Keep Them Coming Back

Every dollar of new business costs you a small war. Ads, content, follow ups, proposals, the whole courtship. Meanwhile, sitting in your records right now is a list of people who already trusted you once, already paid you once, and already know you deliver. Selling to them again costs a fraction of the war, converts at multiples of the rate, and most businesses never do it, because nobody owns the job of staying in touch and awkward messages are easy to postpone forever.

That is a writing problem wearing a strategy costume. The win back note you never send, the upsell you never float, the referral you never request, they all die in the same place: the blank message box. So today the Vault is stocked with seven prompts that clear the box. Each one takes a situation you have been avoiding and produces a message worth sending in under two minutes. Copy them, adapt the bracketed parts, and go collect the revenue that has been waiting politely this whole time.

For the good customer who quietly drifted. Not angry, not gone, just faded, and the longer the silence stretches the weirder it feels to break it.

The prompt: Write a short, warm reconnection email to a past customer I have not spoken to in [timeframe]. We previously [what you did for them]. Do not guilt them, do not say long time no talk, do not fake an occasion. Reference one specific detail of the past work, share one thing that is new or improved on our end that is relevant to them, and end with an easy, no pressure question. Under 120 words. Plain, human, zero salesman energy.

The magic instruction is the ban list. Left to its instincts, AI writes reconnection emails the way a timeshare company would. Telling it exactly which cliches to avoid is what produces a note the customer actually finishes reading. Send five of these a week to your faded list and watch what comes back. The typical response is not annoyance. It is oh good, I have been meaning to call you.

For the heated message sitting in your inbox that you have read four times and answered zero times. Every hour it ages, it gets harder.

The prompt: A customer sent this complaint: [paste it]. Draft a reply that does four things in order: acknowledge their specific frustration in plain words without weasel phrases, take responsibility for what was actually our fault without groveling, state concretely what happens next and when, and give them a direct way to reach me personally. Do not offer excuses. Do not use the phrase we apologize for any inconvenience. Keep it under 150 words.

The reason this works is temperature. You are angry or defensive when you read a complaint, and everything you draft in that state leaks it. The AI is not angry. It produces the calm, accountable version of you on your best day, and you get to review it once your pulse settles. A well handled complaint is one of the strongest loyalty events that exists. Customers forgive problems. They do not forgive silence.

We built the Review Machine on Tuesday. This is the ammunition it runs on.

The prompt: Here are this week's reviews: [paste]. Write a reply to each in my voice, which is warm, brief, and specific. For positive reviews, thank them for one concrete detail they mentioned, never a generic thanks. For critical ones, acknowledge plainly, state one thing we changed, invite direct contact. Remember these replies are read mostly by future customers, not the reviewer, so each one should quietly demonstrate how we treat people.

That last sentence is the whole trick. Review replies are a public performance for an audience of prospects. Framing the prompt that way turns routine acknowledgments into small trust advertisements.

For anything on a cycle: contracts, maintenance plans, annual services, reorders. The revenue is scheduled. The reminder rarely is.

The prompt: Write a renewal reminder for a customer whose [service or plan] wraps up on [date]. Lead with a specific result or benefit they got this cycle, not with the expiration. Make renewing feel like continuing something that is working, not like a bill arriving. Offer one small reason to act this week. Under 100 words, friendly, no countdown clock urgency.

Leading with the delivered value is what separates this from a dunning notice. People renew things that feel like wins. Your job is to remind them it was one.

Quick pause. These prompts, plus about two hundred more covering every message a small business sends, live in the AI Workflow Blueprint, organized so you can grab the right one in ten seconds. It is $47. Reply with BLUEPRINT and the whole vault is yours.

For the customer who bought one thing and has no idea you sell the other thing. This is stunningly common, and it is not their fault. It is yours.

The prompt: Write a short note to a customer who bought [what they bought]. People who get that usually run into [the next problem] within [timeframe]. Introduce [your other offer] as the natural next step, framed entirely around preventing or solving that problem for them, not around us having something to sell. Include one sentence of proof, a result another customer got. End with a soft question, not a pitch. Under 110 words.

The frame is service, not sales, and it is not a costume. If your second offer genuinely helps people who bought the first, staying quiet about it is the opposite of politeness. The prompt just gives you the words that feel as helpful as the intent.

The most profitable message in business and the one owners avoid hardest, because it feels like asking for a favor. Done right, it is not. It is giving a happy customer a way to look smart to a friend.

The prompt: Write a referral request to a customer who just told us they were thrilled with [the work]. Thank them briefly, then ask if they know one person dealing with [the problem we solve]. Make it easy: offer a single line they could forward or copy, written from their perspective, recommending us. No incentives, no schemes, no do you know anyone at all vagueness. One specific person, one forwardable line. Under 90 words plus the forwardable line.

The forwardable line is the piece almost everyone misses. A referral usually dies not from unwillingness but from effort. Your customer likes you fine, but composing an introduction is homework. Hand them the sentence, and the homework disappears. Timing note: this prompt fires best in the twenty four hours after praise. When someone says you did great work, that is not the end of the conversation. That is the beginning of this one.

For the day you actually blew it. Missed deadline, wrong order, dropped ball. It happens to every real business, and the message you send in the following hour usually decides whether the story ends with a lost customer or a more loyal one.

The prompt: We made this mistake with a customer: [describe it honestly]. Write an apology that names the mistake specifically in the first sentence, takes full ownership with no passive voice and no mistakes were made phrasing, explains in one sentence what we changed so it will not repeat, and offers a concrete make good of [your gesture]. Do not over apologize or grovel. Two short paragraphs, sent from me personally.

Notice the ban on passive voice. It is doing heavy lifting. The difference between the order was delayed and I delayed your order is the difference between a corporation hiding and a human standing there, and customers stay loyal to humans who stand there.

In case the strategy needs a spreadsheet to feel real, here is the back of the envelope. Say you have three hundred past customers and your average sale is five hundred dollars. Run Second Sale Hour weekly, touching five win backs, five upsell candidates, and five referral asks. That is fifteen warm conversations a week, roughly sixty a month, against people who already know you deliver.

Now use embarrassingly conservative numbers. If just five percent of those sixty monthly touches turn into a sale, that is three extra sales a month, fifteen hundred dollars, eighteen thousand a year, from one hour a week and zero ad spend. And that ignores the compounding parts: the referral customer who becomes a repeat customer, the win back who signs an annual plan, the complaint you defused who now tells the story of how you handled it. Most owners who run this honestly for a quarter report numbers well above the conservative case, because past customers do not convert at five percent. Strangers do.

Compare that to what eighteen thousand dollars of new revenue costs through ads, and the conclusion writes itself. The cheapest customer you will ever acquire is the one you already acquired. The prompts above are just the tooling that makes acting on that fact take minutes instead of willpower.

One habit makes the whole system stick: track it lightly. A single running list, who you touched, which prompt, what came back. Not for bureaucracy. Because the first month's list is what convinces you to never skip the hour again.

One at a time, these prompts fix individual moments. Wired into your operation, they become a retention system. Keep the context handy: a lightweight relationship tool like Clay quietly remembers who you know, what you talked about, and when you last spoke, which is exactly the raw material the win back and referral prompts feed on. Put the recurring ones on rails: renewal nudges and review replies can trigger automatically through Make.com or run natively inside Go High Level, with AI drafting and you approving. And block one hour a week, call it Second Sale Hour, where you run the win back, upsell, and referral prompts against five names each. That single hour, most weeks, will out earn a day of chasing strangers.

The customers you already served are the warmest market you will ever have. This week, stop leaving them on the table. Open the vault, pick the prompt that matches the message you have been avoiding longest, and send it before lunch.

And if you want these installed as a running system instead of a good intention, that is the work we do inside the AI Business Accelerator, $97, where we build your retention engine around your actual customer list. Reply with ACCELERATOR and bring your five names.

Jordan

The AI Newsroom | Jordan Hale | ainewsroomdaily.com