Customer Retention Campaign: Start With Who Stopped Buying

A customer retention campaign starts with a date, not a discount: the date of each customer’s last purchase. Most small businesses can list who bought last month in a few clicks. Far fewer can list who used to buy and has quietly stopped, because no system flags a sale that did not happen. The campaign below is mostly the work of building that list. The message you send comes second, and the discount, if there is one, comes last.

The case for doing it rests on arithmetic that is older than most of the software sold to run it. Writing in Harvard Business Review, Amy Gallo put it this way: “depending on which study you believe, and what industry you’re in,” acquiring a new customer costs five to 25 times more than retaining one, and she cited Bain’s Frederick Reichheld on a 5 percent gain in retention lifting profits by 25 to 95 percent (hbr.org). Read the hedge in that sentence as carefully as the numbers. The ranges are wide because every business is different, which is exactly why the campaign has to begin with your own sales records rather than anyone’s benchmark.

This is the follow-on to the first 30 days. Our playbook on stopping early customer churn with a four-touchpoint cadence covers the customer who just arrived. This one covers the customer who has been with you for a while and has gone quiet.

Step 1: Decide what “stopped” means in your business

There is no universal number of days after which a customer counts as lost. A coffee shop’s regular who has not been in for three weeks has stopped. A roofing company’s customer who has not called in three years may be exactly on schedule. The only honest definition comes from your own records: how long your customers normally go between purchases.

Pull a list of every customer with at least two purchases and the dates of those purchases. For each one, note the gap between purchases. The middle value of all those gaps, the median, is your normal rhythm. Any customer whose silence now runs past about twice that gap has, by your own numbers, stopped. Twice the median gap is our working rule, not a published standard, and you should tighten or loosen it after one campaign shows you how the list behaves.

Hypothetical, for illustration: a two-chair dog grooming salon exports 18 months of appointments and finds that its repeat clients come back about every six weeks, with a median gap of 45 days. Twice that is 90 days. Every client whose last appointment is more than 90 days old goes on the list. The ones at 60 days do not, because many of them are simply due next week. The salon’s list is shorter than the owner feared and far more specific than a “we haven’t seen you in a while” blast to everyone.

Step 2: Build the list, then sort it by what each name is worth

Where the list comes from depends on where your sales are recorded.

If you sell through Shopify, the customer segment editor does this directly. Shopify’s help pages describe each segment as a ShopifyQL query, say a segment can be sorted by last order date, and give this exact example of a plain-language request to its Sidekick assistant: “Create a segment of customers who last ordered more than 90 days ago and are subscribed to marketing emails” (help.shopify.com). Swap 90 for the number you found in Step 1.

If you invoice or run a point-of-sale system, export your sales or invoices for the last two years to a spreadsheet, one row per sale with the customer, the date and the amount. Sort by customer, then by date, and the last row for each customer is their last purchase. A pivot table, or an AI assistant you paste the export into, can turn that into one row per customer with three columns: last purchase date, days since, and total spent. Replace names and contact details with a customer number before pasting anything into an AI tool; the number is enough to match the rows back afterwards.

If you have neither, the appointment book, the job calendar or the folder of emailed receipts is the record. Thirty minutes with a highlighter finds the names that used to appear every month and no longer do.

Then sort by total spent, highest first. The campaign is not one message to everyone on the list. The top of the list gets a phone call from the owner. The middle gets a personal email. The bottom gets the automated sequence. Our piece on customer acquisition cost counted in hours, not just cash is the companion calculation. An hour of calls to the ten biggest lapsed accounts is often cheaper than a month of ads aimed at strangers, because a returning customer already knows what you sell and what it costs.

Step 3: Write three messages, in this order

Write the phone script first, then the personal email, then the automated sequence. Writing in that order keeps the automated version honest, because it has to say what the owner would say on the phone, only shorter.

The call. Two questions and no pitch: did something go wrong, and what would bring you back. The answers are the most valuable output of the whole campaign, because a pattern in them is a problem in the business, not in the marketing. If three lapsed customers mention the same thing, fix that before sending anything to the rest of the list. It is the same logic as building a customer journey map from the people who left instead of from the ones who stayed.

The personal email. One paragraph, from your own name, that says you noticed, asks the same two questions, and makes replying easy. No coupon in the first message. A customer who left because an order arrived late is not won back by 10 percent off, and a customer who simply forgot you exist does not need the discount to come back.

The automated sequence. Two or three emails, a week or two apart, to everyone further down the list. The first mirrors the personal email. The second adds something that is new since they last bought: a product, a service, longer hours. The third, if you want a discount anywhere, is where it goes, with an end date. The offer comes last because it is the most expensive line in the campaign and the only one you cannot take back.

Our view: the discount is where most retention campaigns start and where this one ends, on purpose. A discount sent first teaches your best customers that going quiet is how they get a deal. The phone call and the two questions cost an hour and teach you something instead.

An AI assistant such as ChatGPT, Claude or Gemini can draft all three messages from a short description of your business and the number you found in Step 1, and can rewrite the automated emails in your own voice from a few past messages you paste in. The owner reads every line before it goes out, and the two questions in the call stay exactly as written.

Where the line is on email and text

Email to past customers is legal, with rules. The FTC’s CAN-SPAM guide for business says a marketing email’s “From” line must accurately identify the business that sent it, that the message must tell recipients how to opt out, that an opt-out request must be honored within 10 business days, and that each separate email in violation can draw a penalty of up to $53,088 (ftc.gov). The tools below add the opt-out link for you. The accurate “From” name is your job.

Text messages are stricter. The FCC says its rules ban text messages sent to a mobile phone with an autodialer unless the phone’s owner previously gave consent, and that commercial texts require written consent (fcc.gov). A phone number on an old invoice is not that consent. If you want texting in a future campaign, collect the written opt-in now, at checkout or on the intake form. None of this is legal advice; for anything beyond these basics, read your email provider’s compliance pages or ask an attorney.

What the tools cost

The spreadsheet and the phone cost nothing, and for a list under a few hundred names they are enough. For the automated sequence, here are three options with prices read from each vendor’s pricing page on November 6, 2026.

Mailchimp has a Free plan that its pricing page suggests for businesses with fewer than 250 contacts. Essentials starts at $13 a month and Standard at $20 a month, each with a 14-day free trial, and the price rises with your contact count (mailchimp.com).

Klaviyo is free for up to 250 active profiles and 500 emails a month. Its Email plan is $20 a month for 500 profiles and 5,000 emails, and its Marketing plan is $25 a month, currently promoted at $1 for the first month (klaviyo.com).

HubSpot offers its free marketing tools at $0, and Marketing Hub Starter is listed at $20 a month per seat, shown with a $7 promotional rate for new customers, including 1,000 marketing contacts (hubspot.com). HubSpot also stores contacts you do not market to at no charge, which suits a list where only the bottom tier gets automated email.

Any of the three can hold the lapsed list as a segment and send the two or three automated emails on a schedule. None of them can make the phone call.

Step 4: Count what came back

Two numbers tell you whether the campaign worked, and both come from the same spreadsheet.

The first is recovered customers: how many names on the list bought again within 60 days of the first message, and what they spent. Set that against the hours the calls took and the monthly fee of the tool. A customer who came back has just told you, by buying, that the relationship is worth something, which makes them the natural first person to hear about your small business referral program in three sentences.

The second is your customer retention rate, which is worth knowing even if you never run a second campaign. Take the customers who bought in the twelve months that ended a year ago. Count how many of them bought again in the twelve months since. Divide the second count by the first. That share is the proportion of customers you keep from one year to the next. Write it down with today’s date. Next year’s figure, set against this one, is the only retention benchmark that is actually about your business.

The prompt behind this playbook

This article grew out of a Pro prompt on our sister site BusinessPrompter, Customer Retention Campaign, whose one-line summary is “Create campaigns to reduce churn and increase customer lifetime value.” It is a five-step chain that walks an AI assistant from the principle to a campaign plan. It sits behind the Pro upgrade, so we have not run it here and will not describe its steps. One disclosure: the “About This Prompt” panel on that page currently describes an operations-data analysis for manufacturing and logistics leaders rather than a retention campaign, so judge the prompt by its title and summary. The library is at BusinessPrompter.com.

Before any of that, tonight: open your sales records, find the customer who has spent the most with you and has not bought in twice your normal gap, and call them tomorrow morning. The campaign’s first message is not an email. It is that call, and what they say decides what every other message needs to fix.