The Retention Audit: 12 Checks Before You Buy More Traffic

Every merchant who calls us about scaling ad spend gets the same question first: what's your 90-day repeat purchase rate? Most don't know, off the top of their head. That's the tell. You cannot responsibly turn up the volume on acquisition if you don't know what happens to the customers you already paid for.

This matters more now than it did three years ago. When AI assistants mediate discovery — ChatGPT, Claude, Gemini, Perplexity, Google AI Overviews off-site, Rufus on Amazon, and Shopify's Brand Agents and Copilot Checkout on-site since the Spring '26 edition shipped June 17 — the first purchase gets shortlisted on comparable, machine-readable evidence: price, attributes, reviews, policies. Story-driven markup is harder to defend at that exact moment. First-order margin compresses. Retention has to do more of the work it always should have been doing.

Why retention has to come before acquisition math

The old logic was simple: if a dollar of ad spend returns more than a dollar of first-order profit, spend it. That logic quietly assumed first-order profit was stable. It isn't anymore. When an assistant shortlists your product against two or three comparable options on structured data alone, the presentation-driven premium that used to protect margin erodes. You still win orders — but the first one is worth less than it used to be.

That doesn't make acquisition unprofitable. It makes it dependent on what happens after the sale. If a customer orders once and never returns, you're financing growth on a shrinking first-order margin with no second act. If a customer orders three times over a year, the same acquisition spend suddenly pencils out fine — because the math that matters is LTV:CAC, not first-order ROAS.

That's the case for auditing retention before you touch your ad budget. Twelve checks, five areas. None of them require new software — most of them require pulling a number you already have and looking at it honestly.

Area 1 — Repeat purchase rate (checks 1-3)

Repeat purchase rate is the single number that tells you whether your business compounds or resets every month. It's also the number most merchants haven't looked at recently, because most platform dashboards bury it under acquisition-flattering metrics like total orders or revenue growth.

1. 1

   Check 1 — Pull 60-day and 90-day repeat purchase rate

   Of customers who ordered in a given month, what percentage ordered again within 60 and within 90 days? Segment by acquisition channel if you can — agent-referred, paid social, organic, email. A blended number hides which channels are actually bringing back buyers.

2. 2

   Check 2 — Compare against your own trailing baseline

   You don't need an industry benchmark to start; you need your own trend. Pull the same 60/90-day repeat rate for the trailing four quarters. Flat or declining repeat rate while acquisition spend rises is the clearest early warning sign in DTC — it means you're buying volume, not growth.

3. 3

   Check 3 — Segment repeat rate by first product purchased

   Some SKUs create repeat customers; others are one-and-done by nature (a gift, a single durable good). If your acquisition spend skews toward low-repeat products, your blended retention number will always look worse than your business actually is — or worse, you'll optimize acquisition toward products that structurally can't retain.

Area 2 — Email and SMS list health (checks 4-6)

List size is the vanity metric; list health is the number that predicts revenue. A 200,000-subscriber list with a 12% open rate is worth less than a 20,000-subscriber list at 40% — and it's actively hurting you, because a disengaged list drags deliverability down for every message you send, including the ones your best customers actually want.

1. 1

   Check 4 — Engaged-subscriber percentage, not raw list size

   Define "engaged" as opened or clicked in the last 90 days, then calculate what share of your list qualifies. Most merchants who haven't pruned in over a year find the number is uncomfortably low.

2. 2

   Check 5 — Opt-in quality: how addresses actually entered the list

   Trace your last 1,000 signups back to source. Addresses earned at a moment of value — order tracking, a sizing guide, a restock alert — convert and engage far better than addresses surrendered to close a popup. If most of your growth comes from popup pressure, expect the engagement numbers in Check 4 to be weak.

3. 3

   Check 6 — Deliverability trend, not just send-time metrics

   Check inbox placement and spam-complaint rate over the last two quarters, not just the open rate on your last campaign. A slow deliverability decline is often the first visible symptom of an unpruned, low-engagement list — and it silently taxes every flow you're about to audit next.

Engaged subscriber

A subscriber who opened or clicked at least one email or SMS message within a defined recent window (commonly 90 days) — as opposed to a subscriber who is merely present on the list but functionally unreachable.

Area 3 — Lifecycle flow coverage (checks 7-9)

Flows are the automated messages that do retention work without a human sending anything manually. Most merchants have some flows live and assume that's coverage. The audit question isn't "do flows exist" — it's whether the handful that actually move repeat purchase rate are live, current, and converting.

1. 1

   Check 7 — Confirm which of the five core flows are actually live

   Not drafted, not "in the platform somewhere" — live and triggering. Send yourself a test signup and a test abandoned cart today and see what actually arrives.

2. 2

   Check 8 — Pull conversion rate per flow, not just open rate

   A flow with a great open rate and a poor conversion rate usually has a weak or buried call to action. This is where most quick wins live, because fixing the offer or the CTA is cheap compared to building a new flow from scratch.

3. 3

   Check 9 — Check flow content against summarization, not just against a human reader

   Inbox assistants increasingly summarize marketing email before a person reads it. If a flow's message only lands with the hero image loaded and a full scroll, an assistant compresses it to "a promotion from a brand" and the substance is lost. State the one thing you want the reader to know in the first line.

If the welcome series specifically is where you're seeing the weakest numbers, we cover the rebuild in detail in the welcome flow rewrite for the agentic era. For flow architecture across the full customer lifecycle, see lifecycle email for agentic commerce.

Area 4 — Post-purchase experience (checks 10-11)

Post-purchase used to be a cost center: fulfillment, support, the occasional refund. It's now also an evidence-generation system, because the residue of what happens after checkout — reviews, ratings, dispute rates, delivery consistency — is exactly what AI assistants read before recommending you to the next shopper. A weak post-purchase experience doesn't just lose the customer in front of you; it quietly damages the evidence that wins future shortlists.

1. 1

   Check 10 — Delivery-promise accuracy

   Pull the percentage of orders that arrived within the window you promised at checkout over the last quarter. A gap here shows up later as reviews and complaints — evidence that's public and machine-readable, not just a private customer-service problem.

2. 2

   Check 11 — Review-request timing and support resolution rate

   Confirm review requests fire after delivery confirmation, when the product has actually been used — not immediately at purchase, when there's nothing to review yet. Separately, pull your support ticket resolution rate; unresolved complaints become the negative reviews and public text that suppress future recommendations.

On Shopify Plus stores, this window is also where a Brand Agent can keep working after the sale — answering order and setup questions without a support ticket. We cover that pattern in post-purchase Brand Agents.

Area 5 — Subscription and replenishment fit (check 12)

Not every catalog should have a subscription option, and forcing one onto a low-repeat, high-consideration product usually just adds churn to your dashboard. But for consumable, replenishable, or predictable-cadence products, the absence of a subscription or reorder mechanism is a retention leak hiding in plain sight — you're relying on the customer to remember to come back instead of building that behavior into the product.

1. 1

   Check 12 — Test whether your catalog has replenishment-fit products without a reorder mechanism

   List your top 20 SKUs by repeat-purchase frequency. For any product reordered by more than a small share of customers within a predictable window, check whether a subscription, auto-reorder, or at minimum a timed replenishment reminder exists. If the product predicts its own reorder timing and you have no mechanism capturing that, you're leaving a retention lever unpulled.

Reading your results: the 12-check scorecard

Score each check pass or fail, then look at the pattern rather than the total. A few specific fail clusters tell you where to start.

• Fails concentrated in Area 1 (repeat rate). You may not have a retention program at all yet — start with post-purchase and flow coverage before spending more on acquisition, since there's nothing yet to protect the spend.
• Fails concentrated in Area 2 (list health). You likely have retention infrastructure that's decayed. Prune the list, fix opt-in quality, and rebuild deliverability before launching new campaigns into it.
• Fails concentrated in Area 3 (flows). The fastest fix on this list — flows are usually a content and logic problem, not a build-from-scratch problem, and conversion-rate gains here show up within weeks.
• Fails concentrated in Area 4 (post-purchase). This is the slowest to fix and the most consequential for agentic visibility, since the evidence it produces compounds or decays over months, not days.
• Fails concentrated in Area 5 (subscription fit). Lowest urgency unless your catalog is genuinely consumable — don't force a mechanism your products don't support.

What to do with a failing audit

A failing audit is not a reason to stop growing. It's a reason to sequence growth correctly: fix the leak before you turn up the pressure. In practice that means treating retention fixes as the higher-priority spend for the next quarter, even when the acquisition team has a ready-to-launch campaign sitting on the shelf. The campaign will perform better against a retention system that's actually catching customers than one that isn't.

If the gaps concentrate in catalog or agent-visibility issues rather than retention mechanics — shoppers aren't finding you at all, agent or otherwise — that's a different diagnostic, and the Agentic Commerce Readiness Score is the faster starting point. But if the audit above turns up real retention gaps, fix those first. More traffic into a broken retention system is the most expensive way to learn that your retention system was broken.