PLG Metrics Every B2B Product Team Should Track
Activation rates and time-to-value separate revenue-driving signals from vanity metrics.
Picture a board meeting with a slide full of activation rates, engagement scores, and month-over-month signup growth. A board member asks what it means for revenue next quarter, and the room goes quiet, because the slide was built to describe behavior, not to defend a growth strategy. That scene plays out at B2B software companies constantly, and it points to a real gap in how PLG gets measured. Most product teams investing in product-led growth are optimizing for metrics like signups, traffic, and feature clicks that have no reliable relationship to revenue, while the one metric that does, activation, goes untracked by a large majority of them. The stakes of that gap keep rising. Roughly 58% of B2B SaaS companies now run some form of PLG, and 91% of them plan to increase that investment. Despite that scale of commitment, only about 34% of PLG companies actively track activation, the single metric that best predicts whether a free user ever becomes a paying one, even though between 40% and 60% of free users never reach activation at all. Teams aren't short on data; they're measuring the surface of the product instead of the moment inside it that actually predicts who pays.
Activation as the controlling variable in a PLG funnel
Activation sits at the center of a PLG funnel because every outcome that matters, conversion, retention, expansion, has to pass through it first. Spend on acquisition, traffic growth, and signup volume only turns into revenue if users cross the activation threshold. Improve any metric downstream of activation without fixing activation itself, and the growth strategy compounds a leak instead of sealing it. OpenView's benchmark data gives this problem a name: "zombie users," people who sign up, click around, and leave without ever touching the product's core value, and OpenView estimates they make up 40 to 60% of a typical PLG free-user base. That share is close to half the funnel, sometimes more, producing signups and usage logs that look like progress but predict nothing about revenue.
The timing of activation carries its own signal. Users who reach a value realization event within the first 72 hours convert to paid at a materially higher rate within 30 days than users who activate late. That single fact reorders the whole measurement hierarchy: D3 activation rate becomes a leading indicator of expansion revenue, while 7-day retention, the metric most teams default to, only confirms what already happened. In practice, that means defining activation as a specific, measurable in-product event, a workflow completed, a report sent, a teammate invited, rather than a page view or session length, and ensuring a new user can reach that aha moment within 15 minutes without assistance, since a product that fails this test isn't yet ready for a self-serve PLG motion. Acting on a leading indicator gives a team months of runway to fix a broken onboarding flow. Waiting on a lagging one means finding out after the damage is done.
Cursor illustrates how far reducing time-to-value can carry a company. That speed is remarkable, but Cursor also sits in a category where individual developers already have high willingness to pay and almost no switching cost, conditions most B2B products don't share. The month-over-month doubling isn't a template to copy. The lesson that generalizes is the cause behind it: cutting time-to-value as close to zero as the product allows matters because every hour a new user spends before reaching value is an hour the funnel can lose them.
Notion makes the same point from a different angle. The company grew its user base 5x in 2020, driven in part by a community-built template library that gave new users something to start from immediately, removing the blank-workspace problem that stalls so many productivity tools. Instead of staring at an empty page, a new user found value inside the first session, and Notion went on to reach a $10 billion valuation with more than 20 million users. Once it's clear that activation is the variable controlling everything downstream, the next question is mechanical: how does a team actually find and define that moment inside its own product?
Defining your activation event before building a dashboard
No correctly defined activation event means no metric framework holds up. Skipping this step means a team ends up building dashboards that track the wrong moment and then optimizes its entire product toward it. Getting the definition right starts with the data already sitting in product analytics. Pull a cohort of accounts that converted to paid over the past 12 months, then find the single feature interaction or workflow completion that appears in the highest percentage of those converted accounts and the lowest percentage of accounts that churned. That interaction, the one that separates the accounts that stuck around from the ones that didn't, is the activation event.
The definition has to describe an action rather than a session. "Workflow completed with at least two steps" or "report sent" tells a team something concrete happened. "Logged in" or "visited pricing page" tells them almost nothing, because those events happen whether or not the user ever touched real value.
Once the event is defined, it has to survive a timing test. If a new user cannot reach that first moment of value within 15 minutes without help from a support rep or a sales call, the onboarding flow needs to be redesigned before activation tracking will ever produce a clean signal. That 15-minute window is a forcing function, not a stretch goal for a design team to aspire to someday: if the product can't clear it, no dashboard built on top of the current onboarding flow will tell the team anything useful, because the underlying event it's measuring is too slow to matter.
The tiered metric stack: which numbers belong at which cadence
With the activation event defined, seven metrics is a ceiling, not a starting point: a team just building this system should pick three or four that map to their biggest current gap and prove the reporting cadence before adding more. Tier 1 is the activation layer, tracked weekly at the cohort level, and it's the tier a team acts on in real time. Activation rate belongs here, with 40% counting as a good benchmark and 70% or higher counting as elite by 2026 standards. Time-to-value is the metric alongside it: under 24 hours is good, under 15 minutes is elite, and this number correlates directly with how fast users convert and how quickly a company recovers its customer acquisition cost. Free-to-paid conversion rounds out the tier, with 2 to 5% counting as a solid baseline and top performers reaching 5 to 10% or higher. Inside this tier, activation rate deserves more weight than 7-day retention, since activation predicts what's coming while retention only confirms what already happened. A team just starting to build this system doesn't need every metric in the stack at once. Three or four numbers that map to the biggest current gap, tracked consistently, prove the reporting cadence works before anything else gets added to the stack.
Tier 2 is feature adoption depth, reviewed monthly. This tier answers a different question than Tier 1: not whether users got to value once, but whether the product is becoming part of a real workflow or just getting tolerated, which makes it the leading indicator for renewal risk. The framing matters as much as the number itself. Measuring the percentage of all users who touched a given feature hides the real signal, because it mixes users for whom the feature is irrelevant in with users for whom it matters. The sharper version asks what percentage of users for whom a feature is relevant used it repeatedly, a distinction that aggregate adoption numbers tend to bury. Product-qualified lead rate also belongs in this tier: it measures the share of users whose in-product behavior signals they're ready to buy or expand, handing sales a pipeline verified by actual usage instead of a marketing-scored guess. PQL definitions typically combine an activation milestone with a buying-intent signal, like a teammate invite, a usage threshold crossed, or access to a paid feature, though some frameworks qualify users on usage threshold alone.
Tier 3 is revenue retention, reviewed quarterly, and it functions as the scorecard rather than the dashboard. Net revenue retention is the compounding number that separates durable PLG growth from growth that leaks as fast as it fills: 100 to 110% counts as good, and anything above 130% counts as elite. NRR carries more weight with most SaaS boards than almost any other figure in the stack, because every other metric in the framework either feeds into NRR or helps explain why it moved. Self-service conversion rate belongs in this tier too: a rising share of paid conversions closing without a sales rep touching the deal is direct evidence the product motion is working, and it's a number CFOs can tie straight to sales efficiency. For context, 2026 median NRR across B2B SaaS sits in the same good-to-elite range, while enterprise segments above $100k in annual contract value post a noticeably higher median. A team that tracks only Tier 3 and skips Tier 1 finds out its problems once a quarter, months after the activation data would have already shown the same story forming.
How the right metrics shift as ARR grows
The metrics worth the most attention at any given moment depend on the growth constraint a company is actually up against, and that constraint changes materially as ARR climbs, so no single universal dashboard serves every stage well.
Below a certain point in ARR, the motion is pure PLG, self-serve by default with a founder stepping in to close the occasional deal by hand. The entire focus at this stage is defining and reaching the activation event, with conversion rate and time-to-value as the weekly metrics. NRR targets move to a range of 90 to 105%, and building out the PQL framework becomes the top investment, since this is the stage where thinking in old-style marketing-qualified-lead terms starts capping how much the funnel can convert.
As ARR grows into the $10M–$50M range, the motion turns hybrid, PLG combined with sales-led growth, with sales proactively owning expansion conversations rather than waiting for signals, the NRR target rising to 108% and roughly 67% of companies above $10M ARR running a hybrid motion whether they call it that or not. The real question at this stage is when to bring in sales, not whether to.
Further up the scale, named account executives take over the largest accounts, and the NRR target rises again, higher than the hybrid-motion target. Datadog is the clearest reference point for what this looks like at scale: the company reached $3.4 billion in revenue for fiscal year 2025, up 28% year-over-year, with 603 customers generating $1M+ in ARR by Q3 2025. The product still opens the door. Sales widens it from there.
Stage isn't the only variable at play. ICP segmentation interacts with ARR directly: pure self-serve PLG holds up well for deals up to roughly $10,000 in annual contract value, but anything above $25,000 ACV needs a sales-assist layer from day one, because buying committees and security reviews slow down self-serve conversion no matter how well activation is performing.
What founder-level customer attention reveals that dashboards cannot
None of the tiers above exist without a correctly identified activation event, and no dashboard finds that event on its own. It gets found through direct contact with customers: watching where they stall, listening to what they describe as the moment the product finally clicked, and noticing which feature appears repeatedly in the accounts that renewed. A product analytics platform can confirm a hypothesis once it exists. It can't generate the hypothesis itself.
This is why the earliest and most important work in a PLG metrics program tends to happen away from any dashboard at all: in sales calls, support tickets, and onboarding sessions, where a founder or an early product lead watches a real user try to get value out of the product for the first time. That direct exposure is what turns a vague sense of "users seem to like this feature" into a specific, testable definition, the kind that can survive the cohort analysis described earlier and hold up as a number tracked weekly. Dashboards take over from there. They don't replace the attention that found the signal to begin with, which is the problem Userlens, an AI CSM platform whose agent Lumi reads each customer's actual product usage and database state then writes that person a grounded next step rather than producing another chart, was built to address at scale.
