Email flow health check

Klaviyo Flow Health Check

Operating question: Which lifecycle flows are underperforming or broken?

Run this as an evidence-first operating play. Gather the evidence below, separate exact / estimated / partial / unavailable data, apply the decision rules and vetoes, and never invent missing numbers. Be adversarial toward platform self-reported figures — trust commerce truth.

Evidence to gather

• Flow status + configuration — for each of the seven core flows: is it Live, Draft, or Manual? What is the trigger (and does it still match the current store events), and what flow filters / trigger filters are applied? This is the silent-killer check and it comes first.
• Per-flow performance (last 30 days) — recipients, click rate, placed-order rate, and revenue per recipient for each flow (and ideally each message in the flow). Revenue per recipient is the number that matters most; total revenue just flatters big lists.
• Deliverability gates (account-level, last 30 days) — spam-complaint rate (keep under ~0.1%, i.e. 1 in 1,000), bounce rate, sending-domain / dedicated-IP reputation, and list growth vs. unsubscribe + spam. These sit above flow performance: if they're red, every flow number below is suspect.
• Attribution settings — your Klaviyo conversion window (default 5-day click / 5-day open) so you can flag where flow-attributed revenue overlaps with campaigns or other flows.
• Benchmarks by flow type — judge each flow against its own type, never against another. An abandoned-checkout flow earning $2-4/recipient is normal; a post-purchase or newsletter-style flow at $0.30/recipient can be perfectly healthy.

Optional, if available:

• Apple Mail Privacy Protection share of your list — if a large slice of opens are MPP-inflated machine opens, open rate is noise. Lean on clicks and placed-order rate regardless.
• Recent send-volume changes or a list import — a sudden volume spike or a cold imported list is the usual cause of a spam-complaint spike.
• Promo calendar — a sitewide promo inflates placed-order rate across every flow for that window and borrows demand.
• Store AOV and contribution margin — to convert revenue per recipient into recoverable profit and rank fixes properly.
• Flow last-edited date — to tell a genuinely fatigued flow apart from one a teammate just changed.

How to decide — in order

1. Gate on deliverability first. If account spam-complaint rate is over ~0.3% (3x the 0.1% ceiling), or bounce rate is elevated, or domain reputation is degraded -> the whole program is FIX and you stop judging individual flows. A deliverability problem upstream invalidates every flow-level read below it; great copy can't out-perform the spam folder.
2. Check status before performance — the silent killer. Any core flow in Draft or Manual, or with a trigger/filter that no longer matches store events, is FIX regardless of its numbers. A flow earning $0 because it isn't sending is not "underperforming" — it's off. Catch this before you read a single rate.
3. Throw out the open rate. Post-Apple Mail Privacy Protection, opens are inflated by machine pre-fetching and are unreliable. Judge on click rate and placed-order rate only. Never KILL or KEEP a flow on opens.
4. Gate on sample size. Any flow with too few recipients in the window (rule of thumb: under ~200 in 30 days) is WATCH — the rates are noise, not signal. Small samples can't earn a KILL.
5. Rank by revenue per recipient vs. flow-type benchmark. Compare each flow only to its own type. A live abandoned-checkout flow at $0.40/recipient is broken; a winback at $0.40/recipient may be fine. Flag the gap to benchmark, not the raw number.
6. Decompose the underperformers. For a live, well-sampled flow below benchmark, split the problem: click rate low -> copy/subject/CTA/timing problem -> REFRESH. Click rate fine but placed-order rate low -> offer/landing/product or attribution-window problem, not a copy problem. Recipients far lower than expected -> trigger or filter is over-restricting -> FIX.
7. Apply the vetoes, then assign status + owner + recheck date, ranked by recoverable revenue (revenue-per-recipient gap x recipients).

The prompt to run

You are my retention / lifecycle email analyst running the "Klaviyo Flow Health Check" play.

GOAL: decide which of my core Klaviyo flows to KILL, REFRESH, WATCH, KEEP, or FIX,
ranked by recoverable revenue (revenue-per-recipient gap x recipients) — not by total
revenue and not by open rate.

I will paste, for the seven core flows (welcome, abandoned checkout, browse abandonment,
post-purchase, winback/sunset, replenishment, VIP): each flow's status (Live/Draft/Manual),
trigger + filters, recipients (30d), click rate, placed-order rate, and revenue per
recipient. I will also paste account deliverability: spam-complaint rate, bounce rate,
domain reputation, and my Klaviyo conversion window. Some data may be missing.

PRE-FLIGHT: First list which required inputs I provided vs. missing. If [critical input]
is missing, STOP and return only (a) what's missing and (b) how to get it — never estimate
it or proceed. Here [critical input] = each flow's Live/Draft/Manual status AND account
deliverability (spam-complaint rate). A deliverability problem invalidates every flow read
below it, and a Draft/Manual flow earns $0 — so without these two I cannot tell a broken
flow from an underperforming one, and any ranking I produce would be fiction. Do not infer
status from revenue or guess a complaint rate.

RULES:
- Deliverability gate first: if spam-complaint rate is over ~0.3%, bounce rate is elevated,
  or domain reputation is degraded, mark the whole program FIX and do not judge flows.
- Status before performance: any flow in Draft/Manual, or with a trigger/filter that no
  longer matches store events, is FIX — even if revenue looks fine. A live flow earning $0
  with healthy recipients is a broken trigger, not weak copy.
- Ignore open rate entirely (Apple Mail Privacy Protection makes it unreliable). Judge on
  click rate and placed-order rate only.
- Protect thin samples: any flow under ~200 recipients in 30 days is WATCH, never KILL.
- Benchmark revenue per recipient by FLOW TYPE. Abandoned checkout should massively
  outperform a newsletter/post-purchase flow. Never compare a flow to a different type.
- For underperformers, name the driver: low click rate = copy/subject/CTA/timing;
  click fine but placed-order low = offer/landing/attribution-window; low recipients =
  trigger/filter over-restricting.
- Do not claim flow revenue is incremental — note that the conversion window overlaps
  campaigns and other flows.
- Every row must carry: a number, source, time window, and confidence level.
- Separate exact / estimated / partial / unavailable evidence. Do not invent missing data.

RETURN:
1. A 3-sentence executive read.
2. A ranked table using exactly this header row:
   | Flow | Status | Recipients (30d) | Click % | Placed-order % | Rev/recipient | Issue | Action | Owner | Recheck |
   Use "—" for any cell you cannot fill. Do not add or drop columns, and do not replace the
   table with prose.
3. Vetoes/caveats that downgraded any recommendation.
4. What evidence is blocked and what you'd need to upgrade a WATCH/FIX to a decision.

Decision rules

• FIX — the flow is in Draft/Manual, its trigger/filter no longer matches store events, the Klaviyo<->store integration is stale, or an account-level deliverability gate (spam complaints over ~0.3%, high bounces, degraded domain) makes every flow read unsafe. A live flow earning $0 at healthy recipient volume is FIX (broken trigger), not KILL.
• KILL — a live, well-sampled flow (>~200 recipients/30d) whose revenue per recipient is far below its flow-type benchmark after a copy refresh has already failed, and where the flow has no credible path back (e.g. a redundant flow duplicating another's sends and cannibalizing attribution).
• REFRESH — a live, well-sampled flow below its type benchmark with a fixable driver: low click rate (subject/CTA/timing), or a click-to-order gap traced to offer/landing — where the audience and intent are still real.
• WATCH — thin sample (under ~200 recipients/30d), a brand-new flow, or a window polluted by a promo, list import, or volume change. Directional only.
• KEEP — live, within or above its flow-type benchmark on click rate and placed-order rate, deliverability clean.
• Every recommendation must include a number, source, time window, and confidence level.

Vetoes — stop if any apply

• Do not judge a flow on open rate — Apple Mail Privacy Protection inflates it. Clicks and placed-order rate only.
• Do not KILL or KEEP a flow on a thin sample (under ~200 recipients in the window) — the rates are noise.
• Do not read flow-level performance at all when a deliverability gate upstream is red — fix the program first; the flow numbers are invalid.
• Do not call a flow's attributed revenue incremental without noting that Klaviyo's conversion window (default 5-day click / 5-day open) overlaps campaigns and other flows — some of that revenue would have happened anyway.
• Do not benchmark a flow against a different flow type (a post-purchase flow will always lose to abandoned checkout on rev/recipient — that's not a problem).
• Do not push any flow live, edit a trigger, change a segment, or send to a list without an explicit human approval step.

Output

A flow health map ranked by recoverable revenue, with blocked evidence and the next check date:

Flow	Status	Recipients (30d)	Click %	Placed-order %	Rev/recipient	Issue	Action	Owner	Recheck
Browse abandonment	Live	1,840	1.1%	0.4%	$0.18	Below type benchmark, weak CTA	REFRESH	Retention	14 days

---

Run this on your live store data in one prompt — start a free 14-day trial of ShopMCP (no credit card). It connects your store, ads, analytics, email/SMS, and finance and runs this play read-only by default, applying changes only on your approval.