Two cards on Subscriptions → Overview answer the questions a subscription discount raises: is this offer worth what it gives away, and why are subscribers leaving. Both follow the date range picked at the top of the page.

​
Offer test

An offer is one plan with one discount — the same plan at 10% off and at 15% off are two offers, and a plan sold at full price shows as No discount. For every offer that customers subscribed to in the selected period, the card shows:
  • Cancel in 72h — the share of those subscribers who cancelled within 72 hours of signing up. Only subscriptions at least 72 hours old are counted, so a sign-up from this morning never counts as “kept”. A discount that buys a sign-up followed by an immediate cancel is a coupon, not a subscription.
  • 90-day value — what a subscriber on this offer paid in their first 90 days: the first order plus every paid renewal inside those 90 days, averaged per subscriber. Only subscribers who signed up at least 90 days ago are counted, because the answer is not known before then.
  • One-time buyer, 90 days — the same figure for customers who bought without subscribing: their first one-time order plus any repeat one-time orders within 90 days. It is worked out per currency and shown once at the top when every offer shares it.

​
Verdicts

Each offer gets one verdict. Once the offer has a 90-day value it is compared with one-time buyers; until then the card falls back to the early 72-hour signal, compared with the store’s other offers over the same period.
  • Nothing is reported from fewer than 20 customers. A cancel rate, a 90-day value, a one-time baseline and a verdict each need at least 20 subscribers (or one-time buyers) behind them. Below that the figure is left out rather than shown from a handful of orders.
  • A store with a single offer gets its 72-hour rate but no early verdict — there is nothing to compare it with. The 90-day verdict still arrives once the cohort is old enough.
  • Offers are compared in their own currency. A subscriber paying in euros is compared with one-time buyers who paid in euros.
Use the early verdict to catch a discount that attracts cancel-right-away sign-ups within days, and the 90-day verdict to decide whether to keep the discount at all. A Trails one-time offer is giving away more than the extra renewals bring back.

​
Early-warning alert

You don’t have to keep checking the card. Once a day PlatformDTC looks at every offer, and when one newly turns High early cancels your store’s staff get an email — “Too many people cancel this offer in the first 3 days” — and the same warning appears in the bell at the top of the dashboard for a week. You are told once per offer; it only comes back if the offer recovers and then slips again. Nothing on your store is changed.

​
Subscription health

Four tiles with a target each, so you can see at a glance whether subscriptions are working. A tile reads On target or Below target; a tile with too little data (fewer than 20 customers or orders) is not shown.

​
Profit per customer: subscribers vs one-time buyers

Revenue can flatter a discount: a subscriber pays more over time, but every renewal also carries product cost, shipping and payment fees. This card adds up profit — what the customer paid, minus product cost, shipping, handling, payment fees and refunds (tax excluded) — for every order of every customer, from their first order on.
  • First 90 days — the average profit a subscriber leaves you in their first 90 days, next to a one-time buyer. Only customers whose first order is at least 90 days old are counted.
  • The table — one row per month you won customers, split into subscribers and one-time buyers. Month 1 is profit by the end of their first 30 days, Month 2 by the end of 60 days, and so on to Month 7. A cell is left empty until at least 20 customers in that row have reached that month.
  • A customer counts as a subscriber if they ever started a subscription, including after a one-time first order.
  • Product costs come from the costs recorded on each order, or your default cost % where none was recorded; the card says how much of your sales had a recorded cost. With neither, the card is not shown — revenue is never presented as profit.

​
Setting up a plan

On Subscriptions → Plans:
  • New plans start at every 4 weeks. That is 13 charges a year instead of 12, and each renewal arrives sooner. One click switches to every month, every 2 months or every 3 months. Existing plans are not changed.
  • Every 3 months shows a reminder: billing every 3 months delays your cash — every 4 weeks gets you paid sooner.
  • A discount of 30% or more shows a warning: big discounts attract bargain hunters who cancel and dispute more — a higher one-time price usually works better than a deeper subscriber discount.
  • When a plan is for one product, the editor shows the one-time price next to what subscribers pay each delivery.
These are suggestions only — you can save any plan you like.

​
Why subscribers left: the trend

The Why subscribers left card splits cancellations into four causes — Customer decided, Payment failed, Fraud & disputes and Order cancelled — with a table of the exact reasons underneath. Click any cause to open its trend:
  • Bars — that cause’s cancellations per day, week or month (following the length of the range), bucketed by the date the subscription was cancelled and stacked by reason: the four most common reasons, then Other reasons.
  • Share of cancels — the line shows that cause as a share of all cancellations in the same bucket. It catches a shift in the mix that a raw count hides when overall volume moves.
  • A gap in the line is a bucket with fewer than 5 cancellations. A share of three cancels is noise, so the line skips it instead of drawing a 0% or 100% point.
A spike in one bucket (for example after a batch of renewal reminders went out) reads very differently from a steady climb — the trend tells you which one you have. Related: Failed subscription payments, Subscription renewals by payment method.