Should You Extend a Free Trial? The SaaS Extension Decision Grid

Yes, you should extend a free trial when a user asks, but only once, only for a fixed number of days, and only with something else changing. Grant a booked call to fit accounts that have not activated, a convert-now offer to fit accounts that have, and a self-serve extension to everyone else. At a $3,000 contract value and an assumed $45 cost per call, a call that lifts conversion by 1.5 points pays for itself.

Rules of thumb

• Allow one extension per account. GitLab’s 2020 public proposal for an in-app trial extension specified it “one time per namespace”, whether the trial is active or expired (GitLab issue 271627).
• Never extend with nothing else changing. Lincoln Murphy of Sixteen Ventures names this as the failure: “Just another 7 days and then what?” (Sixteen Ventures).
• A call per request pays when contract value x conversion lift is bigger than the call’s cost: at $3,000 and an assumed $45 call, the lift needed is 1.5 points (worked below).
• At a $300 contract value the same assumed $45 call needs a 15-point lift, so low-price plans get the self-serve extension, not a human.
• Handling requests by hand adds up: in 2020 GitLab said its sales and support teams were averaging about 60 requests a month at about 5 minutes each, which is 5 hours a month.

Should you extend a free trial? The Extension Decision Grid

The Extension Decision Grid decides whether to extend a free trial from two facts: ideal customer profile (ICP) fit and whether the account has activated.

ICP fit means the company size, role and use case your paying customers share. An activation signal is one of three actions: the user connected their own data, invited a teammate, or completed the key action your best customers complete early. Each cell gets one response.

The grid makes the response depend on evidence instead of on who asked politely. It also stops your sales team from spending calls on accounts that would never pay at your price.

Free trial extension FAQ

Should you extend a free trial automatically?

Only for low-value or non-fit accounts. An automatic, one-time extension costs no human time and suits the self-serve cell of the grid. For fit accounts, make the extension depend on a booked call, because the reason they stalled will not fix itself.

How many times should you extend a SaaS trial?

Once. GitLab’s 2020 public proposal for an in-app trial extension specified it one time per namespace, for active or expired trials (GitLab issue 271627). A second request calls for a sales conversation or a different plan.

When should you offer a trial extension?

Offer the extension or the call a few days before the trial ends. Stripe’s trial_will_end webhook fires 3 days before the trial ends by default, which is a ready-made trigger (Stripe Billing trials documentation).

Is an extension request a good sign or a bad sign?

Both. It shows intent, since the user bothered to ask, but Lincoln Murphy of Sixteen Ventures treats it as a symptom that the trial is not getting users to success. Grant it, then find out why.

Should you offer a discount instead of an extension?

Only in the convert-now cell, where the account has activated and fits. A discount to an account that never activated pays for a product it has not used, and often leaves the stall in place.

Which activation signals should decide a trial extension?

Trial extension decisions should rest on three activation signals in your product data: connected data, an invited teammate, and the key action that predicts payment.

• Connected data. An import, an integration or a synced account. A user who has put real data in has sunk effort into the trial and is often stuck on configuration, not on interest.
• Invited teammate. A second user means a second person who has to agree. Murphy lists “need to get other folks to try it” and an internal demo among the common reasons for extension requests (Sixteen Ventures).
• Key action. The first report, campaign, deployment or invoice your converted customers complete early. Pull it from your own paid cohort: the action most paying accounts did in week one that most churned trials never did.

If you have not defined an activation event yet, do that before you build an extension policy; our guide to raising SaaS trial activation rate covers how to choose it.

Why do trial extensions fail when nothing else changes?

Trial extensions fail when the only change is a new end date, because whatever stopped the first trial converting is still there.

Lincoln Murphy set this out in his 2014 essay on SaaS free trial extension requests. He agrees that a user who asks “should probably” be given the extension, but calls the request “a symptom” of a deeper problem. In his experience the main reason is that the user got busy, the trial ran out, and they now want to “really try it this time”. His conditions for granting one: find out why, agree what a successful trial looks like, plan the steps to get there, and bring in anyone else who has to approve. He is also against automated and incentivised extensions.

Churnkey takes a softer line. Its guide to trial extensions in cancel flows says to always offer an extension to trial users who are cancelling, and suggests extending by the same length as the original trial. The two views reconcile in the grid: the automated extension Churnkey describes suits the lower-value cells, and Murphy’s conversation suits the fit accounts where a sale is worth a human’s time.

What is one call per trial extension request worth?

One call per trial extension request is worth contract value times the conversion lift, minus the call’s cost: $255 at $3,000 and a 10-point lift.

No public benchmark measures how much a call lifts conversion on extended trials, so the model below uses labelled assumptions in three bands. Replace them with your own numbers. Fixed inputs: a $3,000 first-year contract value (a $250 a month plan), and a $45 cost per call (45 minutes of prep, call and notes at a $60 an hour loaded rate). Volume uses the roughly 60 requests a month GitLab reported in 2020 as a reference point.

The break-even is the useful number. Call cost divided by contract value gives the lift a call must create: $45 / $3,000 is 1.5 points. At a $300 contract value it is $45 / $300, or 15 points, which is why low-price plans belong in the self-serve cell. Two cautions: the lift is first-year contract value, not cash this month, and accounts that would have converted anyway inflate it unless you hold out a control group. The live page on trial-to-paid conversion with outbound conversations covers when in a trial those calls land best; this table only prices the extension moment.

How long should a SaaS trial extension be, and how many should you allow?

A SaaS trial extension should be one per account, only as long as the agreed success milestone needs, with a hard end date.

• Length. Churnkey’s rule of thumb is the same length as the original trial. For the booked-call cell, set the length from the milestone agreed on the call instead: if the blocker is a data import, the days to import plus a few days to use it.
• Count. One. GitLab’s 2020 proposal limited its in-app extension to one per namespace. A second request is a sign that the account needs a sales conversation or a different plan, not more time.
• Timing. Ask before the trial ends, not after. Stripe’s customer.subscription.trial_will_end event fires 3 days before a trial ends by default, which gives your system a natural trigger to offer the call or the extension (Stripe docs).

What should happen to trials you let expire?

Trials you let expire should enter a 30-day reactivation sequence, because the account, its data and its contacts stay valuable after the clock runs out.

A simple sequence: day 1, confirm the data is kept and the account can restart on a paid plan; day 7, one short note tied to the key action they never completed; day 14, a case or template from their use case; day 30, a final check-in and a question about timing. Keep each message to one ask. If one of these accounts later raises its fit (more seats, a bigger company), it moves into a convert-now or booked-call cell. For accounts that paid and then left, the separate playbook on winning back churned SaaS customers applies instead, and if expired trials are a large share of your sign-ups, check the diagnosis in why your trial-to-paid conversion rate is low.

What does running trial extension calls in-house cost?

Running trial extension calls in-house costs about 5 hours a month of triage at GitLab’s volume, and about 45 hours if all 60 requests got a call.

The Extension Decision Grid shrinks that. If a quarter of requests land in the booked-call cell, 15 calls at 45 minutes each is about 11 hours a month, plus building the activation flags, the self-serve extension button and the reactivation emails, which is often a few days of product and lifecycle work. Do it yourself if you get a handful of requests a month and a founder or account executive can take the calls within a day. Hand it off when requests outrun the team or calls slip past the trial end date. LeadsNow books these conversations using AI calling and SMS follow-up (50,769+ AI-booked appointments since 2017); its AI appointment setting service is billed on results, as a revenue share, a fee per appointment, or a mix of both, and the broader SaaS lead generation guide shows where trial follow-up fits in the funnel.

Sources

1. Sixteen Ventures, Lincoln Murphy: SaaS Free Trial Extension Requests are a Bad Sign (15 July 2014)
2. GitLab issue 271627: Allow users to extend SaaS trial in-app
3. Churnkey: How to use trial extensions to convert more trialing users (2 July 2024)
4. Stripe Docs: Billing trials