Integration guides
How to connect it — and what breaks.
Setup guides for the tools a Shopify brand doing $3M–$30M actually runs on. Every one of them ends the way the vendor documentation never does: with the three things that break in production, what causes each, and what to do about it before it costs you a billing cycle.
Search
Search index is built by npm run build — unavailable in dev.
Start with
Why every guide has a failure section.
An integration that works on the day you connect it is not the same thing as an integration that works. On a subscription brand nothing fails once — every renewal is an order, so a fault repeats on every billing cycle for every subscriber until somebody happens to notice. In our experience it fails in three ways, and every guide below is organised around them.
- Identity drift
- Two systems each think they own the customer record. A subscriber updates their email in the portal, an agent edits it in the helpdesk, checkout used a different address than the account — and now there are two profiles, each holding half the history, with consent sitting on one of them. Every flow you build downstream inherits the split.
- Duplicate events
- Turning a new integration on almost never turns the old sender off. Three systems can now send the same shipping confirmation, and every subscription renewal is also an order, so anything triggered on a first-purchase event re-fires on every billing cycle for every subscriber.
- Silent partial sync
- The worst class, because nothing errors. A webhook is dropped under load, a SKU has no counterpart on the other side, one location’s inventory stops updating. No alert fires, no red banner appears, and the gap only surfaces as a support ticket weeks later. Anything worth integrating is worth reconciling on a schedule.
None of the three announce themselves. That is the argument for reconciliation: a scheduled job comparing both sides and alerting on the difference is cheaper than any of the three, every time.
The stack these guides cover.
One kind of business, one stack. We write about the tools a Shopify or Shopify Plus brand doing $3M–$30M is actually running — a subscription platform, Klaviyo, a paid helpdesk, a 3PL, a marketplace feed — and about nothing else, which is why these guides can be specific.
- Subscriptions
-
- Recharge
- Skio
- Smartrr
- Stay AI
- Shopify Subscriptions
- Email & SMS
-
- Klaviyo
- Postscript
- Attentive
- Omnisend
- Support
-
- Gorgias
- Zendesk
- Re:amaze
- Fulfilment
-
- Shopify
- ShipBob
- 3PL portals
- Loop Returns
- Channels
-
- Amazon Seller Central
- TikTok Shop
- Google Merchant Center
- Automation & data
-
- n8n
- Zapier
- Make
- Google Sheets
- Triple Whale
- Finance
-
- QuickBooks
- Xero
- Shopify Payments
A pair missing from the list below is usually a pair we have built and not yet written up. Ask for it and we will tell you whether it is a documentation gap or a genuinely hard integration.
The guides.
Each one covers what the integration actually does, the setup step by step, the three things that break, what syncs and what doesn’t, and what we build when the native integration runs out of road.
-
Klaviyo Recharge
Set up the Recharge–Klaviyo integration properly, then fix the three things that break it: split profiles, duplicate sends, and events that never backfill.
Read the guide → -
Shopify ShipBob
Connect Shopify to ShipBob, then fix what actually breaks: bundle SKUs the warehouse cannot pick, inventory that reads high, and split fulfilment emails.
Read the guide →
Nothing published in that slice yet. Clear the filter to see everything.
When the native integration isn’t enough.
Native integrations are built for the average of every merchant on the platform. A brand doing $3M–$30M is not that average — the renewal spike, the bundle that has to explode into components before it reaches the warehouse, the pause-instead-of-cancel path, the decline-code branch and the marketplace feed that has to agree with Shopify are all outside it. Four things we build when the app runs out of road.
- A webhook receiver you own Self-hosted n8n takes the source system’s webhooks directly, with a queue, a retry policy and a dead-letter path, so a spike or a five-minute outage on the far end doesn’t become a permanently missing event.
- Business rules the app has no opinion about Decline-code-aware retry branches, kit explosion before an order reaches the warehouse, hold windows for subscribers who still have edit rights, batching for renewal days. The logic lives in one place instead of being approximated in four settings screens.
- Reconciliation and alerting A scheduled job that compares both sides — charges against events, warehouse on-hand against Shopify counts, profiles against customer records — and alerts on the difference. This is the single highest-return thing most brands are missing.
- The events the native integration doesn’t carry Retry attempt number, decline-code family, days since last successful charge, projected next-reorder date. Once those exist as properties, the flows you couldn’t build become ordinary.
All of it is built in your accounts, documented, and yours if we part ways — the same rule as everything else we do. See how ops automation works →
Before you start
Why does every guide have a section on what breaks?
Because the vendor documentation only describes the happy path, and the happy path is not where the money goes. On a subscription brand every renewal is an order, so an integration fault does not fail once — it repeats on every billing cycle until somebody notices. The failure modes are the part of these guides that took us years to learn and they are the part nobody else writes down.
Are these guides written from the vendor documentation?
The setup steps follow the vendors’ own flows, because there is only one way to paste an API key. Everything after that — the failure modes, what does not sync, the reconciliation you should be running — comes from live builds. Where a menu label or a screen has moved between versions, the guide says so instead of pretending it hasn’t.
What do you do when the native integration isn’t enough?
We build the middleware. Usually that is a self-hosted n8n instance that receives the webhooks directly, normalises them, applies the business rules the native integration has no opinion about, and posts clean events onward — plus a reconciliation job that compares both sides on a schedule and alerts on the difference. It lives in your accounts and it is yours.
Can you just fix ours?
Yes, and that is usually the fastest version of this. The Revenue Recovery Audit maps every integration in your stack, what each one is meant to sync, and where it has silently stopped — with a dollar figure next to each gap. Two weeks, fixed fee, credited against any build.
Related reading
Find out what you’re losing.
Before you commit to anything, we tell you exactly what you’re losing and what it costs to stop it. Two weeks. Fixed fee. Credited in full against any build you go ahead with.
- Fee
- $1,500–$3,000, fixed
- Duration
- Two weeks
- Credited
- In full, against any build
- You supply
- Read access + one 45-minute call