PQL scoring · Built for B2B PLG

PQL scoring that reads the whole account

A usage threshold tells you one person got busy. Trailspark evaluates every product user at an account together with the marketing, CRM, and web evidence around them, against the definition of good you write in plain English, then sends the propensity and the reason behind it to your CRM.

Defining PQL

What is a Product Qualified Lead

A Product Qualified Lead is a user who has shown buying intent through how they use your product rather than through marketing engagement. Someone who connects a production data source is a PQL. Someone who downloads an ebook is an MQL. Same company, different evidence, different qualification story.

PQLs exist because product-led growth changed the funnel. A free trial signup can reach the buying decision before any marketing touchpoint fires. If your evaluation only reads marketing engagement, those users are invisible to sales until someone notices them. By that point, the buying window may have closed.

The term is useful for naming where the evidence came from. It is weaker as a verdict. A PQL is one person's activity inside one product. The decision your team actually makes, whether to work this account now and for what, belongs to the account rather than to whichever user happened to cross a threshold.

Where marketing scores run out

Why marketing-only scoring misses PLG conversion

Three patterns that show up repeatedly in teams running a product-led funnel through a marketing-only scoring model.

Marketing engagement lags product activity

A user connecting a production data source on day three of their trial is stronger evidence than that same user downloading an ebook two weeks later. Marketing-only scoring sees the ebook and never sees the data source.

Product users skip the funnel

A developer signs up with a personal email, gets the product running, and only pulls in their company stakeholders two weeks later. No form fill, no opt-in, no MQL. By the time marketing sees them, the buying group is already forming.

A fixed point value ages badly

Points-based PQL models assign one permanent value to an event. Ship a free tier and the feature that used to signal commitment becomes table stakes, but the point value does not know that. What good looks like should be something you can restate in a sentence.

What counts as product data

Product usage is evidence, not a universal score

Two vendors can both say they read product data and mean completely different things. A plan field copied into the CRM and a timestamped record of what every user built this week are both called product data. They are not the same substrate to evaluate an account from.

Product data fidelity, thinnest to richest

Plan and CRM fields

Tier, seat count, trial end date, written onto the account record. This tells you what the account signed up for. It says nothing about whether anyone opened the product this month.

Usage rollups

Monthly active users, total event count, a last-seen date. Direction without detail. A rollup cannot separate four people steadily building something from one person clicking around.

Milestone flags

Booleans: activated, invited a teammate, connected a source. Closer to meaning, but a flag carries no timestamp, no order, and no sense of how much work the step took.

Event streams

Trailspark evaluates here

Named, timestamped actions attributed to individual users. Now the order matters: what someone did, how fast, and what they did next.

Workspace and account telemetry

Trailspark evaluates here

Integrations connected, permissions configured, data volume committed, and which capabilities are live across the whole workspace. The state an account has actually built, not just the clicks it produced.

Trailspark is built for the bottom of that ladder. Forward events from Segment, RudderStack, Amplitude, Mixpanel, your warehouse, or your own backend, and the account story keeps the names, the timestamps, and the order they happened in.

The evidence

What full-fidelity product evidence looks like

Named actions and workspace state, not a usage number. Your product will have its own vocabulary for these. You describe which ones matter for your motion in plain English, and Trailspark evaluates the account against that description.

Activation milestones

User completed setup flow within 24 hours of signup
Connected a production data source, not sample data
Invited their first teammate
Created a project or workspace that matches production naming

Collaboration breadth

Three or more teammates active in the same workspace
Cross-functional invites, sales plus marketing or engineering plus ops
Shared outputs internally via external links
Recurring usage across multiple users, not one account owner

Workspace commitment

Uploaded real volume, not sample content or test data
Connected upstream or downstream integrations
Built automations or workflows beyond the template library
Configured role-based permissions beyond defaults

Sequence and velocity

Active on multiple business days in the first two weeks
Session depth increasing week over week
Configuration steps completed in quick succession, not spread over a month
API or webhook usage indicating a production integration

Account-level scoring

The threshold is per user. The decision is per account.

Score each user on their own and every person gets measured against a bar built for one person. Several individually modest users at one account can be a far stronger account story than one highly active tire-kicker. Per-user scoring hands you the tire-kicker, because the tire-kicker is the only one who cleared the bar.

Four people at the same company, each moderately active in the same workspace, each sitting below your PQL threshold, each invisible. Trailspark resolves product users into the account they belong to first, so the unit being evaluated is the account and the denominator is the group.

Today · scattered across your stack

Product analytics

3 trial users building, integrating, hitting limits

all below PQL · invisible

Marketing automation

1 VP, whitepaper + webinar

MQLs · worked alone

CRM · web · billing

more signals, never joined up

siloed

Different platforms, separate scores, no shared view of the account.

With Trailspark

Brightwave, Inc· buying group of 4Propensity HIGH

Roles inferred from behavior + firmographics

ChampionPriya Nair · VP Platform
Technical EvaluatorSam Cole · Staff Engineer
Decision MakerJ. Chen · VP Engineering
⚠Economic Buyermissing

An engineer runs a live technical evaluation while a VP-level champion engages your scaling and pricing content and a known decision maker re-engages in your CRM. No Economic Buyer yet; nurture Finance now to de-risk the close.

Same signals you already collect. A completely different thing to act on.

Lenses

Same product activity. Different meaning.

The events do not change. What they are worth depends on the account they came from and the question you are asking about it.

Segment context

Forty activated users at a 150-person company can be deep adoption. The same forty at a 20,000-person company can still be a small foothold. Identical usage, very different share of the account.

Motion context

Three people visit pricing and an admin connects an enterprise integration. On a free trial that supports an acquisition call. On an existing customer the relevant question is expansion. The activity is the same. The business question is not.

Surrounding evidence

Low event throughput on its own reads as disengagement. Next to a workspace that just connected four integrations and turned on alerting, it reads as a team still building. The evidence means something different depending on what sits next to it.

Lifecycle timing

Light usage on day 4 of a trial and light usage on day 40 are not the same evidence. Sequence, velocity, and recency change the call, not just whether an event ever fired.

You define as many evaluation lenses as your business needs. Trial acquisition, free-plan conversion, expansion, renewal, a segment, a user type. Each one is a plain-English description of what good looks like, and each account is evaluated through the lens that applies to it.

Reasoned evaluation

The reason is part of the evaluation

A factor list shows what moved a number. It does not tell you why conflicting evidence supports this result for this account. The written reason arrives with the propensity, and it has to survive a rep reading it out loud.

Trial account, day 6Lens: trial acquisitionHigh propensity

“One identified user and throughput is thin: no log volume, and APM event counts well below an established workspace. Against that, this workspace connected four applications, enabled deploy tracking, turned on log ingestion, and configured alerting inside six days. The breadth and speed of that setup outweigh the low throughput here, because throughput is exactly what a workspace still being wired up would not have yet. The open gap is people: no second user has been invited and no economic buyer has appeared.”

The weak throughput is in the reason. It was not dropped because the call came out positive, and it did not sink the call because the evidence around it explains why it is missing at this point in the trial. That is the difference between a factor breakdown and a reasoned evaluation.

Step by step

From product event to a call your CRM can act on

Ingest, resolve, evaluate, activate. Trailspark evaluates. The systems your team already works in execute.

Ingest the product evidence

Forward events from Segment, RudderStack, Amplitude, Mixpanel, your warehouse, or directly from your backend. Any event structure, mapped to named signals during setup, alongside marketing engagement, CRM history, and web behavior.

Resolve users to the account

Product signups, CRM contacts, and marketing leads at the same account resolve into one organization view, so every product user is evaluated as part of the account they belong to. Anonymous history attaches once the person is identified.

Evaluate against the lens that applies

Trailspark picks the evaluation lens for this account, trial acquisition or expansion or a segment you defined, and judges the full account story against the plain-English definition of good you wrote for it. Re-evaluation runs when the evidence changes enough to warrant a fresh look.

Activate in the tools you already use

Propensity, the written reason, the lens that was applied, and buying-group role coverage land as structured fields in Salesforce or HubSpot, with Slack alerts routed per ICP. No new workspace for your reps to live in.

How PQL scoring works: common questions

A Product Qualified Lead is a user who has shown buying intent through product usage. Unlike a Marketing Qualified Lead who signaled intent through marketing engagement (a content download, a form fill), a PQL signaled intent through the product itself (creating a project, inviting teammates, connecting data sources). The term names where the evidence came from. It does not settle whether the account behind that user is worth working, which is an account-level question.

MQL, PQL, and SQL are stages in a funnel. MQL captures marketing intent; PQL captures product intent; SQL is a human-qualified MQL or PQL a rep has accepted. Trailspark reads product activity and marketing engagement together as one account story rather than routing each in isolation, so a buying group forming across both shows up as one account instead of several separate, sub-threshold leads.

Points-based models assign a fixed value to isolated events and add them up. Trailspark reads the whole account story, every product user plus the marketing, CRM, and web evidence around them, and evaluates it against the definition of good you wrote in plain English for the motion that applies. The output is a propensity with account-specific reasoning that reconciles supporting and counter-evidence, not a point total with a factor list attached.

No. Product usage is evidence, and it means different things in different places. Forty activated users at a 150-person company can be deep adoption; the same forty at a 20,000-person company can still be a foothold. Heavy activity from one user with no one else at the account behind them is a weaker story than several moderately active users across a workspace. Trailspark evaluates the product evidence in the context of the account, the segment, the motion, and where the account sits in its lifecycle.

As much fidelity as you can send. A plan field in the CRM, a monthly usage rollup, and a stream of named, timestamped user actions are all called product data, but only the last one shows sequence and velocity. Add workspace telemetry on top of it: integrations connected, permissions configured, data volume committed, and which capabilities are live. Trailspark evaluates from that end of the range, and you describe in plain English which of it matters for your motion.

Yes. Trailspark writes the propensity, the reason, and the evaluated lens to dedicated custom fields in your CRM, separate from existing native scoring. A sensible rollout: run both in parallel for a month, compare the resulting pipeline against closed-won outcomes, and consolidate on whichever your team trusts.

Product analytics tools (Amplitude, Mixpanel, Heap) measure what users do. Trailspark takes that activity, resolves it to the account, and judges what it means for the decision in front of you. They complement each other: analytics tells you your funnel shape, account evaluation tells you which accounts are worth sales attention now and why. Trailspark ingests events from your analytics tool and returns a call your CRM and sales team can act on.

Yes, and they usually deserve separate lenses. Good on a 14-day trial and good on a free plan that someone has held for eight months are different descriptions, even when the underlying events look similar. You write each one in plain English, set which accounts each lens applies to, and Trailspark evaluates every account through the lens that fits it.

Keep reading

Article

Product Usage Milestones That Predict Conversion

Seven specific product usage milestones and how to read them at account level.

Product

AI Lead Scoring That Evaluates the Whole Account

Product activity, marketing engagement, CRM history, and account context evaluated together against your definition of good.

Evaluate your product users as the accounts they belong to.

Connect your first event source on the free plan, write what good looks like for your trial motion in plain English, and read the calls Trailspark returns.