Multi-Touch Attribution in HubSpot: A Practical B2B SaaS Setup
Most attribution setups in HubSpot are configured wrong — usually because someone enabled all attribution models and never picked one. Here's the model B2B SaaS companies should default to and exactly how to wire it.
There is a version of this post that goes through every HubSpot attribution model in detail, with diagrams and pros/cons. We will not write that post. Most B2B SaaS companies should pick one model, wire it correctly, and move on with their lives. Picking is the hard part; almost everything written about attribution exists to delay picking.
This post is the opinionated version. The model we recommend is W-shaped. The setup that makes it work is below. The reasons we do not recommend the alternatives are at the bottom.
If you came here for “it depends,” try a different post.
The problem with attribution in HubSpot today
Most attribution setups in HubSpot fail in one of four predictable ways. Pattern-match before you read the rest of the post.
Pattern 1: All models on, no model selected. HubSpot lets you enable first-touch, last-touch, linear, U-shaped, W-shaped, time-decay, and full-path simultaneously. Many setups have all of them on. Nobody knows which dashboard reflects “the truth,” so different leaders quote different numbers. This is the most common failure.
Pattern 2: Last-touch only. Default HubSpot installs lean last-touch. For SaaS sales cycles longer than two months, last-touch is mathematically a lie — it credits the channel that closed, not the channel that opened. Pipeline forecasts get optimised against the wrong channel mix.
Pattern 3: First-touch only. The opposite mistake. Often imported from a marketing-led culture that wants credit for the top of the funnel. Equally wrong for SaaS — it ignores everything between first contact and close.
Pattern 4: The attribution dashboard nobody trusts. Often the result of incomplete UTM tracking, missing campaign records, or inconsistent source-of-truth properties. The dashboard exists, the numbers update, and nobody quotes it in board meetings because the numbers are visibly wrong.
If you recognise your setup in one of these four, you do not have an attribution problem. You have a configuration problem. Configuration is fixable.
Why first-touch and last-touch both lie for SaaS
A typical B2B SaaS deal touches 15–40 marketing-owned interactions before close. The buyer sees an ad, downloads a guide, attends a webinar, reads three blog posts, gets sequenced by an SDR, takes a sales call, sees follow-up nurture, attends a customer reference call, and finally closes after a procurement review. That is the median, not the long tail.
First-touch credits the ad that started the journey. It tells you nothing about what kept the buyer engaged or what closed them.
Last-touch credits the SDR sequence or the bottom-of-funnel demo nudge. It tells you nothing about what made the buyer aware in the first place.
Either one, used in isolation, will steer you toward overspending on one half of the funnel and underspending on the other.
The W-shaped model — what it is, when it fits
W-shaped attribution credits three touchpoints heavily and distributes the rest:
- 30% to the first touch (the channel that created awareness).
- 30% to the lead-creation touch (the moment the contact became known — usually a form fill).
- 30% to the opportunity-creation touch (the moment the deal was created in the pipeline).
- 10% distributed across every other touch between those three.
This is the default for B2B SaaS for one reason: those three moments are the ones the marketing team can actually influence. Awareness, lead capture, opportunity creation. Channel performance against those three buckets is decision-grade data.
W-shaped fits when:
- Sales cycles are 30+ days (any deal with multiple touches before close).
- The marketing team owns at least the first two touches (awareness and lead capture).
- The marketing-to-sales handoff (opportunity creation) is documented and instrumented.
W-shaped does not fit if:
- Most deals close within a single touch (rare in B2B SaaS — common in transactional ecommerce).
- You sell purely through outbound and marketing has no touch on the early funnel (use a different model, or admit you are sales-led and stop attributing).
- You have not yet wired campaigns properly. W-shaped is data-hungry. Bad input is worse than no model.
The exact HubSpot setup
This is the configuration. Run it as a checklist.
1. Confirm the lifecycle definitions. Every contact must have a documented lifecycle progression: Subscriber → Lead → MQL → SQL → Opportunity → Customer. The “MQL” and “SQL” definitions must be agreed by Marketing and Sales, written down, and enforced by workflow. Without this, “lead-creation touch” is meaningless.
2. Wire the campaigns. Every paid channel — LinkedIn, Google, retargeting, sponsorships — must have a HubSpot Campaign record. Every email, every landing page, every form must be associated with a Campaign. Without Campaign records, attribution can show “Paid Social: 30%” but cannot break it down further. The Campaign record is the unit of attribution.
3. UTM hygiene.
A documented UTM convention enforced on every paid link. We use utm_source for platform (linkedin, google), utm_medium for type (paid, organic, email, referral), utm_campaign for the HubSpot Campaign name (verbatim), utm_content for the asset variant. Inconsistent UTMs are the single biggest source of attribution noise.
4. Custom property: First-Touch Source / Last-Touch Source / Lead-Creation Source. HubSpot has built-in original-source properties; we usually augment with three explicit custom properties so reporting is unambiguous. Populate them by workflow at the right lifecycle moments — not via formula, which fails on edge cases.
5. Deal property: Opportunity-Creation Source. When a deal is created, capture the Marketing Campaign or Sales Activity that created it. This is the third W-shape touch. It should be required at deal creation by stage gating.
6. Configure the W-shaped attribution report in HubSpot. Under Reports → Attribution, set the model to W-shaped and the conversion event to Closed-Won deal. The first version of the report will look chaotic; that is normal until the historical data has filled in. Fresh setups should expect 60 days before the report is useful.
7. Build three reports off it. The three are below in the next section.
8. Disable the other attribution models from the default dashboards. Leave them available for ad-hoc analysis but remove them from any default dashboard the team looks at. The “all models on” failure mode is solved by hiding the alternatives.
The 3 dashboards every revenue team needs after the wiring is done
These three are non-negotiable. Do not build any other attribution dashboard until these are ironclad.
Dashboard 1: Channel-level revenue contribution. W-shaped revenue attribution by channel, last quarter, with QoQ comparison. The single answer to “where is our pipeline coming from?” The CMO quotes this in every QBR.
Dashboard 2: Campaign-level efficiency. Cost per pipeline dollar, by Campaign, last 90 days. Includes paid spend by Campaign / W-shaped pipeline contribution by Campaign. The CFO quotes this when the marketing budget conversation comes up.
Dashboard 3: Funnel-stage attribution. First-touch, lead-creation, and opportunity-creation breakdowns separately, by channel. This is the diagnostic dashboard — when the channel-level dashboard shows something surprising, this one tells you which of the three stages is driving the change.
If your attribution dashboards do not include those three, you are operating with a partial picture.
Common mistakes we see during setup
Skipping the campaign hygiene step. Teams want attribution dashboards before they have wired campaigns. Without Campaign records, the dashboards show channels but not the work the team is actually doing. Useless.
Using formula properties for source-tracking. Formulas re-evaluate retroactively and produce incorrect history when source data changes. Use workflows that stamp values at the right moment and never recalculate.
Inconsistent UTM enforcement. A single team member running a paid campaign without UTMs corrupts a quarter of attribution data. The fix is process — UTM convention documented, link-builder tool, enforcement via channel managers.
Confusing “channel” with “campaign.” Channel is paid social; campaign is “Q1 ABM Push.” Reporting needs both. Tracking only channel loses the actual work the team did.
Skipping the 60-day data-fill window. Teams look at the W-shaped dashboard a week after configuration and pronounce it broken. It is not broken; it is empty. Waiting two months is part of the rollout.
How to roll attribution data into board reporting without overpromising
Two rules for board decks.
Rule 1: Single attribution model in board materials. The CMO picks one model — almost always W-shaped — and uses it consistently across every board update. No “depending on how you measure it” caveats. Pick once.
Rule 2: Attribution as direction, not precision. Multi-touch attribution is directionally accurate, not numerically exact. Use the data to make budget decisions, not to claim that “LinkedIn paid produced exactly $1.4M of pipeline last quarter.” That precision is fictional. The right framing: “LinkedIn paid is our largest pipeline-contribution channel and accounts for roughly a third of attributed revenue. Q3 spend should reflect that.”
Boards respect the second framing. They trust the team less every time the first one slips into a deck.
When you are ready to do this properly
If your current setup matches one of the four failure patterns at the top — all models on, last-touch only, first-touch only, or untrusted dashboard — you are looking at 4–6 weeks of work to fix. The fix is not glamorous; most of it is the campaign and UTM hygiene step, which is the unglamorous foundation everyone wants to skip.
We do this as a focused engagement — typically 4 weeks, paid as a project rather than a retainer. The deliverables are: campaign hygiene completed, the eight-step setup above run, the three dashboards live, and a one-page attribution policy your team can hand to anyone running a campaign in the future. Get in touch if you want us to scope it.
Or, if you are running a fresh implementation, this is included by default. We do not ship Marketing Hub implementations without a working W-shaped attribution setup. The post explaining what that 90-day implementation looks like is here.