Rebuilding attribution around closed-won, not MQLs

Why MQL attribution lies to you

MQL-based attribution tells you which campaigns generate leads. That sounds useful until you realise that the campaigns that generate the most MQLs are often the ones that generate the worst pipeline.

The mechanism is straightforward: MQL attribution rewards any touch that precedes a form submission. Top-of-funnel content, broad audience targeting, low-intent keywords - these all produce form fills. A webinar on a generic topic produces a thousand registrants. Your CRM shows a thousand MQLs. Your attribution model credits that webinar as a high-performing channel.

Meanwhile, the niche LinkedIn campaign targeting 200 accounts by title and seniority produced 12 form fills. Same attribution model, different conclusion: low performance. The campaign gets cut. The result: you've just removed the one channel that was reaching actual buyers and doubled down on the one reaching everyone else.

The IRONSCALES case is instructive: when we moved from MQL-based reporting to CRM-connected attribution, they saw 73% more conversions matched versus native CRM connectors - not because the channel performance changed, but because they were finally measuring the right thing.

What closed-won attribution looks like in practice

Closed-won attribution connects the dots backwards from a won deal to every marketing touch that preceded it. The starting point is not the lead - it's the opportunity marked Closed Won in your CRM.

From there, you trace back: which campaigns touched this account before they became a lead? Which ad did the champion interact with? Which content did the economic buyer consume? Which keyword did they search when they were evaluating vendors?

The result is a very different picture of channel performance. The broad webinar with 1,000 registrants may have zero representation in your closed-won deals. The niche LinkedIn campaign targeting 200 accounts may appear in 40% of them. That's the insight that changes budget allocation - and it's invisible in MQL-based reporting.

"The campaigns that generate the most MQLs are often the ones that generate the worst pipeline. Closed-won attribution makes that visible."

The three data layers you need

Building closed-won attribution requires connecting three data layers:

First, CRM closed-won data. The opportunity must be marked with a close date and deal value, and it must have a contact record with a trackable history. If your CRM hygiene is poor, this is the first thing to fix.

Second, ad platform import. Offline conversion import lets you push closed-won events back to Google Ads and LinkedIn as conversion signals. This is how the ad algorithm learns that the person who engaged with campaign X later became revenue - and starts finding more people like them.

Third, GA4 offline conversion upload. For session-level attribution, you need to connect GA4 sessions to CRM deal records via GCLID and LINKER_ID parameters captured in the CRM at lead creation. Without this, GA4 attribution stays at the lead level.

The combination of these three layers gives you a full attribution model: which channel, which campaign, which session, and which ad unit appears in your closed-won deals.

Making the switch without breaking everything

The safest way to move to closed-won attribution is to run it in parallel with your existing MQL reporting for 30 days. Don't rip out the MQL model - add the closed-won layer alongside it.

After 30 days, compare: which campaigns appear in MQL reports but not in closed-won? Which appear in closed-won but not in MQL? The differences tell you where your current model is misallocating budget.

Once you're confident in the closed-won model, you can start shifting optimisation targets. Push closed-won signals back to your ad platforms as the primary conversion event. Within 4–8 weeks, the algorithms start finding audiences that look like buyers rather than audiences that look like form-fillers.