Guide

Free trial optimization checklist

A stage by stage checklist for a free trial with a card on file, from the decisions before launch to the first renewal, with one number to watch per stage.

By Meralume Labs. Published . 7 min read

Work down the trial in time order: decisions before launch, the first minutes, the first days, the middle, the days before the charge, the charge itself and the first renewal. At each stage fix what is broken, write down one number to watch, and move on.

How to use this list

Most trial advice is organized by tactic: urgency, social proof, emails. A checklist organized by time is more useful when something is already live, because it matches how a customer experiences the trial and makes gaps obvious. A gap at the start of the list is nearly always worth more than a polish at the end, since every later stage only sees the people who survived the earlier ones.

Each stage below has a short list of checks, a reason for each and a measurement. You will not need all of it on day one. If you are short of time, do the stages in order and stop when you find the first thing that is plainly wrong. The trial page checklist goes deeper on the signup page, and the failed payments guide goes deeper on the charge. This guide ties the whole path together.

Stage 0: decisions before launch

  1. Write the one sentence that says what a person should have achieved by the end of the trial. Everything later is judged against it.
  2. Choose the trial length from the time that sentence takes, not from habit. In the ChartMogul study of 200 products the most common length is 14 days, but that records what others picked.
  3. Decide whether to ask for a card. If you do not, choose what the subscription does when the trial ends without one: Stripe offers cancel, pause or invoice.
  4. Decide who gets a trial. A trial per company, per email domain or per card is a policy choice that you should make before the first abuse, not after.
  5. Name the metric for the trial: paying customers per thousand visitors, with trial to paid rate by start week underneath it.

The ChartMogul report is blunt about the reason for the third check: it recommends asking three questions before changing a trial, namely whether you want more people to use the product or more people to buy it, when users reach the first moment of value, and what each free user costs you. Write the answers down. They decide the rest.

Stage 1: the first minutes after signup

  1. The confirmation screen repeats the offer in the same words as the page, including the end date and the first charge date.
  2. A welcome email arrives within minutes and contains one next step, not five.
  3. The card, if taken, is checked at signup so a bad card fails here and not in two weeks. A hold of the plan price does this, as the failed payments guide explains.
  4. The customer lands inside the product, not on a settings page.

Measure the share of new trials that reach a first session within a day. If it is low, nothing downstream can be fixed with emails alone, because the people you would be emailing have not seen the product.

Stage 2: the first days

  1. Use sample data or a guided task so the product shows a result before the person has done any setup.
  2. Mark the moment the product has done something useful, as an event you can count.
  3. Send a message only when the customer has not reached that moment, and make the message point to the step they are missing.
  4. Make help easy to reach from inside the product.

The measurement here is time to that first useful result, and the share of trials that ever reach it. This is the stage where trials are won or lost for most products, and it is a product and onboarding problem, not a billing one. Onboarding is a subject of its own, and the point here is only to measure it.

Stage 3: the middle of the trial

  1. Tie messages to behavior. A person who invited a teammate gets a different message from a person who has not logged in since day one.
  2. Show progress inside the product so the person can see what they have done with it.
  3. Offer a short call to the people whose accounts look like a fit for your larger plans, and keep that offer plain.
  4. Do not hide the end date. A person who knows when the trial ends makes a decision. A person who finds out from a charge makes a complaint.

Count return sessions per trial and the share of trials that did the key action at least twice. A trial with one visit is a trial that will not convert, whatever the emails say.

Stage 4: the days before the charge

  1. Send a reminder that names the plan, the amount, the date and how to cancel. Stripe can send a trial ending reminder 7 days before the end and link it to a page where the customer updates a card.
  2. Listen for the event Stripe sends a few days before the trial ends, and use it to confirm a payment method exists.
  3. If you hold the card, make sure the hold will still be valid at capture. Stripe lists validity windows of days for card holds, and a trial longer than the window needs a renewed hold.
  4. Send the expiring card notice, which Stripe offers a month before a card on file expires.
  5. Make cancelling as easy as signing up. Anything harder produces refund requests and disputes.

Measure the share of trials with a valid, usable card at five days out. That number tells you how many charges will fail before they have been attempted.

Stage 5: the charge itself

  1. Confirm the charge went through and tell the customer, with a receipt.
  2. For a failure, branch by type. Soft declines go to retries. Hard declines go to a card update request, because Stripe does not retry them until a new payment method exists.
  3. Set the retry policy deliberately. Stripe recommends 8 tries within 2 weeks for Smart Retries and allows a custom schedule of up to three retries.
  4. Choose what happens to the subscription when recovery fails: cancel, mark unpaid, leave past due or pause.
  5. Alert a person when a high-value trial fails, so someone can write to the customer.

The numbers are the failed charge rate at first charge, the recovery rate within the retry window, and the share of failures that were hard declines. The dunning and involuntary churn glossary pages define the terms so your team measures the same thing.

The failed payment recovery calculator shows how much the recovery rate matters. Holding the other defaults fixed, the table varies only the share of failed renewals you win back.

Recovery rateRecovered each monthLost each monthLost each year
20%$500$2,000$24,000
40%$1,000$1,500$18,000
55%$1,375$1,125$13,500
70%$1,750$750$9,000
85%$2,125$375$4,500

Use the failed payment recovery calculator with your own recurring revenue and failure rate to replace the defaults. The table shows the shape of the relationship, not your result.

Stage 6: after the first renewal

  1. Ask people who cancel in the trial why, with one question and a short list of answers, and read every response for a month.
  2. Compare the first-month retention of trial customers against customers who bought without a trial.
  3. Look at refunds in the first month. A high refund count points to a charge notice that was not clear.
  4. Feed what you learn back to Stage 0.

What to measure, in one table

StageNumber to watchWhere it comes from
0 Before launchPaying customers per thousand visitorsVisits, trials and payments by start week
1 First minutesTrials with a session in the first dayProduct events
2 First daysTime to first useful resultProduct events
3 MiddleReturn sessions per trialProduct events
4 Before the chargeTrials with a usable card at five days outBilling records
5 ChargeFailed charge rate, recovery rate, hard decline shareBilling records
6 AfterFirst-month retention and refundsBilling records

A monthly review

Once a month, put the seven numbers on one page by start week, read them from left to right and pick the earliest stage that moved the wrong way. Fix that one first. A rising failed charge rate while return sessions fall is not a billing problem; it is the same trial losing interest earlier, and the charge is where you notice it.

Which stage should I fix first?

The earliest one with a clear problem. Later stages only see the people who got through the earlier ones.

Do I need all seven stages?

No. Start at Stage 0 and work down. Stop at the first thing that is plainly wrong and fix it.

How many numbers should I track?

One per stage, so seven in all. More than that and nobody reads them.

Does this apply to trials with no card?

Mostly. Stages 0 to 3 apply as written. Stages 4 and 5 become a request to add a card before the end, and the end of trial setting decides what happens without one.

Where do I get the retry settings?

In your Stripe revenue recovery settings, where you pick Smart Retries or a custom schedule and what happens when recovery fails.

Sources