
Why B2B Attribution Is Fundamentally Broken
The attribution problem is not limited to last-click models. It is structural. B2B buying behaviour is fundamentally incompatible with the way attribution tools work.
Consider what a typical B2B buying journey actually looks like. A buyer researches for months — often three to nine months for enterprise software. They get recommendations in private channels: Slack communities, WhatsApp groups, LinkedIn DMs, conversations at events. They consume content across untrackable devices — listening to a podcast in the car, reading a LinkedIn post on a personal phone, watching a webinar on a shared office screen. Buying committees involve three to ten stakeholders, most of whom never interact with your marketing directly. And by the time someone fills in a form, they have already decided you are worth evaluating.
By the time a buyer converts, roughly 70% of their journey happened in places your attribution dashboard cannot see. You are making budget decisions based on the 30% that is visible — and that 30% is disproportionately weighted towards demand capture rather than demand creation.
Based on a composite of self-reported attribution data from B2B SaaS companies. Dashboard figures represent last-click attribution; reality figures represent customer-reported first meaningful touchpoint.
The Three Questions Your Attribution Tool Can't Answer
If attribution dashboards can't tell you what's actually driving pipeline, what should you be asking instead? There are three questions that matter more than any metric your dashboard can produce — and no attribution tool on the market can answer any of them.
What first put us on this buyer's radar?
Attribution tools can't answer this because the first touch often happens in places they can't see — a podcast mention, a Slack recommendation, a conversation at a conference. By the time someone interacts with a trackable touchpoint, they already know who you are.
What convinced them we were worth evaluating?
The shift from awareness to active consideration usually involves consuming long-form content, reading case studies, or hearing a trusted peer vouch for you. These moments rarely generate a click or a form fill. They generate intent — which is invisible to dashboards.
What content or interaction built enough trust to reach out?
Trust is the final gate before a buyer takes action. It's built over weeks or months through repeated exposure, perceived expertise, and social proof. No attribution model can assign a percentage to trust.
These are not abstract questions. They are the questions that determine where you should invest your marketing budget. And the only way to answer them is to ask the people who actually bought from you.
Self-Reported Attribution: The Missing Piece
The solution is deceptively simple: ask your customers how they first heard about you. Not with a dropdown menu — with an open text field.
Dropdowns constrain the answer to options you've already thought of. They create false precision. A dropdown that says "Google" doesn't tell you whether someone searched your brand name after hearing you on a podcast or found you through a cold keyword search. These are completely different journeys with completely different implications for your marketing strategy.
Open text captures the actual story. "My colleague Sarah mentioned you in our team Slack." "I heard James on the Revenue Architects podcast." "Someone in the Pavilion community recommended you." These are the answers that change strategy. And they only appear when you stop forcing buyers into your pre-defined categories.
The objection I hear most often is that open text is hard to analyse at scale. This was true five years ago. It is no longer true. AI can categorise hundreds or thousands of free-text attribution responses in minutes — identifying patterns, clustering similar responses, and surfacing channels that your dashboard never knew existed.
The dropdown tells you which bucket to put the conversion in. The open text tells you the actual story — and the story is what changes strategy.
Your Dashboard Isn't Measuring What Matters.
Attribution dashboards are designed to measure demand capture — the moment someone clicks, fills a form, or converts. They are very good at this. The problem is that demand capture is the last 10% of the buyer journey.
The other 90% — the podcast that made them aware, the peer who recommended you, the content that built trust over months — is where your marketing budget should be optimised. But it's the part your dashboard can't see.
When you make budget decisions based only on what the dashboard can measure, you systematically defund the channels that create demand and overfund the channels that capture it. You starve the top of the funnel to feed the bottom — and then wonder why pipeline is shrinking.
What To Do Now
The answer is not to throw out your attribution dashboard. It still provides useful signal about demand capture — which ads generate clicks, which landing pages convert, which email sequences drive replies. That data has value.
The answer is to stop using it as the primary input for strategic budget decisions. Your dashboard can tell you how to optimise within a channel. It cannot tell you which channels to invest in. For that, you need to hear from the people who actually bought.
Add a "How did you first hear about us?" field to every conversion point
Contact forms, demo requests, sign-up flows. Use open text, not a dropdown. Make it optional if you must, but you will find that most buyers are happy to tell you — because they remember the moment. It was a real experience, not a click.
Resist the urge to create categories too early
Let the data tell you what the categories are. You will discover channels and touchpoints you didn't know existed. AI can help you cluster responses once you have enough volume — but start by reading them yourself.
Compare self-reported attribution to your dashboard data monthly
The gaps will tell you where you are overinvesting and underinvesting. If your dashboard says podcasts drive 2% of pipeline but your customers say podcasts are how they found you, that is the most important signal your marketing team will see all quarter.
Stop using dashboard attribution for budget allocation
Use your dashboard for tactical optimisation within channels. Use self-reported attribution for strategic decisions about which channels to fund. These are different questions and they require different data sources.
Use AI to analyse at scale
Once you have hundreds of open-text responses, use AI to categorise, cluster, and identify patterns. This is where machine learning genuinely helps — not in tracking the untrackable, but in making sense of what your customers are telling you.
The businesses that figure this out first gain a significant advantage. While their competitors are optimising based on incomplete data — pouring money into demand capture while starving demand creation — they are investing in the channels that actually build pipeline. Not because they have better tools. Because they asked a better question.
Frequently Asked Questions
What is marketing attribution?
Marketing attribution is the process of identifying which touchpoints contribute to a conversion. Most tools only capture demand capture (the moment someone fills a form or clicks an ad) rather than demand creation (the weeks or months of exposure that made them want to buy in the first place).
Why do traditional attribution models fail in B2B?
B2B buying cycles are long (often 3-9 months), involve multiple stakeholders, and rely heavily on untrackable channels like peer recommendations, private communities, and in-person conversations. Traditional models can only credit what they can see, which is a small fraction of the actual buyer journey.
What attribution model works best for B2B?
The most effective approach combines traditional analytics (for demand capture signals like ad clicks and landing page conversions) with self-reported attribution (for demand creation signals like podcasts, word-of-mouth, and community recommendations). Neither alone tells the full story.
Can machine learning improve marketing attribution?
Machine learning can't track what's fundamentally untrackable — a conversation in a Slack community, a recommendation over coffee, a podcast listened to on a morning commute. Where AI genuinely helps is in analysing self-reported attribution responses at scale, categorising open-text answers into meaningful patterns.
What questions can and can't attribution tools answer?
Attribution tools CAN tell you which ads get clicks, which landing pages convert, which emails get opened, and which campaigns drive form fills. They CAN'T tell you what first put you on a buyer's radar, what built enough trust for them to reach out, or which channels created demand that later converted through a different channel.
Your attribution dashboard is not lying out of malice. It is lying because it was built to measure a world that no longer exists — a world where buyers clicked an ad, visited a website, and filled in a form. The modern B2B buyer journey is messier, longer, and more human than any dashboard can capture. The sooner you accept that, the sooner you can start measuring what actually matters.
James Kevan is the co-founder of First Signals, where he helps B2B companies understand what's actually driving their pipeline. If you want to find out what your attribution dashboard is missing, the AI Opportunity Audit starts by asking the questions your tools can't answer.
From the same series: Islands. Good Tools. No Bridges. · Your Business Isn't Broken. Your Processes Are. · The AI Brain Freeze · The Quiet Businesses.
© 2026 James Kevan / firstsignals.ai. Share freely with attribution.