How to Improve Repeat-Purchase Rate

Repeat-purchase rate is the metric most stores report and least often move, usually because the lever they reach for is a discount. Last updated: October 2026.

Omniconvert has measured post-purchase behaviour across the CROBenchmark dataset of 7,000+ websites in 15+ industries, against 248+ audit criteria, over 13 years in eCommerce, and the stores with strong repeat rates are rarely the ones with the most generous returning-customer offers. They are the ones whose second order is easier to place than their first. Omniconvert Explore is the CRO platform behind the testing and survey work that tells you why a customer did not come back, and Explore averages a 23.2% conversion uplift across 70,000+ experiments. Start from a good conversion rate if you want the acquisition-side figure first.

Measure the second order, not the repeat rate

This is a measurement problem before it is a marketing problem, and getting it wrong wastes the rest of the effort.

Picture a store that doubles its advertising spend in March. In April its repeat-purchase rate falls, because the denominator just filled with people who have had no time to come back. Nothing about the store's retention changed. The opposite happens when spend is cut, and a team can take credit for a retention improvement that is entirely an acquisition artefact.

The cohort version removes this. Take everybody whose first order was in March, and ask what proportion placed a second order within sixty days. That figure is comparable with February's and with last March's, and it responds to things you changed rather than to how much traffic you bought.

Pick the window deliberately and then leave it alone. Sixty days is a reasonable default for most categories, ninety for considered purchases, thirty for consumables. The number matters less than the consistency, because the comparison is the point.

Finding your second-order window

This is the highest-value hour available in this whole topic.

Run the query: for customers with two or more orders, the days between the first and the second. Take the median rather than the mean, because a handful of customers who came back after a year will drag an average into uselessness.

What you get is a number with an obvious meaning. If the median is twenty-four days, every piece of useful contact needs to land before day twenty-four, and a sequence whose second email goes out on day thirty is arriving after the decision.

Look at the spread as well as the median. A tight cluster means a replenishment rhythm you can almost schedule. A wide spread means the second order is triggered by an event rather than by elapsed time, and the right response is a trigger rather than a timer: a reorder prompt when the customer returns to the site, not on day twenty-one.

Segment the window by first product where you can. A store whose entry product is a consumable and whose second product is a durable has two windows, and averaging them produces a schedule that fits neither.

Five levers that move repeat-purchase rate

The table sets out each lever with who it reaches, the effort involved and the usual mistake.

The last row is in the table as a warning. A loyalty programme is a reasonable thing to build once the first four are in place and a very expensive way to avoid building them.

Why the reorder route matters more than the campaign

If you only do one thing from this article, do this one, because it reaches every returning customer rather than a segment.

Consider what a returning customer currently faces. They remember the product but not its exact name. They search, land on a category page, find three similar variants and have to recall which one they bought. Then they re-enter whatever the browser did not save. For a routine repurchase of a known item, that is a surprising amount of work.

The replacement is a reorder action in two places: the account order history, and the delivery-confirmation email, which is the message with the highest open rate your store sends. One action, pre-filled with the exact variant.

Two implementation notes decide whether it works. It has to carry the variant, not the product, because sending somebody to a product page to choose again defeats the purpose. And it should confirm rather than complete, since a one-click order with no confirmation step produces support tickets that cost more than the orders.

For the broader conversion-rate levers that this sits alongside, improve your conversion rate covers the first-order side, and improve email conversion rate covers the channel most of this runs through.

What to stop doing

Each of these is a default that arrived without a decision, which is why they persist.

The second-order discount is the most common and the most damaging. It raises the measured rate immediately, which makes it look like a success, and it teaches a cohort that your prices are negotiable. Bain and Reichheld's finding that a five percent retention improvement can raise profits by twenty-five to ninety-five percent only holds if the retained orders carry their margin, and a trained discount expectation removes exactly that.

The fortnightly schedule is the next. It exists because an email tool asked for a cadence and somebody picked one. If your window is twenty-four days, a fortnightly sequence sends a generic message at day fourteen and the useful one at day twenty-eight, four days after the decision was made.

The aggregate repeat rate is the quietest of the three. It will move for acquisition reasons every month, and a team watching it will attribute those movements to their own work in both directions. Marketing Metrics' observation that selling to an existing customer is far likelier than converting a new prospect is the reason this work is worth doing at all, and you cannot tell whether you are doing it without the cohort view.

What a store team should do this week

Three of the four are a day's work between an analyst and whoever owns the email tool.

• Run the interval query. Median days from first to second order, among customers who reordered. Segment it by first product if you can.
• Re-time the sequence so the genuinely useful message lands before the median, and say plainly which message is the useful one.
• Specify the reorder route. Variant-level, in the account and in the delivery email, with a confirmation step. Hand it to development as a one-page spec.
• Replace the dashboard metric with first-to-second-order rate by monthly cohort at sixty days. Keep the old number if you must, below the new one.
• Audit last month's first-order failures. Late, damaged, wrong. These customers have the highest intent and the worst experience, and most stores treat the queue as refunds rather than as recovery. The CRO audit checklist, step by step covers the structured version.

Where order history, email engagement and support themes have to be read together before anybody can tell which customers are drifting, that unification is the problem Nexus by Omniconvert addresses, with every proposed campaign staying something a person approves before it goes live. Baymard Institute's usability research is the external reference for the account-area and checkout friction that a reorder route removes.

FAQ: improving repeat-purchase rate

What is repeat-purchase rate and how is it calculated?

Repeat-purchase rate is the share of customers who have placed more than one order, usually measured over a fixed window such as twelve months. The version that is actually useful is narrower: the proportion of a single monthly cohort that placed a second order within a set number of days. Measured in aggregate across all customers, the number moves whenever acquisition volume moves and tells you nothing about what you changed.

Why does the second order matter more than later ones?

Because the gap between one order and two is where almost all of the loss happens. Customers who order twice are substantially likelier to order again, so the second purchase is the step that converts a buyer into a customer. Spending effort on the fourth and fifth order while the first-to-second rate is weak is optimising a stage very few people reach.

Do discounts improve repeat-purchase rate?

They pull orders forward and train customers to wait for the next one, which raises the measured rate for a quarter and lowers the margin on every order after it. A discount is the right tool for clearing stock and the wrong tool for building a habit. If a discount is the only thing producing second orders, the product or the timing is the actual problem.

When should you contact a customer after their first order?

Inside your own second-order window, which you find by measuring the median days between first and second order among customers who reordered. For a consumable that may be three weeks before the product runs out. For a durable good it may be two months, and the useful contact is about using the thing rather than about buying another. A generic fortnightly schedule misses both.

What is the fastest change that improves repeat purchases?

Making the second order shorter to place than the first. Most stores make a returning customer repeat the whole flow: find the product, configure it, re-enter details. A one-action reorder route from the account page and the delivery email removes that, costs a few days of development, and needs no discount, no campaign and no new copy.