SaaS Trial Conversion Optimisation

Most trial conversion problems get treated as a pricing or sales problem when they're actually a product-instrumentation problem. Teams argue about whether the trial should be 14 days or 30, whether to require a credit card upfront, whether to gate the "pro" features or the "power user" features — and almost none of them can tell you, for a trial that ended yesterday, whether the account ever reached the point in the product where the value became obvious. If you can't answer that question, changing trial length is just moving furniture.

Trial length is a time-to-value question, not a preference

There's no universally correct trial length. What matters is how long it genuinely takes a new account to reach their first real "aha" — the point where the product has done something for them they couldn't easily do without it. For a tool where that happens in a single session (an image converter, a scheduling widget), a 7-day trial is generous and a 30-day trial just gives people more time to forget you exist. For something that needs a full billing cycle, a data import, or a team of five to actually get set up — most B2B workflow and ops tools fall here — a 14 to 30 day trial is closer to reality, and cutting it to 7 days mostly produces trials that expire before the user did anything meaningful. The mistake I see most often is copying a competitor's trial length instead of measuring your own activation curve.

Feature gating: gate habit, not price

The instinct is to gate whatever features look expensive to build or whatever you'd charge extra for on the paid tier. That's backwards during a trial. The features worth gating are the ones that create a switching cost — integrations that pull in a customer's existing data, automations that run on a schedule, anything that means the user would lose real work by walking away. Gating those during the trial (with a clear "you get this on day 1 of paid") gives the user a reason to convert instead of just a reason to be annoyed. Gating cosmetic or edge-case features instead — a color theme, a rarely-used export format — saves you nothing and teaches the trial user that upgrading just removes friction rather than adding value.

Engagement signals that actually predict conversion

Login count is a weak signal on its own — it tells you someone opened a tab, not that they got value. The signals worth watching are closer to the core workflow: did the account complete the setup steps that unlock the main use case, did more than one person from the account log in (multiplayer usage is one of the strongest single predictors of B2B conversion), and did usage happen on more than one day. An account that logs a burst of activity on day one and then goes silent is not on track, even if the raw event count looks fine at a glance. The useful exercise is pulling trial accounts that converted and trial accounts that didn't, and looking at what genuinely differed in the first week — not assuming you already know.

SELECT
  t.converted,
  COUNT(DISTINCT t.account_id)                         AS accounts,
  ROUND(AVG(a.distinct_users_active), 1)               AS avg_active_users,
  ROUND(AVG(a.distinct_days_active), 1)                AS avg_active_days,
  ROUND(AVG(CASE WHEN a.completed_setup THEN 1 ELSE 0 END) * 100, 1) AS pct_completed_setup
FROM trials t
JOIN (
  SELECT
    account_id,
    COUNT(DISTINCT user_id)                            AS distinct_users_active,
    COUNT(DISTINCT DATE(event_at))                      AS distinct_days_active,
    BOOL_OR(event_name = 'setup_completed')             AS completed_setup
  FROM product_events
  WHERE event_at BETWEEN trial_start AND trial_start + INTERVAL '14 days'
  GROUP BY account_id
) a ON a.account_id = t.account_id
GROUP BY t.converted;

In-trial intervention beats end-of-trial emails

By the time the "your trial ends in 3 days" email goes out, the outcome is usually already decided — the account either got to value in the first week or it didn't. The higher-leverage move is watching for the accounts that stall early (no second login, setup never finished, only one user ever touched it) and intervening while there's still time to change the trajectory: a check-in from a human, a nudge toward the specific unfinished step, or in self-serve products, an in-app prompt tied to the exact place they got stuck. This only works if the engagement data is visible daily, not reviewed once a month in a dashboard nobody opens.

Credit card upfront changes what you're measuring

Requiring a card at trial start filters out a chunk of low-intent signups before they ever start the clock, which inflates the conversion percentage of the remaining pool without necessarily producing more revenue. It's a legitimate choice, but it changes what "conversion rate" even means, and it isn't directly comparable to a no-card trial's number. If you switch from one model to the other, expect the percentage to move for reasons that have nothing to do with product changes, and don't let that swing get read as a win or a loss it isn't.

Wrapping up

Trial conversion is decided early, mostly by whether the account reaches real value before the clock runs out — not by the exact number of days on the trial or how aggressively features are gated. Set trial length to match actual time-to-value, gate the features that create switching cost rather than the ones that feel premium, and build the habit of watching engagement signals daily so you can intervene while a stalled trial is still salvageable. Everything else is tuning around the edges of that core loop.