A UK electrolytes brand

01 · Where they started

Growth that was treading water

This brand had done the hard part: a single hero product people genuinely loved, and a fast-growing, subscription-first business with the subscription programme at its centre.

Underneath, the engine was leaking almost as fast as it filled. In the month before we started, the brand was losing nearly as many subscriptions as it gained. Acquisition spend was mostly backfilling churn, and net subscriber growth was heading for zero.

The product retained. The infrastructure around it had never been built to match.

02 · What we found

The exit interview nobody was reading

We started where the loss was happening: the cancellation flow itself. We pulled every cancellation session from the funnel's history and read what customers had been saying on their way out.

1. The cancel flow was a formality, not a funnel

   Three generic reasons, one treatment each, no awareness of how long someone had subscribed or what plan they were on. It was saving very few of the people who reached it.

2. The number-one cancel reason was not about the product

   The most common answer by far was too much product. The flow answered with a skip, and that treatment rarely worked. The real cause sat upstream: rigid delivery schedules that did not match how fast people actually used the product, so stock piled up until cancelling felt like the only option.

3. The exit was the only door with no defence around it

   Cancellations clustered before the first renewal. There were no billing reminders, so customers were charged, surprised, and cancelled reactively.

03 · What we built

Rebuilding the exit as a conversation

The rebuild treated the cancellation flow as the centre of a system, not a page to polish. Everything that follows shipped inside the first ten weeks.

• The first fortnight

  Quick wins while the strategy was written. Smart payment retries switched on and aligned to paydays. Pause enabled in the subscriber portal. Skip options changed from confusing multipliers to real timeframes. Subscriber segmentation converted from a stale static list to a live segment, so subscribers stopped receiving prospect offers that undercut their plan.

• The early weeks

  A reason architecture that listens. Three vague reasons became six precise ones, with sub-reasons underneath (taste alone split three ways) and a required free-text option, so every exit produces usable data. Each reason got a treatment menu matched to the actual problem: delivery-frequency changes, pauses and skips for surplus; a flavour swap plus sample pack for taste; a multi-order discount for people who had not yet felt a difference, built to extend the usage window rather than buy a single renewal. Discounts stayed in the toolkit, but as one matched treatment among several, capped, and never the default.

• Go-live

  The new cancellation flow went live. The save rate more than doubled within days, and kept climbing as the treatments were tuned.

• Alongside it

  The upstream fixes for the number-one cancel reason. Longer delivery options, so cadence finally matched real usage. A billing reminder flow with one-tap skip, delay and flavour-swap actions, so nobody gets charged by surprise. Two-way SMS through Skio, letting subscribers skip, pause or move a renewal by text the day before billing, while cancelling deliberately stays in the portal so every exit passes through the flow. And milestone rewards placed exactly at the drop-off cliffs, giving the cancel flow a loss-aversion lever: pausing preserves your streak and the gift waiting in your next order, cancelling resets them.

• The refinement loop

  Then the flow started teaching us. A splash-screen video now fronts the funnel before any reason is asked: a real person who mirrors the brand's customer, walking through what the programme includes and what cancelling gives up. Offers began branching on order count, so a first-cycle wobble and a tenured subscriber are treated differently. Payment recovery was extended and tiered. Each iteration moved the save rate again, to almost four times where it started, with most saves now coming from a skip, pause or plan change rather than a discount.

04 · The numbers

What it added up to

• ↗ 3.9x cancellation save rate
• ↘ -17.35% total churn rate, engagement period vs the period before
• ↘ -14.39% active churn rate, customer-initiated cancellations
• ↗ +81% failed-payment recovery rate
• ↗ +12% recurring revenue base growth in a single month