Product Qualified Leads: PQL Definition, Scoring & Examples

A product qualified lead (PQL) is a lead that has demonstrated buying intent through product usage — a trial or freemium account that crossed a meaningful value threshold. ProductLed defines it as *"a lead who has experienced meaningful value using your product through a free trial or freemium model"* — qualification by what they did, not what they downloaded.

The concept exists because product-led companies noticed their best pipeline was not coming from whitepaper forms. It was coming from accounts already using the product — and the qualifying event was invisible to a marketing-qualification lens. The PQL makes that event first-class: usage is the new form fill.

The comparison is not a verdict — the two measure different funnels. A company with no free product cannot mint PQLs; a PLG company drowning in signups needs PQL discipline precisely because raw signups are the new vanity metric. Hybrid teams run both and route them differently.

And the failure modes mirror each other: MQLs over-count the curious, PQLs over-count the free-forever. The fix in both cases is the same — tie the definition to behavior that historically correlated with paying and staying, not just activity.

The canonical examples, as cataloged by ProductLed: Slack — an account reaching its 2,000-message limit. Facebook — a user adding 7 friends. Drift — 100 conversations on the website. Each encodes the same discovery: a usage threshold past which users stuck around, found empirically and then promoted to the qualification bar.

Finding yours is correlation work, not creativity work:

1. 1Pull the accounts that converted to paid and retained.
2. 2Pull the accounts that signed up and evaporated.
3. 3Find the early behaviors that separate the two populations — invited a teammate, connected an integration, hit a usage volume, returned three days in a row.
4. 4Pick the 2–4 with the strongest correlation and the clearest product meaning: those are your milestones.

Analytics platforms like Amplitude and Pendo exist substantially to make this correlation work tractable; in BusinessMCP the same idea ships as product milestones — you mark which tracked goals are product milestones, and accounts crossing them surface as product-qualified with the behavior attached.

A milestone alone is half the score. The other half is the same fit axis every scoring system needs: a student hitting your usage threshold and a 200-person ICP-fit company hitting it are not the same lead. PQL scoring = product milestones × account fit — the multiplicative logic from our lead scoring guide applied inside the product.

Score at the account level, not just the user level. Three users from one company each poking around is a stronger signal than any of them individually — and the buying conversation will happen with the account, likely involving people who never touched the trial (the buying committee problem, inside PLG).

Decay applies too: a milestone crossed yesterday outranks one crossed last quarter. A PQL is a trigger with a shelf life, like everything else in the signal hierarchy.

Product led sales is the routing layer on top: which PQLs get a human, and what the human does. The rule that keeps it from ruining the self-serve experience: the touch helps with what they are already doing. A user who just connected an integration gets setup help and a relevant next step — not a discovery call script.

A practical routing split: high-fit accounts crossing strong milestones get a human touch within a day (this is a same-day trigger — speed logic applies); medium signals get in-product nudges and lifecycle email; low-fit or early accounts stay purely self-serve. The touch itself references the milestone: congratulate, unblock, and offer the step that compounds their momentum.

One more PQL flavor deserves its own alert: the trial that goes suddenly quiet after strong activity. Something blocked them — and a short, helpful "did something break?" note is among the highest-reply messages in existence. An AI SDR handles both flavors well precisely because the trigger, the context, and the right small ask are all machine-readable.