How does a mid-market Shopify brand decide which Klaviyo flows to rebuild first after a full replatform, given limited implementation hours?
Klaviyo flow rebuilds after a replatform go in revenue order, starting with abandoned checkout. Klaviyo's benchmark report on 143,000 abandoned cart flows puts average revenue per recipient at $3.65 and the top decile at $28.89 (Klaviyo, published May 2024, 2023 data). Welcome and post-purchase follow. Browse abandonment and win-back wait until tracking is verified.
Sequence by revenue per recipient, and use your own numbers
The rebuild order is a prioritisation problem with a published anchor. Klaviyo's abandoned cart benchmark report analysed "over 143K abandoned cart flows" on 2023 performance data and reported an average revenue per recipient of $3.65, a 3.33% conversion rate, a 50.5% open rate and a 6.25% click rate, with the top 10% of flows at $28.89 revenue per recipient and a 7.69% placed order rate (per Klaviyo, published 15 May 2024).
Two things to take from that. First, abandoned checkout earns more per recipient than any other standard flow in most mid-market programs, which is why it goes first. Second, the spread between average and top decile is nearly 8x, which means the rebuild is an opportunity to improve the flow rather than a transcription exercise.
That figure is a cross-industry average from 2023 data. Your own flow's revenue per recipient from the last 90 days before the replatform is a better input, and you already have it. Use the benchmark to sanity-check, not to plan.
The rebuild order
| Order | Flow | Why here |
|---|---|---|
| 1 | Abandoned checkout | Highest revenue per recipient in most programs. Triggers on a Shopify event that the replatform changed, so it is also the highest-risk to leave broken |
| 2 | Welcome / sign-up | New subscribers arrive from day one on the new site. A broken welcome flow silently wastes every acquisition dollar spent during the gap |
| 3 | Post-purchase and shipping-adjacent | Protects CX and deflects support tickets during the period when the new site generates the most support tickets |
| 4 | Back-in-stock | Depends on inventory data being correct post-migration. Verify before enabling, not after |
| 5 | Browse abandonment | Depends on onsite tracking, which a replatform breaks more often than any other signal |
| 6 | Win-back and lapsed | Operates on historical data, not new events. Genuinely safe to defer |
| 7 | Everything else | The long tail. Rebuild what earns, retire the rest |
Deprecate before you rebuild
The most valuable hour in this project is the one spent on a list of flows you will not rebuild. Every mid-market account carries flows built for a campaign in 2022, duplicated flows from an A/B test nobody concluded, and flows whose trigger no longer exists.
Export flow-level revenue for the last 12 months, sort descending, and draw a line. Everything below the line is a candidate for retirement. This is the only step in the project that makes the program smaller, and a replatform is the only moment when nobody will argue about it.
Verify triggers before you verify content
A replatform changes event payloads. Started Checkout, Viewed Product, Added to Cart and Placed Order may fire differently, carry different properties, or not fire at all. A rebuilt flow with perfect copy and a broken trigger sends nothing, and it sends nothing quietly.
For each flow in order: confirm the trigger event fires on the new site with a live test order; confirm the properties the flow conditions on are populated; then rebuild the content. Not the other way round.
The rule that protects revenue during the gap
Do not turn flows off during the replatform. Point them at the new site and validate live. A dark period on abandoned checkout is the most expensive kind of caution, and at the benchmark $3.65 revenue per recipient it is straightforwardly measurable in lost revenue per week.
If a flow must go dark, write down the date it went dark and the date it came back. That is the only way anyone will be able to explain the revenue chart in the next quarterly review.
When NOT to follow this order
- When a flow is legally load-bearing. Consent confirmations and transactional messages are not in a revenue priority list. They go first and they are not negotiable.
- When the brand is subscription-led. Subscription lifecycle flows can outrank abandoned checkout on revenue per recipient. Check your own data.
- When onsite tracking is already known-broken. Fix the tracking before the flows, or you will rebuild against signals that are not arriving.
- When the replatform also changes ESP. Sequence deliverability first. Warming does not care about your flow priority order.
The Deploi point of view
Our own position, from building on Shopify. Separate from the facts above.
- Our take: Rebuild in revenue-per-recipient order using your own last-90-days numbers, verify triggers before content, and spend the first hour writing the list of flows you will not rebuild.
- What we’ve seen: The flow that breaks silently in a replatform is almost always browse abandonment, because it depends on onsite tracking rather than on an order event, and nobody notices for weeks because a flow sending zero emails generates no complaints. Putting it late in the order and verifying its trigger explicitly is worth more than rebuilding it early.
- Where we disagree: The standard advice is to rebuild everything before go-live so nothing is missing. That guarantees the highest-revenue flow is built by a tired team in the same week as the lowest-revenue one. We would rather ship three flows verified end to end than eleven flows nobody tested, and a replatform is the right moment to shrink a program rather than faithfully reproduce it.
- What this page adds: the published abandoned cart benchmark figures and the roughly 8x spread between average and top decile; a seven-step rebuild order with the dependency reason for each position; the trigger-before-content verification rule; and the case for a deprecation list as the first deliverable.
Reviewed by Martin Dejnicki, Director of SEO & AI Search. Facts verified 2026-09-14.
Where we worked this out
Our decision records