1. Getting started
Installation runs as a three-step guide. The app unlocks at the end of it, and the whole thing is designed to be finished in one sitting.
See it on a real order
The first screen renders your blocks against your store's most recent real order — real product titles, real prices, real customer name. Not a stock mockup.
This is deliberate: the fastest way to know whether an upsell belongs on your thank-you page is to see it sitting under an order you recognise.
Publish your blocks
Each surface has its own switch. Flipping one publishes the blocks on that surface to your live store.
The switch stays in the state your store is actually in until the server confirms the change — it does not flip optimistically. A spinner appears on the switch you touched, so you always know which one is working.
Choose a plan
Pick Starter or Unlimited to start a 7-day free trial, or continue on Free. If Encore can read your store's order volume, the tier that fits is pre-selected for you — and only for a store that has not already chosen one, so nobody's existing bill changes by itself.
Free is not in the radio list. It sits in its own block below, with its cap stated up front.
2. Surfaces and blocks
Encore runs on the two post-purchase surfaces every Shopify plan can still customise. Each block type belongs to exactly one of them.
| Surface | When the customer sees it | Blocks available |
|---|---|---|
| Thank you page | Immediately after checkout, once per order. | Recommendations, Discount code |
| Order status page | Every time they come back to check on the order. | NPS survey, Reorder |
The split follows intent. The thank-you page catches someone with a card still in hand, so it carries the selling blocks. The order status page is visited later, repeatedly, by someone thinking about the order itself — which is when a survey gets an honest answer and a reorder makes sense.
Blocks on a surface can be reordered, and a block can be a draft or live independently of the surface switch.
3. Block reference
What each block does, what it ships with, and what you can change.
Recommendations
Thank youShows products to buy next. Live from install.
- Number of products
- 3 by default. 1, 2 or 3.
- Source
- Static rules by default; switch to the AI picker on a paid plan.
- Discount offered
- 10% by default.
- Exclude what they bought
- On, and we recommend leaving it on.
Discount code
Thank youA code for the next order. Starts as a draft.
- Code source
- Encore mints a real discount in your admin when the block goes live — or point it at a code you already run.
- Discount
- 10% by default.
- Expiry
- 30 days by default.
NPS survey
Order statusAsks how it went. Live from install.
- Scale
- Standard NPS, 0–10.
- Open question
- On. A free-text follow-up after the score.
- Once per order
- On. A customer who comes back to check tracking is never asked twice.
Reorder
Order statusBuy it again, pre-filled. Starts as a draft.
- Number of products
- 2 by default.
- Basis
- The customer's past purchases, or your best sellers.
4. Running an A/B test
Open a live block and start a test. Branch A is the block as it is now; branch B is one change from a fixed list. Traffic splits at the ratio you set, and the branch is assigned per order by the server, so a customer never sees two versions of the same block.
What you can vary depends on the block:
| Block | What branch B can change |
|---|---|
| Recommendations | Let the AI pick products instead of rules · show 1 or 2 products instead of 3 |
| Reorder | Suggest best sellers instead of past purchases · show 1 or 2 products instead of 3 |
| NPS survey | Drop the open question |
| Discount code | Not testable — see below |
Why a discount block cannot be tested
Varying a discount block means varying the offer, and the two arms then cost you different amounts. A test whose branches have different costs is not measuring the change it claims to — the winner is whichever one you subsidised harder. Rather than let you run it and misread it, Encore does not offer the test at all, and the button is not there to click.
How a winner is decided
Conversion is defined per block type, because the block types do fundamentally different jobs:
- Selling blocks count an order. The signal arrives from Shopify's
orders/createwebhook, matched by the attribution the block's own cart link carried — not from anything the browser reports. - Survey blocks count an answer. Their base rate is a response rate, and it is never averaged together with a purchase rate.
Until the sample is large enough, the results screen says so and withholds a verdict. A lead that is really noise is the most expensive kind of number a dashboard can show you.
5. Reading the analytics
The performance screen answers one question — how much money Encore made you — and then shows you the cost side of it.
| Metric | What it means |
|---|---|
| Attributed revenue | Revenue from second orders that an Encore block led to. |
| Conversion rate | Share of orders with an impression where the customer did what the block asked. |
| AOV uplift | Extra value per order, against orders that saw a block. |
| Dismiss rate | How often people actively close a block. Rising here means it is in the way. |
| Bounce rate | How often people leave the page after seeing it. |
Encore does not print a return-on-cost ratio. On a small store that number says more about the size of the app fee than about the app, and the revenue figure is the one you can check against your own Shopify reports.
6. Plans and the cap
Every plan has every feature. What differs is how many orders showing a block are covered per month.
| Plan | Price | Cap | Trial |
|---|---|---|---|
| Free | $0 | 80 / month | — |
| Starter | $12 | 500 / month | 7 days |
| Unlimited | $29 | No cap | 7 days |
Reach the cap and your blocks pause until the next month. There is no overage charge and no degraded mode — your configuration and your history are untouched, and upgrading restores blocks immediately.
One more difference worth knowing: the AI product picker runs on paid plans only. On Free, recommendations always come from static rules.
7. Troubleshooting
Check three things, in this order: the surface switch is on; the block's status is live and not draft; and — for a discount block — a real discount code exists. A discount block with no code renders nothing on purpose. If all three are right, place a test order rather than refreshing an old thank-you page; that page is tied to a specific order.
Conversions arrive from the orders/create webhook, so a block only registers a sale once a customer has completed a second checkout through it. Impressions should appear straight away. If impressions are also zero, the surface is probably off or the blocks are still drafts.
Either the block is not live yet, or it is a discount block — those cannot be A/B tested, for the reason in section 4. Every other block type has at least one variant available.
They are probably coming from static rules. Check that the block's source is set to the AI picker and that you are on a paid plan. Encore also falls back to rules whenever the model is slow or not confident enough — that is normal, and it is what keeps the block from rendering empty.
That should be impossible while “exclude what they bought” is on — the guardrail runs before the ranker and removes the purchased products and their near-identical variants from the candidate list. If you have seen it happen with the setting on, please send us the order number; that is a bug we want.
Reinstall within about 48 hours and your blocks and history are still there. After that Shopify sends the shop redaction request and everything is deleted — see the Privacy Policy.
Did not find it here?
Send us the store domain and an order number and we can usually answer in one reply.