Conversion: win sales.

Getting here, especially the first time, is the hardest thing a brand does. You won someone over. Congratulations, and now the part most brands get wrong: the first order is not the finish line, it is the moment the relationship starts, and the next two months decide whether it was worth anything.

I studied every first-time buyer at a large California retailer from February 2024 to January 2026, grouped by the month of their first order: 24 cohorts, about 1.5 million people. The average first order was $81. Over two years the average first-time buyer was worth $116, so the first order was 70% of everything that customer would ever spend. About 64% never placed a second order. The steepest drop in repeat buying came between the month of the first order and month two; from month six onward the line was close to flat, with one small bump at month twelve that looked like people coming back for the same season.

The research says the same thing from the other side. Reichheld and Sasser's finding from 1990: hold on to 5% more of your customers and profits rise by a quarter to nearly double, depending on the business. Gallo's summary in HBR two decades later: winning a new customer costs five to twenty-five times what keeping one does. Every dollar beyond the first order comes from someone who placed a second one, and the window for the second one closes in about sixty days. That is the whole economics of this stage.

The step: place order

One step, and everything upstream was for it. The mistake is to treat it as the end of the funnel. Nothing is worse than a salesperson who is delighted to take your money and has nothing to say afterwards. If you want them to return, this is the window, because the data says it closes fast.

Here is the hint I give every brand, and it surprises people with how well it works: in the post-purchase series, write content that does not ask them to buy. Explain more. Be timely. Create a different experience depending on how many times the person has bought. Drucker's line about marketing is the standard: know the customer so well that the product sells itself. A first-time buyer does not need a second pitch; they need to be shown they chose well, and told what happens next.

At the retailer, the post-purchase series for a new customer runs 14 to 21 days. The confirmation suggests pieces that go with what they bought. The delivery message asks them to share a photo. Five days later a review request, with 10% off the next order for leaving one. At day 14, an email recommends new arrivals based on their order. Every flow splits by tier, so the same trigger gets a different message depending on where the person is: a first-time buyer who leaves a cart gets the 15% first-order offer at 72 hours; a one-time customer gets 10% off the cart at 24 hours, because the next few weeks are when a second order is most likely; customers with two or more orders get free shipping or perks instead of a discount. Three people, three experiences, one trigger.

What the discount buys

Conversion is where discounting gets its grip, and the research on what a discount buys is thirty years old and still ignored. Ehrenberg, Hammond and Goodhardt went through years of panel data on price promotions in 1994 and found the pattern every brand's own numbers show: a promotion produces a sharp sales peak, the peak is made almost entirely of people who already bought the brand, and when it ends there is no after-effect, no new habit, no lift in the following weeks. The promotion did not create customers. It moved existing customers' purchases into the window and paid them to do it.

A golf and corporate apparel brand had lived that for two years. In 2025, 95% of orders in the Black Friday window used the 30% sitewide discount, and it came to 29.1% of the list value of everything sold. Orders in the window fell 29.5% anyway, and November and December revenue fell 24%. The discount was on nearly every order and did not hold the volume. By August of this year the brand had walked away from promotions and was growing at full price, orders up 8.3% at a flat average order value, which is demand and not a mix effect.

For Black Friday the brand kept every product at full price and added a gift card for the next visit: spend $150, get $20; spend $300, get $35. I sized the rungs on the only full-price month on record, where the median order was $115 and the average item $83, so the first rung sat about one item above the median, where 34% of orders already landed. A gift card costs nothing until someone spends it, and only the orders that reach a rung earn one; planned at 40% redemption, a $20 card costs about $8 per card issued. The redemption rules sat on the second order, with a 45-day window that ran into January, the quietest month, so the return visit landed there at full price. Modelled on the August orders, the tiers' expected cost per order was roughly a twelfth of the discount's, and they matched last year's total margin at just over half of last year's orders.

The objection: but our checkout flow converts at 3%

It does. At a pharmacy brand, abandoned checkout converted 3.22% of recipients and ranked first of four flows, and the suggestion in front of me was to make it the place to grow. The ranking measured intent: the people in that flow had picked items and started to pay. Any flow that fires after a filled cart is credited with orders that were already in motion. The only way to know what the checkout series adds is to hold back a tenth of entrants for two weeks and compare order rates from the store's data; the holdout worksheet on this site does the arithmetic. The flow is probably doing something. The report cannot tell you how much, and the brand that scales a budget on it is scaling on intent it did not create.

The flows: post purchase, upsell

The post-purchase series is the one above, and it is the most under-written flow in most accounts: three emails that end on day three with a review request, then silence until a winback months later. The upsell loop fires earlier, between add to cart and the order, and it is where the counting problem bites first. Write the post-purchase series for the sixty days, not the three. Treat the upsell as a service, the item that completes the one in the cart, and measure it with a holdout before anyone calls it a growth lever.

What I check first

• The reorder curve. Group first-time buyers by the month of their first order and find where the drop is steepest. At the retailer it was inside two months. Yours depends on what you sell.
• Days 0 to 60, message by message. Which were written for a one-time buyer, and when does the series stop? If it stops at day 14 and the next message written for them is a winback months later, the best window is covered only by the general calendar.
• What the discount bought last time. Share of orders that used it, its cost as a share of list value, and whether orders rose. At the apparel brand the answers were 95%, 29.1% and no.
• One holdout. On the flow with the best order rate, because that is where attribution flatters most.

If you take one thing from this stage: the first order is 70% of the customer, and the other 30% is decided in the next sixty days by messages that do not ask for it. The order is placed. Now the brand has to keep a promise, and the next stage is the first time the customer holds the product.