Your attribution report says paid social drove 60% of revenue. Sales says it drove almost none. Both of you are reading real data, and both of you are right, because you picked different models and nobody agreed on the rules first.
That's the moment attribution dies in most companies. Not when the tool breaks. When the trust breaks.
Here's the good news: the fix costs nothing and starts before you open a single vendor website.
I sat down with Tish Millsap on the Marketing Smarts podcast to talk through the revenue attribution puzzle, and she said something in the first five minutes that reframed the whole conversation for me. She's been called into organizations to clean up attribution projects that already had a tool, already had budget, and already had a report nobody trusted.
The Short Answer: Attribution Is a Question Problem, Not a Tool Problem
Revenue attribution means connecting the activities you run to the revenue those activities helped create. That's it. Tish put it plainly.
"We have activities that are happening and we want to connect those to the outcome of revenue being created. And that's really the essence of revenue attribution." Tish Millsap
Most teams break this by starting at the end. They buy a platform, configure it fast, ship a report, and then discover that nobody in the building agrees with what it says.
"A lot of people that I work with, when we start talking about revenue attribution, they jump right to the tool. Like, I need to buy a tool to do this. And really, the tool is kind of further down in the decision chain." Tish Millsap
The industry data backs her up. The IAB's State of Data 2026 report, built on a survey of more than 400 senior planning and analytics decision-makers at U.S. brands and agencies, found that between 60% and 75% of buy-side users say their advanced measurement approaches fall short on rigor, timeliness, trust, and efficiency. Only 39% use attribution, incrementality testing, and marketing mix modeling together, even though those approaches complement each other.
Read that again. This is an industry that has bought the tools. It still doesn't trust the numbers.
So let's build this the other way around.
The Who, What, When, Where, and Why Framework
Tish uses a five-question framework, and the order matters more than the questions do. You answer these before you shop.
Start With Why
Why are you doing this? What specific question are you trying to answer?
Tish described walking into a new engagement where the client had already purchased a tool. When she asked what questions they wanted answered, they didn't have any. They just wanted "revenue attribution."
So she pushed: do you want to understand paid channels versus free channels? Do you want to know what creates opportunities, or what closes them? Those are different questions, and they need different setups.
Write your questions down. Three to five of them, in plain language, before anything else happens. Every decision after this one gets easier when you have them.
Then Who
When an opportunity gets created, which people count?
You can connect just the primary contact. You can connect the full buying group. You can connect every contact at the account. Whatever you choose determines which activities get credit, and the broader you go, the broader your model becomes.
This decision carries more weight in 2026 than it did even two years ago. Forrester's The State of Business Buying, 2026, drawn from a survey of nearly 18,000 global business buyers, found the typical buying decision now includes 13 internal stakeholders and nine external influencers, and that number climbs for complex or expensive purchases.
Think about what that means. If your reps associate one contact to a deal, your model sees one journey out of twenty-two humans. The report isn't wrong, exactly. It's just answering a much smaller question than you think it's answering.
Here's the HubSpot-specific version of this trap. HubSpot's revenue attribution reports only include deals that have at least one contact associated to them. No associated contact means no attributed revenue, and that deal quietly disappears from your report without any error message.
Then What
What activities belong in the model?
Start with the core: paid media, content downloads, events, webinars. Then decide whether you want the wider set, things like page views, BDR calls, and meetings booked.
"The basic bread and butter is the basic activities that you're doing as a marketing team. Where is your focus? What are you executing on?" Tish Millsap
More activities isn't automatically better. Every activity you add is a data hygiene commitment you're making to your future self.
Then When
How far back should the model look?
If an opportunity gets created today, do you want to see three years of history? Or do you want a tighter window, where only the last six months counted?
Run the math on your own sales cycle before you pick. 6sense's 2025 B2B Buyer Experience Report, based on responses from roughly 4,000 buyers across North America, EMEA, and APAC, found the average B2B buying cycle ran 10.1 months, down from 11.3 months the prior year. The same research found buyers now make first contact around 61% of the way through their journey, up from 69%.
So a six-month lookback window on a ten-month sales cycle doesn't just trim the edges. It erases the entire discovery phase, which is exactly where your content and brand work live. You'll conclude that content doesn't drive revenue, and you'll be wrong.
And Finally Where
Where do you want to see this? Which tool?
Last. Not first. By the time you get here, you already know what you're buying and why, which makes you a much harder person to sell to.
The Myth: You Only Need One Model
When I asked Tish what she'd most like to debunk, she didn't hesitate.
"I think it's the idea that you only need one model. Different models tell you different things." Tish Millsap
Different questions need different models. If you want to know what pushed something over the line into becoming an opportunity, a last touch model answers that. If you want a general read on how everything works together, an even spread multi-touch model answers that instead.
HubSpot gives you nine models to work with. As of its June 2026 documentation, those are First Interaction, Last Interaction, Linear, U-shaped, W-shaped, Time decay, Full path, J-shaped, and Inverse J-shaped.
A few of the practical details that trip up even experienced admins:
- U-shaped gives 40% to the first interaction and 40% to lead conversion, splitting the remaining 20% across everything else.
- W-shaped gives 30% each to first interaction, contact creation, and the interaction that created the deal, with 10% spread across the middle. It requires a deal-based interaction.
- Full path gives 22.5% each to first interaction, lead creation, deal creation, and last interaction, with 10% across the middle. It's available only in revenue attribution reports.
- Time decay uses a seven-day half-life, so an interaction eight days out gets half the credit of one from the day before conversion.
And here's the gotcha worth writing on a sticky note. With W-shaped and Full path models, if a deal's create date falls before the create dates of its associated contacts, the model returns null. Not zero. Null. Your report will look broken, and the cause is a data sequencing problem, not a tool problem.
Run two or three models side by side. The gap between what first touch says and what last touch says isn't noise. That gap is information about where your funnel actually does its work.
The Part Nobody Talked About Three Years Ago: What Attribution Can't See
This is where I need to go past what we covered on the podcast, because the ground moved.
Your attribution model can only credit what it can observe. In 2026, a growing share of the buying journey happens somewhere your tracking code will never reach.
Forrester's 2026 research found that 94% of business buyers now report using AI during their buying process. When a buyer asks an AI assistant to compare three vendors, reads the answer, and then types your company name into their browser, your analytics records that as direct traffic. The conversation that built the shortlist leaves no trace.
6sense has reported that only about 3% of website visitors self-identify through form fills, which means the overwhelming majority of research on your site is anonymous by default.
Three practical moves close part of that gap.
Turn on AI Referrals in HubSpot. HubSpot now categorizes traffic from AI assistants and chatbots including ChatGPT, Claude, Perplexity, and Gemini under a dedicated AI Referrals source, and it's available as both Contact Original Source and Latest Source. Note the boundary: regular Google and Bing searches still land under Organic Search, and only AI-specific domains get the AI Referrals label.
Add a self-reported attribution field. One required "How did you hear about us?" field on your high-intent forms captures what no pixel can. It's unreliable for any single record and quite useful in aggregate, which happens to be the level you make budget decisions at anyway.
Watch the AEO signal. HubSpot launched its AEO tool at Spring 2026 Spotlight, and in that announcement HubSpot reported that organic traffic for its customers was down 27% year over year. That's a number worth sitting with. If a quarter of your organic traffic evaporated, your attribution report would show declining content performance, and the underlying content might be performing better than ever inside answers you can't measure.
Reflective question for your next team meeting: if your attribution report can't see the channel that built your shortlist, what decisions are you making with it that you shouldn't be?
The Two Things That Actually Break Attribution Projects
Tish named a yin and a yang here, and I've watched both play out in HubSpot portals for years.
One: Your Data Isn't Clean
Vendors will tell you they can have you running in 30 days. Tish is careful here: that claim is usually true, but only if your underlying data is already clean.
"Are you using the same kind of campaign methodology? Are you using the same kind of form submissions? Is all this stuff cleaned up and organized in your system? Because garbage in, garbage out." Tish Millsap
Garbage in, garbage out. It's the oldest rule in reporting and it's still undefeated.
For HubSpot specifically, revenue attribution reports only include deals that meet all three of these conditions: the deal sits in a closed-won stage, it has at least one associated contact, and it has known values in the Amount, Create date, and Close date properties. Miss any one of those and the deal doesn't count.
HubSpot also applies sampling, processing up to 100,000 associations or activities per deal so that a handful of high-activity records don't distort everything else. Good to know before you go hunting for a page view that isn't showing up in a drilldown.
Data likes to be uniform. Your job is to make it boring and consistent before you make it fancy.
Two: Nobody Agreed to the Rules
This is the one that quietly kills more projects than dirty data does.
"The sales organization has to be on board or they're never going to believe what it is that the model puts out." Tish Millsap
Tish shared something that stuck with me. On a recent engagement, she was brought in by the CFO rather than the CMO or VP of Marketing, which was a first for her. That CFO turned into the strongest internal advocate on the project, because he already cared about how money got spent and what it returned.
If you're driving this from marketing, get sales ops to agree to the methodology first. The model. The inputs. The window. All of it, before configuration starts.
"You'll implement this fancy tool, you'll do all the hard work, and then nobody believes the reports once they come out." Tish Millsap
Nothing frustrates a sales leader faster than being handed a scorecard they never got to review. Bring them into the conversation while it's still a conversation.
You Probably Already Own Enough Tool to Start
Tish gave two tips here that have aged well, though one needs a 2026 correction.
First tip: watch where you're getting your education. Every vendor teaches attribution through the lens of their own product philosophy. Read widely, and look for guidance that isn't tied to a purchase.
Second tip: check your existing stack before you shop.
"You probably already have something in your tech stack that does attribution." Tish Millsap
That's still true, with an important asterisk for HubSpot humans. Attribution reporting in HubSpot isn't available across all tiers. Contact create attribution requires Marketing Hub Professional or Enterprise (or Content Hub Professional or Enterprise). Deal create attribution and revenue attribution require Marketing Hub Enterprise.
So if you're on Starter or Free and someone tells you revenue attribution is already included, that's not accurate. Check your tier before you build a plan around it.
Now the correction. On the podcast, the entry cost for a dedicated attribution tool got framed as roughly $50,000 to $100,000. That was a fair read at the time. It's no longer the whole picture. Several B2B attribution platforms now publish free tiers and monthly plans well under four figures, and Google's Meridian, an open-source marketing mix model, became generally available to everyone in February 2025 at no license cost, with a no-code Scenario Planner interface added in February 2026.
The honest version for 2026: enterprise attribution platforms still run into serious money, and the floor has dropped a lot. Verify current pricing directly with any vendor before you budget, because this category is moving fast.
The point Tish was making holds either way. Cost was never the real barrier. Clarity was.
What Success Actually Looks Like
I asked Tish how you know you've done this well. Her answer had two halves, and the second half is the one I keep thinking about.
The formal half: you hold real meetings where you look at attribution reports and make decisions from them. SEO is performing, so you invest more there. That trade show produced no closed-won revenue last year, so you skip it this year.
The informal half is the real tell.
"You're in a conversation and people randomly bring up the tool and say, hey, let's look and see how that white paper performed." Tish Millsap
When somebody who doesn't own the dashboard pulls it up mid-conversation out of curiosity, attribution has stopped being a marketing project and started being how your company thinks. That's the finish line. And like every finish line, it's really just the start of the next race.
Your First 30 Days
We didn't cover this on the podcast, so let me add it. Here's how I'd sequence the work if you're starting Monday.
Week one: write the questions. Three to five specific questions in plain language. Not "we want revenue attribution." More like "which channels create opportunities that actually close?"
Week two: hold the buy-in conversation. Get sales ops and your finance partner in a room. Walk them through your questions and your proposed model. Ask what would make them distrust the output, then design around those answers.
Week three: audit the data. Check deal Amount, Create date, and Close date completeness. Check contact association on closed-won deals. Check campaign naming consistency and UTM discipline. Fix what's broken before you report on it.
Week four: run two models side by side. Use what you already own. Compare first touch against last touch and talk about the gap as a team.
Notice that buying anything doesn't appear on this list. That's on purpose. Do these four weeks first and you'll either discover you don't need a new tool, or you'll walk into vendor conversations knowing exactly what you're buying.
Frequently Asked Questions
What's the difference between revenue attribution and regular reporting?
Revenue attribution is a type of report, not a separate discipline. Tish frames marketing reporting in five buckets: backroom metrics that only marketing cares about, campaign performance, operational metrics, volume and velocity and conversion rates, and attribution. Attribution is the bucket that connects the other four to money.
Which attribution model should I use?
The one that answers your question. Last touch tells you what closed. First touch tells you what created awareness. Multi-touch models tell you how things work together. Most teams need at least two running side by side.
Do I need to buy an attribution tool?
Probably not to start. If you're on Marketing Hub Professional or Enterprise, you already have attribution reporting available, and you'll learn more in your first 90 days of using it than any demo will teach you. Buy a dedicated platform when you've outgrown what you own, not before.
Why does my HubSpot revenue attribution report look empty?
The usual causes are structural. Deals need to be in a closed-won stage, have at least one associated contact, and have values in Amount, Create date, and Close date. If you're using W-shaped or Full path models, also check whether your deal create dates fall before your contact create dates, which produces null values.
How do I track leads that came from ChatGPT or other AI tools?
HubSpot now has an AI Referrals traffic source that captures visits from AI assistants including ChatGPT, Claude, Perplexity, and Gemini, available as both Contact Original Source and Latest Source. Pair it with a self-reported "How did you hear about us?" field, since many AI-influenced buyers arrive as direct traffic with no referrer at all.
Here's the Bottom Line
Revenue attribution isn't a tool you install. It's an agreement you build.
Agreement on the questions. Agreement on who counts, what counts, and how far back you look. Agreement with the sales team on the rules before anybody sees a number. The software just does arithmetic on top of the agreement your humans already made.
You don't need permission or budget to start. You need a whiteboard, three good questions, and one honest conversation with sales.
Let's Build the Report Your Team Will Actually Use
If you're staring at a HubSpot portal wondering whether your attribution reporting is telling you the truth, we can look at it together. Our team runs a free portal audit that examines your attribution setup, your deal data completeness, your campaign structure, and the tracking gaps quietly distorting your numbers.
No pitch deck. Just a clear read on what's working, what's broken, and what to fix first.
Book your free portal audit and get a straight answer about your attribution data.
Not ready for a conversation yet? That's completely fine. Grab our HubSpot reporting checklist and run the audit yourself. We give away the playbook, not just the results.
