Acquisition: earn contact trust.

Someone spends time in your store and decides to subscribe, or at least to let you have some data: a cart, a browsed product, an address. At the very least they have consented to a profile, and from here on the brand can speak. Every flow in this stage exists to turn that consent into a first order, and the quality of all of them comes down to one thing: how well the brand understood why the person came.

Peter Drucker put the purpose of a business in one sentence in 1954: to create a customer. Seth Godin, forty-five years later, put the condition on it: the messages have to be anticipated, personal and relevant, or the permission goes away. Acquisition is the stage where both get tested, four times, by four flows, and the research on what happens between the first click and the order is not flattering. Baymard's running average across 49 studies puts cart abandonment at about 70%. Seven in ten people who put something in a cart do not buy it. In Baymard's own survey, the two biggest reasons were not doubt about the product: extra costs that appeared at checkout, 48%, and being made to create an account, 26%. Most of the people lost at this stage were not unconvinced. They were inconvenienced, and no flow fixes a shipping surprise.

The steps: subscribe, browse, add to cart, start checkout

Four steps, each one a little more intent than the last. The flows map onto them: the welcome series after a subscribe, the browse series after a product view, the cart series after an add to cart, the checkout series after a started checkout. The names are standard. What is not standard is whether they answer the reason the person came.

At an architectural products retailer, the reason was easy to find once we asked. Finishes are a big purchase and buyers want to see the finish in their own light before they commit. So the funnel sold a small sample first and treated the full order as the second sale. Hopkins built campaigns on exactly this a century ago: get the product into the hand and let it argue for itself. Kahneman, Knetsch and Thaler gave it a mechanism in 1990, the endowment effect: once a person holds a thing, they value it more than they did a minute before. A sample on the kitchen table is a finish the buyer already half owns.

The Meta campaign was built on four customer worries: will it look real, will it fit, is it the right finish, can I install it. Each worry had an ad, a landing page and, after the sample shipped, an email that picked it up. Followed forward from their first sample, buyers placed a full order a median of about 25 days later, and 11% of sample buyers in the lead category ordered in full within 60 days. The flows were built for that wait, not for the day the sample arrived. The launch nearly failed on day one, when a CRM integration rewrote the tracking on all 45 ads and they came back from review with their media stripped; we rebuilt from one canary ad, and within days sample orders went from 15.4 a day to 19.0.

At a hand-dyed yarn brand, people choose yarn by feel, so getting it into someone's hands became the acquisition event. The offer was one free mini in a colorway they could reorder. The brand's old birthday gift had been running as an ad to new people, and only 6% of its 740 redemptions over 13 months were first-ever orders, with baskets 38% smaller than the store average because the gift had no minimum. We agreed to judge every offer on cost per first purchase, not cost per lead, and the gift moved to lapsed customers where it belonged. A free pattern had looked cheapest on cost per lead, until we priced in that most patterns pay the designer about half the list price per copy.

The flows are not an email thing

Many people think flows live only in email, text and WhatsApp. They are omnichannel. The welcome series at the retailer was an ad, a landing page and an email, in that order, and the sample buyer's worries were answered on all three. The brand's paid team built the campaign and pages to my brief; the measurement and the Klaviyo flows were mine; the brand saw one funnel with one set of numbers. A browse series that only exists in email is a browse series for the share of visitors who gave you an address, which at most brands is a small minority of the people who browsed.

The flows also break in ways the dashboard does not show. At a kidney-health brand, both welcome email flows triggered on the master list, and the SMS welcome fired on any signup, including back-in-stock alerts, with smart sending off. Nothing kept one person out of two welcome sequences at once. At the yarn brand, the birthday signup page used the wrong form, so people who signed up joined no list and never got a welcome email. Neither problem appeared in a report; both appeared when I signed up myself and watched what arrived. Godin's permission is withdrawn the day the brand sends two welcomes and a text to someone who asked for a restock alert.

The objection: our ads say the cost per customer is fine

The ad platform's cost per new customer is the number everyone plans on, and at the yarn brand it was wrong in a way nobody could see from inside the platform. Meta could match 65 to 76% of the purchaser list, so about a third of existing customers could not be excluded from acquisition ads. In September, ads to people who already buy were credited with about 60% of Meta-attributed revenue on 23% of the spend. The platform was reporting a cost per new customer while a share of the customers were not new. The only clean number was the store's: first-order dates, counted by month, next to spend. That is the number acquisition should be judged on, and it is the one the sample funnel was built to produce.

When you move platforms, these come first

The acquisition flows are the ones to rebuild first after a migration, because each one answers something a person just did, and a gap there means someone signed up or left a cart and heard nothing. When a large California retailer moved 9 million profiles during peak Q4, the welcome, browse and cart abandonment flows were live again inside 30 days of cutover, before any finer segmentation. The sequence was deliberate: get the four answers back, then make them better.

What I check first

• Sign up yourself. With a new address, from the main form and from every other form on the site. Count what arrives, from which flows, and whether two of them collided.
• The reason, written down. The four worries at the retailer fit on an index card. If the brand cannot say what stops a first order, the flows cannot answer it.
• The checkout surprises. Shipping, tax, account creation. Baymard's two biggest reasons for leaving are both fixable before any email is written.
• Where the first order comes from, by cost. Cost per first purchase, from the store's orders, not cost per lead and not the ad platform's number.
• The wait. Follow one month's new subscribers or sample buyers forward and write down the median days to a first full order. Then check that the flows are still talking to them on that day.

If you take one thing from this stage: find out why they came, and make sure every flow answers it on every channel. When the first order lands, the hardest thing in the funnel has happened. The next stage is about not wasting it.