The board deck is due next week and the same question keeps coming back: which channel is bringing in deals, and what do you tell the board? Marketing attribution is how you answer it.
Put simply, it’s how you decide which marketing touches get credit for a lead, a deal or revenue. At post-seed that mostly means deciding where the next euro goes.
The catch? B2B buying looks nothing like the tidy funnels in most attribution models. Deals take months and involve several people, and a lot of the research happens where no tracking pixel can see it.
Below: the main attribution models and where each goes blind, the one I’d use at your stage, what HubSpot Starter can realistically do, and a one-slide template for the board.
The short version, and a bit of a relief: no model is “true”. The goal is a setup that’s honest enough to make better budget decisions than you make today.
The quick version
- Attribution decides which touches get credit for a lead, deal or revenue. No model is “true”; aim for better budget decisions.
- GA4 now offers only data-driven and two last click models; build rule-based multi touch models in the CRM.
- Long cycles, buying groups of five to 16 people and dark social all hide touches from software.
- On HubSpot Starter: original source plus a required “how did you hear about us” field. Multi-touch reports need Marketing Hub Professional.
- For the board: one slide comparing self-reported and software-attributed deals, pipeline in euros, and next month’s move.
What is marketing attribution?
In practice, marketing attribution means giving credit for a conversion to the touchpoints before it: an ad click, an organic search visit, an email click, a webinar sign-up.
An attribution model is the rule that decides how that credit gets shared out. Single-touch models give all the credit to one touch. Multi touch attribution spreads it across several touches, either by a fixed rule or by an algorithm.
Attribution is also only one input into your reporting. It sits alongside the pipeline and efficiency numbers I covered in the B2B SaaS marketing metrics worth tracking.
Marketing attribution models compared
There are seven attribution models most B2B teams will bump into. They differ in where they put the credit, which quietly changes which channels look like heroes.
| Model | How credit is assigned | Good for | Blind spot | When a post-seed team should use it |
|---|---|---|---|---|
| First touch | 100% to the first touch | Seeing what creates awareness | Ignores everything that closed the deal | To check which channels bring new accounts in |
| Last touch | 100% to the last touch before conversion | Seeing what triggers the demo request | Over-credits branded search and retargeting | Paid campaign optimisation, with caution |
| Linear | Equal credit to every touch | A neutral view of the whole path | Treats a newsletter open like a demo call | Rarely; useful as a comparison view |
| Time decay | More credit to touches closer to conversion | Short, sales-assisted cycles | Under-credits early awareness work | When most deals close within weeks |
| U-shaped (position based) | 40% first touch, 40% lead creation, 20% spread across the rest | Lead generation reporting | Stops at lead creation, ignores the deal | Default choice once you have repeatable lead flow |
| W-shaped | 30% first touch, 30% lead creation, 30% deal creation, 10% the rest | Connecting marketing to pipeline | Needs clean CRM stages and deal data | Once opportunities are tracked consistently |
| Data driven | An algorithm assigns credit based on converting and non-converting paths | Optimising paid channels at volume | A black box; weak with low conversion volume | Later, in GA4 or ad platforms, for paid spend |
The percentages for U-shaped and W-shaped come from HubSpot’s attribution documentation. HubSpot offers nine marketing attribution models, including full path, J-shaped and inverse J-shaped variants, if you fancy going deeper.
Which attribution models GA4 still supports
This one catches people out: GA4 now offers far fewer models. According to Google’s GA4 attribution help page, the first click, linear, time decay and position-based models “are no longer available as of November 2023”.
What remains is data-driven attribution, paid and organic last click, and Google paid channels last click. All of them exclude direct visits from credit unless the whole path is direct. So if you want rule-based multi touch attribution models, the CRM is now where they live.
Multi touch attribution: what it can (and can’t) tell you
Multi touch attribution (often shortened to MTA in marketing) shows how channels work together along a buying path. What it can’t do is show touches it never recorded, and in B2B that’s a big share of them.
Where MTA marketing reports shine:
- Showing that a channel which rarely closes deals (for example, LinkedIn ads) often opens them.
- Comparing the same channel across quarters, using the same model.
- Spotting content and pages that appear in many winning paths.
Where they struggle:
- Anything offline.
- Anything on a different device.
- Anything seen but not clicked.
- Anyone in the buying group who never filled in a form (which is most of them).
So treat MTA as a directional view, not an invoice.
Why B2B attribution breaks: long cycles, committees and dark social
Picture a deal that starts with a podcast, runs for months and gets signed off by a CFO who never touched your website. Most setups miss nearly all of that. The cycle outlasts the lookback window, the buyer is a group rather than a person, and much of the influence leaves no click.
Long sales cycles
GA4’s default lookback window for key events is 90 days, per its attribution settings documentation. If your deals take longer than that from first visit to demo, the first touches simply fall off the edge of the report.
Quick test
Compare your typical time from first visit to demo request with GA4’s 90-day window. If the journey is longer, the report is quietly missing the touches that got the whole thing started.
Buying committees
Gartner reports buying groups ranging from five to 16 people across as many as four functions. Usually just one of them fills in the demo form, so contact-level attribution credits that person’s path and misses everyone else’s. The CFO who read three of your posts? Invisible.
Dark social
Plenty of deals start somewhere no tool can see, like:
- A founder’s LinkedIn post.
- A Slack community thread.
- A podcast mention.
- A recommendation in a private chat.
The buyer then types your URL or searches your brand, and the credit goes to “direct” or “branded search”, which is a bit like thanking the front door for the dinner party. The same is increasingly true of answers from AI assistants, which I wrote about in AI search visibility for B2B SaaS.
Self-reported attribution: the “how did you hear about us” field
Self-reported attribution means asking the buyer directly where they heard about you, usually with a required open-text field on the demo form. It’s the cheapest way to see the dark social touches software misses.
Here’s how I’d set it up:
- Open text, not a dropdown. Dropdowns nudge people into picking the first plausible option. Free text gives you “saw your CEO’s post about X”, which is far more useful.
- Only on high-intent forms: demo request, contact sales, trial sign-up. Not on every ebook download.
- Categorise monthly into a short list (LinkedIn organic, referral, podcast, search, events, other) in a separate CRM property.
- Ask again on the first sales call and log the answer. Buyers often remember better once they’re chatting.
Then read the two side by side. When software says “direct” and buyers say “LinkedIn”, the gap is your dark social signal. I don’t fully trust either on its own. Side by side, they’re usually close enough to act on.
When software says “direct” and buyers say “LinkedIn”, the gap is your dark social signal.
One refinement: “LinkedIn” alone isn’t enough. It could mean the founder’s posts or LinkedIn Ads, which are completely different budgets. Before moving money, I split the answer with a follow-up question or separate options like “LinkedIn post” and “LinkedIn ad”.
Which attribution model to use at each stage
With a handful of deals a quarter, a fancy model mostly adds noise. The more deals you have, the more detail a model can carry.
Early: founder-led sales, few deals
Use the self-reported field plus first touch and last touch in the CRM. Read every deal by hand. At this volume, a conversation with sales tells you more than any model.
Post-seed: first marketer, repeatable lead flow
Move to U-shaped attribution in the CRM for lead creation, keep the self-reported field, and use GA4 for website behaviour. This is where most of the teams I work with sit, and it happily covers most budget decisions for demand generation programmes.
Towards Series A: steady pipeline and paid spend
Add W-shaped reporting so deal creation gets its own credit, and use data driven attribution in GA4 and the ad platforms to optimise paid spend. And yes, keep the self-reported field. Then revisit how budget moves between channels, as in how to set a B2B SaaS marketing budget.
How to set up marketing attribution in HubSpot and GA4
Most of a marketing attribution setup is plumbing: clean GA4 tracking and clean CRM stages, plus the self-reported field. The order I’d do it in:
- Agree a UTM convention (source, medium, campaign naming) and write it down. Honestly, most attribution problems start with messy UTMs.
- Mark your key events in GA4: demo request, trial sign-up, contact sales. Leave page views and scrolls out of it.
- Check GA4 attribution settings in Admin. Confirm the reporting model (data-driven, paid and organic last click or Google paid channels last click) and the lookback windows: 30 days by default for acquisition events, 90 days for other key events.
- Install the HubSpot tracking code on every page and connect your ad accounts, so ad interactions land on contact records.
- Clean up lifecycle stages and deal stages so lead creation and deal creation are recorded consistently. W-shaped attribution lives or dies on this.
- Add the “how did you hear about us” field to high-intent forms and create a categorised property next to it.
- Build the attribution report your plan allows (see the HubSpot Starter route below).
- Review monthly: software attribution next to self-reported answers, with sales in the room.
Founder shortcut
Only have an afternoon? Start with steps 1 and 6. A written UTM convention and the self-reported field fix more reports than any shiny new tool.
If this is the bit that keeps slipping down the to-do list, it’s the core of our growth and reporting work.
The HubSpot Starter route: what you can do before Professional
Starter already answers most early attribution questions: where contacts and deals first came from, and what buyers say brought them in. The multi-touch attribution reports are the part it doesn’t include.
Here’s the setup I’d use on Starter:
- Original source. HubSpot’s traffic source properties are available on all plans, and companies and deals associated with a contact mirror them. That’s your software-attributed first touch.
- A required self-reported field on the demo form. Create a “How did you hear about us?” property from the form builder, make it required, and add a second property for the monthly category.
- Deal source. When a deal is created, check the mirrored original source and ask sales to log the buyer’s first-call answer on the deal.
- A monthly export. Pull closed and open deals with both sources into a sheet and compare them side by side.
| What you want | On HubSpot Starter | What needs a higher plan |
|---|---|---|
| Where each contact first came from | Yes: Original Source properties, available on all plans | Nothing extra |
| The same source on deals | Yes: deals associated with a contact mirror its source properties | Nothing extra |
| A required self-reported field on the demo form | Yes: forms work on all plans, and a field can be made required | Nothing extra |
| Multi-touch contact create attribution reports | No | Marketing Hub Professional or Enterprise |
| Deal create and revenue attribution reports | No | Marketing Hub Enterprise only |
Upgrade once you have enough deals for a multi-touch model to tell you something a conversation with sales can’t.
What to tell the board: a one-slide attribution template
Tell the board what buyers say, what the software says, how much pipeline each channel touched, and what you’ll change next month. One slide, the same columns every time.
| Channel | Self-reported deals (example) | Software-attributed deals (example) | Pipeline € (example) | What we’ll do next month |
|---|---|---|---|---|
| LinkedIn, founder posts | 4 | 1 | 60,000 | Keep posting; add a clear next step on the profile |
| LinkedIn Ads | 1 | 3 | 25,000 | Hold spend; test one new audience |
| Search, organic and paid | 2 | 4 | 30,000 | Move budget to high-intent terms |
| Referrals and partners | 3 | 0 | 45,000 | Ask two happy customers for intros |
| Other or unknown | 1 | 3 | 10,000 | Fix UTMs on the newsletter links |
How to read it:
- Where self-reported and software numbers disagree (founder posts vs LinkedIn Ads above), the gap is the story. Explain it in one line rather than picking a winner.
- Show pipeline in euros. Boards react to money moving far more than to form fills.
- Every row ends in a decision, so the conversation stays on next month instead of drifting into a debate about the model.
Board-ready
Keep the columns fixed for at least two quarters. A steady, imperfect slide beats a new clever one every month, because the board can see the trend.
Frequently asked questions
What are the main marketing attribution models?
The main models are first touch, last touch, linear, time decay, U-shaped (position based), W-shaped and data driven. The first two give all credit to one touch. Linear, time decay, U-shaped and W-shaped split credit by fixed rules. Data driven uses an algorithm that compares converting and non-converting paths to assign credit.
What is multi touch attribution?
Multi touch attribution spreads credit for a conversion across several touchpoints instead of giving it all to one. It shows how channels work together along a buying path. In B2B it is directional rather than exact, because it misses offline touches, other devices and people in the buying group who never filled in a form.
What attribution models does GA4 support?
GA4 supports three models: data-driven, paid and organic last click, and Google paid channels last click. Google removed the first click, linear, time decay and position-based models in November 2023. All GA4 models exclude direct visits from credit unless the entire path to the key event consists of direct visits.
What is data driven attribution?
Data driven attribution uses machine learning to compare paths that led to a conversion with paths that did not, then assigns fractional credit to each touchpoint based on its estimated contribution. It is the most sophisticated option in GA4, but it works best with plenty of conversions, which early-stage B2B companies often lack.
Which attribution model is best for B2B SaaS?
It depends on stage and data volume. With few deals, use first and last touch plus a self-reported “how did you hear about us” field. With repeatable lead flow, U-shaped works well. Once deals are tracked consistently in the CRM, W-shaped connects marketing to pipeline. Keep the self-reported field at every stage.
One thing to do this week
Add a required open-text “How did you hear about us?” field to your demo request form, today. It takes about ten minutes and no new tool. Within a month you’ll have a list of real answers to put next to your software attribution. Next month, drop them into the board slide above.
And once you know which channels are pulling their weight, the free B2B Growth Planner helps you turn that into a team, budget and roadmap for the next few quarters.