Attribution mistakes in B2B are self-concealing: wrong models produce data that looks reasonable until the budget decisions built on them fail. Here are the five most common mistakes and exactly how to fix them.
// THE SHORT VERSION
→The most expensive attribution mistake is not running the wrong model. It is running the wrong model and making a budget decision based on it.
→CRM source fields like HubSpot’s "original source" capture one touchpoint each, first or last. Using them as your attribution view means budget decisions are being made on roughly 10% of the actual buyer journey.
→GA4 already defaults most key events, including form fills and demo requests, to a 90-day lookback window. The shorter 30-day default only applies to acquisition events like first opens and first visits. The real risk is an old property setting, or someone narrowing the window without realizing the cost to a long B2B cycle.
→Optimising Google Ads for form-fill volume finds more form fillers, not more buyers. Smart Bidding chases whatever conversion event you define, and if that event is a form fill, quality quietly falls as volume rises.
→Data-driven attribution needs real conversion volume to find genuine patterns rather than noise, and most B2B firms closing single-digit deals a month don’t have it. If channel credit swings meaningfully week to week with no change in spend, that is usually the tell.
→Attribution is not a setup project. UTM coverage drifts, new channels go untagged, and a model that fit last year’s channel mix may not fit this year’s. It needs a standing quarterly check, not a one-time fix.
Marketing attribution mistakes in B2B are not always obvious because they are self-concealing. A wrong attribution model produces plausible-looking data. The problem only becomes visible months later when the budget decisions made on wrong data start producing wrong results. By that point, the model has been trusted and the mistake is embedded in the planning process.
Mistake 1: treating the CRM source field as attribution
HubSpot’s original source field captures how a contact first arrived at your site. The most recent source field captures the last known source. These are single-touch records: one touchpoint at the start and one at the end of the journey, with nothing in between.
For a buyer who found your blog via organic search in January, saw a LinkedIn retargeting ad in March, read your newsletter for two months, and clicked a Google Ads branded result in April: HubSpot original source shows organic search. Most recent source shows paid search. Neither shows the LinkedIn retargeting, the newsletter, or the five blog posts read between January and April. A founder using these fields to make channel investment decisions is working from 10% of the actual touchpoint data.
The fix: understand that CRM source fields are first-touch and last-touch indicators, not attribution. Use them as diagnostic signals alongside a proper multi-touch attribution view from GA4 or a dedicated attribution platform.
Mistake 2: assuming your lookback window is shorter than it actually is
The lookback window determines how far back Google’s own documentation says GA4 looks for a touchpoint to credit when a conversion happens. Here’s the detail worth getting exactly right: for acquisition events like first-open and first-visit, GA4’s default lookback is 30 days. For every other key event, which is where form fills and demo requests live, the default is already 90 days.
That’s better news than most attribution content admits. But it doesn’t mean this setting is safe to ignore. Three ways it goes wrong in practice: a property that was set up or migrated before this default existed, someone manually narrowing the window to 30 days without realizing which event types that touches, or simple confusion between the acquisition-event default and the conversion-event default when reading a report. A 90-day cycle that’s silently only crediting the last 30 days produces exactly the same symptom either way: channels that build early awareness show almost no credit, and the report reads as a reason to cut them.
The fix: in GA4, go to Admin → Data display → Events → Attribution settings, and check the key event lookback window for your actual conversion events, not the acquisition-event default. Confirm it’s set to 90 days if your sales cycle runs that long. This is a two-minute check, and it’s worth doing even if you never touched the setting, because you can’t tell from the outside whether an old default or a past change is sitting underneath your reports.
Mistake 3: optimising on form-fill volume, not pipeline quality
Most B2B paid search campaigns are optimised to maximise form fills. Google Ads Smart Bidding maximises for the conversion event you define. If that event is a form fill, Smart Bidding will find the traffic most likely to fill in a form, regardless of whether those form fillers are your actual ICP.
The fix: set up offline conversion tracking to pass closed-deal events from the CRM back into Google Ads and GA4. Optimise campaigns for qualified opportunities or closed revenue, not form fills. The volume will fall. The pipeline quality will rise.
Mistake 4: running data-driven attribution with insufficient data
Data-driven attribution is the default in GA4 since 2023. The problem for B2B: it needs enough conversion volume to find genuine, repeatable patterns rather than noise. A B2B firm closing 8 deals a month is working with very little signal for a machine-learning model to train on. When the underlying volume is thin, channel credit distributions can shift meaningfully week to week with no corresponding change in actual channel performance, which is the practical tell that something is off.
There’s no single official GA4-specific figure from Google for exactly how much volume is enough; the widely-repeated benchmark in the industry is around 400 conversions per month for the specific event being modeled, a figure that traces back to Google’s older Universal Analytics documentation and gets applied informally to GA4 by practitioners today. Treat it as a useful rule of thumb, not a Google-stated GA4 rule.
The fix: check your monthly trackable conversion event count in GA4. For B2B, the relevant conversion events are form fills, demo requests, and any other online action tagged as a key event. Closed deals are offline conversions and will not appear in this count unless you have set up Measurement Protocol. If your volume is low relative to that few-hundred-a-month benchmark, or if you’re seeing week-to-week channel credit swings with no spend changes, switch from data-driven to last-click within GA4 while volume builds. For position-based attribution, which GA4 no longer offers natively since November 2023, you need a dedicated tool: Ruler Analytics, Dreamdata, or HubSpot’s custom attribution reporting. The Attribution paths report in GA4 shows path data but does not apply a position-based credit distribution. It cannot substitute for the model itself.
Mistake 5: treating attribution as a setup project, not a discipline
Attribution breaks continuously. UTM parameters are missing from new campaign links. New channels are added without being tagged. The CRM is updated and source attribution fields stop populating. An agency launches a campaign without following the naming convention.
The fix: build attribution maintenance into the quarterly marketing discipline. Monthly: check the GA4 source/medium report for new sources appearing as direct or unidentified. Quarterly: audit UTM coverage across all active paid channels, confirm CRM source fields are populating correctly, and review whether the attribution model still fits the current channel mix and conversion volume.
Worked example: three mistakes running simultaneously
Pinnacle Growth is a B2B SaaS firm running Google Ads, LinkedIn Ads, content marketing, and a weekly newsletter. They had been relying on HubSpot’s original source field as their attribution view, running data-driven attribution in GA4 with a narrowed 30-day lookback window on their key conversion events, and optimising their Google Ads campaigns for form-fill conversions. Three mistakes running simultaneously.
The result: Google Ads showed 58% of original sources in HubSpot. Content showed 11%. LinkedIn showed 4%. Newsletter showed 1%. Based on this, they had cut LinkedIn by 40%, reduced content production, and cancelled the newsletter over the prior six months.
A diagnostic run across all three mistakes revealed: the narrowed 30-day lookback was giving no credit to any touchpoint more than 30 days before form submission, which was where LinkedIn and content operated. The data-driven model at 12 conversions per month was finding noise. And the campaigns were optimising for form fills from a population with a 4% sales-qualified rate.
After restoring the lookback window to 90 days, switching to a rule-based model, and running self-reported attribution for one quarter: in the position-based model, LinkedIn moved to 22% first-touch contribution. Content influence appeared in 38% of conversion paths as a middle-journey touchpoint. Newsletter appeared in 31%. Google Ads remained the primary last-touch channel at 41%, responsible for most actual conversion moments. The distinction: Google Ads was converting demand that LinkedIn, content, and newsletter had built. All four channels were working. The previous setup had been crediting only the one that converted and defunding the three that created the demand it harvested.
The most expensive attribution mistake is not running the wrong model. It is running the wrong model and making a budget decision based on it.
If you want to know where your attribution is wrong and what to fix first, that is exactly what a growth diagnostic covers. Thirty minutes. Book a 360 GTM audit with mrktrs →
Frequently asked questions
What is the most common marketing attribution mistake in B2B?
Confusing the CRM original source field with multi-touch attribution. HubSpot’s original source field captures one touchpoint: the first known source of a contact. It tells you nothing about the eight or nine other touchpoints that influenced the buyer between first contact and close.
Why does a narrowed lookback window hurt B2B attribution specifically?
Whatever the cause, a lookback window that’s shorter than your sales cycle gives zero attribution credit to any touchpoint that occurred before it. For B2B firms with 60 to 180 day sales cycles, that compresses the entire attribution picture into whatever window remains, systematically overvaluing bottom-funnel channels and undervaluing awareness and nurture channels. GA4 defaults most key events to 90 days already, but it’s worth confirming rather than assuming, especially on an older property.
Why does optimising for form fills produce bad attribution decisions?
Form fills are a proxy metric for pipeline, not pipeline itself. When paid search campaigns optimise for form fills, the algorithm finds traffic most likely to complete a form, which is not the same population as traffic most likely to become a customer. The fix is to pass closed-deal data back into Google Ads and GA4 and optimise for qualified pipeline rather than form-fill volume.
How do you know if your attribution data is broken?
Four signals: direct traffic rising without a corresponding rise in branded search or revenue; data-driven model outputs changing significantly week to week without change in channel spend; the HubSpot source field showing significant direct and unknown sources; channel credits not matching the qualitative picture from sales calls.
How often should you audit your attribution setup?
Monthly for UTM coverage and quarterly for the full attribution review including model fit, lookback window, offline conversion tracking, and CRM source field population.
THE DIAGNOSTIC
Which of these five is costing you the most?
One question. One personalised recommendation for your setup. No sales call required.