Convert Lane

Chapter 8 of 11

CRO After the Sale

Half of conversion work happens after the order: email capture that does not wreck the first visit, post-purchase real estate, and why LTV is the real rate.

Contents
  1. Email capture without wrecking the first visit
  2. The most wasted screen in ecommerce
  3. The welcome flow is a conversion surface
  4. Post-purchase upsells, done without wrecking anything
  5. Replenishment and win-back timing
  6. Review generation is conversion infrastructure
  7. Subscriptions, honestly
  8. Loyalty programmes, honestly
  9. LTV is the real conversion rate
  10. Measuring retention without fooling yourself
  11. Where the course ends and the work begins

Conversion rate optimization stops, in most people’s heads, at the order confirmation.

That is where the most valuable half of it starts. A customer who has already paid you is the cheapest revenue available, the most likely to convert again, and the least optimized surface on nearly every store we look at.

This chapter is about the parts of conversion work that happen at or after the moment of purchase, plus one idea that reframes everything in the previous seven chapters.

Email capture without wrecking the first visit

Email capture is a conversion problem that competes with your other conversion problem, and most stores resolve the conflict badly.

The default implementation: a modal at three seconds offering 10% off. It captures emails. It also interrupts every first-time visitor at the exact moment they are forming an impression, discounts a customer who may have paid full price, and trains people to wait for a popup before buying anything.

Better, in rough order of how much we like them:

Capture at the point of a real need. Back-in-stock alerts on sold-out variants. Delivery notifications. Size guidance sent to your inbox. These convert at a fraction of a blanket popup and produce a list that actually opens email, because they signed up for a thing rather than for a coupon.

Capture in the footer and the post-purchase flow, always. Free, permanent, no interruption.

Delay the popup and condition it. If you run one, fire it on exit intent or after meaningful engagement, not at three seconds. Never fire it on a visitor who already has items in cart, and never fire it twice.

Offer something that is not a discount. A sizing guide, a comparison, the first chapter of something useful. A discount popup selects for discount buyers; that is a real cost you pay later in margin.

We practise this on this site, which is worth saying plainly because it is easy to write advice you do not follow. Our own capture is one form in the footer, one inline after the article, and one exit-intent that fires once a month. No coupon, because we do not have one to give. If you want to see the full argument for a specific pitch rather than a generic one, the newsletter page is the version we actually run.

The most wasted screen in ecommerce

The order confirmation page. Every single customer sees it, at the moment of peak goodwill, and on most stores it says “Thank you for your order” and nothing else.

Things that belong there, roughly in order of value:

A two-question survey. “What almost stopped you from buying?” and “What were you using before this?” Covered in Chapter 2, and this is where it lives. Response rates on the confirmation page are dramatically higher than anywhere else, because the customer is relieved and briefly generous.

Account creation, now rather than earlier. They have already given you the email and the address. Offering an account here, for order tracking, converts far better than forcing it at checkout, which Chapter 6 argues you should never do.

A referral ask. The moment after purchase is the highest-intent moment for a share. Not a full programme necessarily; a link and a reason is enough.

A genuinely relevant post-purchase offer. One-click, no re-entering payment, complementary rather than a rerun of what they just bought. Someone who bought a coffee grinder wants filters, not a second grinder.

What does not belong there: your full navigation, a newsletter popup, or a discount on the thing they just paid full price for.

The welcome flow is a conversion surface

Most stores treat the welcome email sequence as a marketing asset. It is a conversion asset, and it is measurable like one.

A first-time subscriber who has not bought is a person who did not convert on the site, and you now get a second attempt with more room than a product page allows. Everything from Chapter 5 applies here: answer the questions in order, handle objections from your review mining, set delivery expectations.

What consistently works:

  • Message one arrives immediately and delivers whatever was promised. Nothing else.
  • Message two is the mechanism. Why the product works. This is the section that does not fit on the product page.
  • Message three is proof. Specific customer stories, not a star average.
  • Message four handles the top objection from your research, directly and by name.
  • Discount, if any, arrives last, not first. Leading with a discount to a new subscriber sets your price permanently in their head.

Measure it as a funnel, not as opens. Subscriber to first order, and days to first order. Those two numbers tell you whether the sequence is doing conversion work or just sending email.

Post-purchase upsells, done without wrecking anything

The one-click post-purchase offer is the rare mechanism that is almost pure upside, because it happens after the order is banked. Nothing you do here can cost you the purchase you just made.

That said, it can cost you the customer.

Offer one thing, once. A chain of three offers before the confirmation page turns a good purchase experience into a gauntlet, and the customer remembers the gauntlet.

Make it genuinely complementary. Refills, the matching item, the thing that makes the first thing work. Not a second version of what they bought, and not a random bestseller.

Do not discount below what they just paid. Offering the same product 20% cheaper thirty seconds after full price is how you teach someone never to buy at full price again, and it generates a support ticket.

Keep it one tap. If it re-opens checkout or asks for payment details again, the mechanism does not work and you should not run it.

Make declining obvious. A “no thanks” link in small grey text below a large button is a dark pattern, and it converts people into resentful customers who complete a purchase they did not want.

Replenishment and win-back timing

For consumables, the highest-value automation is arithmetic: work out when the customer runs out, and be there slightly before.

Calculate the actual cycle per product. Not a guess. Look at repeat purchase intervals in your own order data, take the median rather than the mean because a few outliers will skew it, and set the reminder at roughly 80% of that interval. A 30-day supply gets a reminder around day 24.

Segment by what they bought, not by a single store-wide timer. A store selling both a monthly consumable and an annual durable that reminds everyone at 45 days is wrong for both.

Win-back is a different message from replenishment. Replenishment assumes they still want it and are simply running low, so it is a convenience. Win-back addresses somebody who stopped, so it needs to ask why or offer something new. Sending a cheerful “time to reorder” to a customer who left because the product disappointed them is worse than sending nothing.

Stop eventually. A win-back sequence that runs indefinitely is how you end up mailing a list of people who have not opened anything in two years, which is how you damage the deliverability of every message you send to people who do care.

Review generation is conversion infrastructure

Reviews are the raw material for most of Chapter 5 and most of your research in Chapter 2. A store with 40 reviews cannot do this work. A store with 4,000 can.

So review volume is not a vanity metric, it is an input to everything else.

Timing is the whole game, and it is product-specific. Ask when the customer has experienced the outcome, not when the parcel arrived. A consumable might be 14 days. A garment, 10. A mattress, 45. Sending the request on delivery day gets you reviews about the packaging.

Ask for the specific thing you need. A blank “write a review” box produces “Love it, great product,” which is useless as evidence. Prompted questions produce usable material: “How did the fit compare to what you expected?” That is a review that answers a future customer’s actual objection.

Make photo submission trivial, because customer photos outperform studio images in the review block, as covered in Chapter 5.

Do not filter negatives. A product page with only five-star reviews reads as fake, and it removes the exact signal a buyer is scanning for. A 4.4 with visible three-star reviews explaining a real limitation converts better than a suspicious 5.0.

Subscriptions, honestly

If you sell a consumable, subscription is the largest lever in this chapter, and the most abused.

The abuse pattern: pre-select subscription, show the subscription price as the headline price, make one-time purchase a small radio button below, and bury cancellation. It lifts subscription rate immediately. It also produces refunds, chargebacks, support load, and people who will never buy from you again.

The honest version tends to win on the numbers that matter. Show both prices at equal weight. Say the cadence and the saving in plain words. Make cancelling obviously possible, and say so on the product page, because the fear of being trapped is the actual objection.

Switching from a pre-selected subscription to an honest side-by-side usually pushes the subscription rate down and ninety-day retention up. That is the trade, and it only shows up if you measure past the first order, which is the entire argument of this chapter.

Which is the point of this whole chapter.

Loyalty programmes, honestly

Points programmes are widely sold and rarely evaluated. The evaluation question is simple and almost nobody asks it: are you rewarding behaviour you would not otherwise have got?

A programme that gives points to customers who were going to reorder anyway is a discount on your best customers, dressed as retention. It shows up as a successful programme because members have higher lifetime value than non-members, which is a selection effect, not a result. Your most loyal customers join loyalty programmes.

To know whether it works, compare enrolled customers against a matched holdout who were eligible and did not enrol, at similar prior purchase behaviour. If the gap is small, you are paying for something you already had.

Where they genuinely work: categories with low switching costs and frequent purchase, where the programme changes which store gets the next order. Where they do not: considered, infrequent purchases, where nobody remembers they have 340 points.

LTV is the real conversion rate

Here is the reframe.

Every metric in the first seven chapters measures one session. Conversion rate, add-to-cart rate, checkout completion. All single-visit numbers.

The business does not run on single visits. It runs on what a customer is worth over their life, against what they cost to acquire. Every optimization decision should eventually be checked against that ratio, and some decisions that look like wins on a single-visit metric are losses against it.

Concrete examples of the conflict:

  • A discount popup lifts first-order conversion and lowers margin on every subsequent order from that customer.
  • A pre-selected subscription lifts subscription rate and lowers ninety-day retention.
  • An aggressive cross-sell lifts average order value and increases return rate.
  • Softening a return policy lowers conversion rate slightly and raises lifetime value, because people buy again from stores that made returning easy.

None of those are visible in an A/B test that runs for three weeks and measures one session.

What to do about it practically: for any test that could plausibly affect repeat behaviour, tag the cohort and check them at 60 and 90 days. You will not get statistical significance on the long-term number, and that is fine. You are looking for the direction, and for the case where a shipped winner turns out to have cost you money.

This is also the honest answer to why conversion rate is a diagnostic rather than a goal, which is where Chapter 1 started. The loop closes there.

Measuring retention without fooling yourself

Retention metrics are unusually easy to report in a flattering way, mostly by accident.

Use cohorts, not averages. “Our repeat purchase rate is 34%” is not a number, because it mixes customers who bought last week with customers who bought three years ago. The honest version is a cohort table: of customers acquired in a given month, what share bought again within 30, 60, 90, and 365 days. That table shows whether the business is getting better or worse, which the average never will.

A growing store’s blended retention rate always looks like it is falling. If acquisition is accelerating, recent cohorts, who have had less time to buy again, make up a larger share of your customer base and drag the average down. The business can be improving on every cohort while the headline number declines. This trips up almost everyone the first time.

Watch time-to-second-order. It is the earliest reliable signal you have. A cohort whose median second order comes at 40 days instead of 55 is a cohort that will have higher lifetime value, and you will know months before the revenue confirms it.

Segment lifetime value by acquisition channel. This is where the discount-popup argument becomes concrete. Customers acquired through a discount frequently show lower lifetime value than customers acquired at full price, sometimes dramatically. If you only look at blended lifetime value, you cannot see it, and you will keep optimising the popup that is quietly degrading your customer base.

Be careful with “lifetime” value. Most stores do not have enough history to know a true lifetime, so use a fixed window: 12-month value per customer, by cohort, by channel. It is a number you can actually compute and compare rather than an extrapolation that flatters whichever assumption you picked.

Where the course ends and the work begins

Eleven chapters. The short version of all of it:

  1. Conversion rate is a diagnostic. Revenue per session is closer to the truth. Lifetime value is the truth.
  2. Research before ideas. Your reviews and support inbox already contain the answer.
  3. A hypothesis without evidence is a guess, and a guess that loses teaches you nothing.
  4. Most tests lose. Size them before you build them, and do not peek.
  5. The product page has one job: answer every question, in order, without leaving the page.
  6. Checkout is mostly a cost-surprise problem, and most of the fix happens before checkout.
  7. Paid traffic arrives holding a promise. Match it or pay for people who bounce.
  8. The order confirmation is the start of the next conversion, not the end of this one.

If you have read this far, you know more about this than most people who sell it. Go and run the research in Chapter 2 on your own store. It costs a fortnight and nothing else, and it will produce a problem list you did not have this morning.

And the commercial part, stated once as promised in Chapter 1: we do this for a living, for eight to ten ecommerce brands at a time. If you would rather it be done than do it, book a call. Twenty minutes, no deck, and if your traffic is too low for a testing programme we will say so on the call rather than sell you one.

Either way, the method above is the whole method. There is no advanced version behind a gate.

Ready to have the whole funnel run for you?

You have read the method. If you would rather it be done than do it, that is the entire arrangement. Free 20-minute call, no deck, and we will tell you straight whether we can help.

  • 437+ stores built
  • 2,000+ tests shipped
  • No call needed to start

Claim your free audit of one page

Send us one page of your store and we’ll review it on video: what’s leaking, what to fix first, and what to leave alone. Changes you can ship this week to lift CVR, AOV and revenue per visitor.

Get My Free Audit