How to Build a First-Visit Follow-Up Sequence

Why this matters

Most shops have follow-up. Very few have a sequence. The difference is that follow-up is a message someone remembers to send, and a sequence is a fixed set of touches with defined jobs, defined timing, and defined stop conditions, which means it runs whether or not anyone remembers.

This matters most on first visits, because the first-time customer is the one you know least about and the one most likely to leave without a word. A single well-written follow-up email covers about three days of a twelve-month risk window. The rest of that window is where first-time customers quietly become somebody else's regulars.

This article is about designing the sequence itself: how many touches, what each one is for, what each one must not do, and how to stop it. It assumes nothing beyond a calendar and a list.

Step 1: Name one outcome for the whole sequence

Write it down before anything else: the sequence exists to produce a second booked job within 12 months. Not reviews, not referrals, not brand awareness. Those are byproducts and some of them are legitimately worth collecting along the way, but they are not the thing you measure the sequence against.

Skipping this step is why most sequences degrade. Without a single named outcome, every quarter somebody adds a message, the sequence becomes a newsletter, and the customer's unsubscribe is the only feedback you ever get.

Step 2: Give every touch a different job

The design rule that carries the whole thing: no two touches in the sequence ask for the same thing. If two touches have the same job, one of them is a reminder, and reminders are how a sequence starts feeling like pressure.

Each touch gets three lines written down before you write a word of the actual message:

• Its job. The one thing this contact is for.
• Its trigger. A number of days from a specific event, or a named month.
• Its prohibition. What this touch is explicitly not allowed to do, because that is the thing it will drift into.

The prohibition line is the one people skip and the one that keeps the sequence clean two years later.

Step 3: The five touches

Five is not sacred, but the shape is. Two touches inside the first two weeks while the visit is still real to them, one at a quarter, one on the date you actually agreed to, and one at the year mark. Adding a sixth touch between months 4 and 11 almost always produces a message with no job, which is exactly the message that gets you filtered.

Why touch 1 cannot carry the review ask. At 72 hours you do not yet know whether the repair held. Asking for a public review before you have confirmed the outcome means you are occasionally asking an unhappy customer to publish, and you are always making the first post-visit contact about your needs rather than theirs. Confirm first, ask second. The two-step version reliably produces both a better review rate and fewer surprises.

Why touch 4 is a call and the others are not. It is the only touch asking for a scheduling decision, and scheduling decisions have back-and-forth in them. An email asking someone to pick a date generates a reply, then a counter-reply, then silence. A two-minute call ends with a date.

Step 4: Write the stop rules before the messages

A sequence without stop rules eventually contacts a customer who booked last week, or a customer who told you to stop, and one of those two errors costs you the relationship outright. Four rules cover almost everything:

• A booking ends the sequence. The moment a second job is on the calendar, the customer leaves this track entirely. They are now a repeat customer on a different rhythm.
• A stated no-contact request stops everything, permanently, across all channels. Record it where anyone can see it, not in one person's inbox.
• A complaint or a callback pauses the sequence. It resumes only after the issue is closed, and touch 2's review ask is skipped for that customer unless the recovery went genuinely well.
• Two consecutive non-responses on one channel demote that channel. If two texts get no reply, touch 3 goes out by a different channel or not at all. Continuing to push a dead channel is the single most common way a sequence turns into noise.

Write these on the same page as the sequence itself. A stop rule filed somewhere else is a stop rule that does not exist.

Step 5: Match the channel to what you are asking for

Channel is not a style preference, it is a function of the size of the ask.

• Confirmations and records go on whatever channel the customer used to reach you. They contacted you that way; that is their revealed preference.
• Anything with an attachment or a photo goes to email, because it needs to survive in a place they can find it in six months.
• Anything asking for a decision or a date goes to a phone call.
• Nothing important goes only to a channel you have not seen a reply on.

The failure mode is a shop that puts the entire sequence on email because email is easy to send in bulk, then concludes that follow-up does not work. What did not work was asking for a scheduling decision by bulk email.

Step 6: Build the calendar that runs it

This is the part that decides whether the sequence survives contact with a busy month.

Take a wall calendar or a spreadsheet with one row per first visit and five date columns. When a first visit closes, the office fills all five dates at once, right then, from the visit date. Touch 4's date comes from the return month the tech recorded; if there is no return month, that cell is left blank and the customer skips touch 4 entirely.

Then the daily routine is one question: what is due today? Not "who should we follow up with," which requires judgment and therefore gets skipped, but a list you read off a page.

Running this from software changes nothing about the design. It changes who fills the dates. The design failure and the software failure are the same failure: nobody owns the daily "what is due today" read.

Step 7: Make touch 3 something only you could have written

Touches 1, 2, and 5 can be templated with a few merge fields and they will read fine. Touch 3 cannot. Its job is to reference the specific item your tech flagged on that specific property, which means it needs one sentence a template cannot produce.

The template carries the frame and the person writes the sentence:

"When we were out in March, the item we flagged was [the specific thing, in plain language]. It is not urgent, but it is the kind of thing that gets decided either in the fall or on the worst day of the winter. Do you want us to take a look at it while the calendar is open?"

That one sentence is the difference between touch 3 converting and touch 3 getting filtered. It takes about ninety seconds to write from the job notes.

Worked example: 30 first visits through a full sequence

A shop logs 30 first visits in one month and runs the whole cohort through the five touches, tracking what each touch produced.

• Touch 1, 72 hours. 30 contacted, 21 replied, which is 70 percent of the 30 contacted. Two of those 21 replies reported the problem was not actually resolved. Both route to the callback process and leave the sequence. 28 remain.
• Touch 2, day 12. 28 contacted, 9 reviews posted, which is 32 percent of the 28 asked.
• Touch 3, month 3. 28 contacted, 4 booked a second job, which is 14 percent of the 28 contacted. Those 4 exit as repeat customers. 24 remain.
• Touch 4, the return month. A return month had been captured on only 13 of the 30 original visits, about 43 percent of the cohort, and one of those 13 had already booked at touch 3. So touch 4 reached 12 people. Of those 12 calls, 5 booked, which is 42 percent of the 12 called. 19 remain.
• Touch 5, month 12. 19 contacted, 2 booked, about 11 percent of the 19 contacted.

Total second jobs from the cohort: 4 plus 5 plus 2, which is 11 second jobs out of 30 first visits, about 37 percent.

Now read what the numbers actually say. Touch 4 converted 42 percent of the people it reached, against 14 percent for touch 3 and 11 percent for touch 5. It is the strongest touch in the sequence by roughly three times, and it produced 5 of the 11 total bookings while reaching fewer than half as many people as any other touch.

And it reached only 12 of 30 people, not because the sequence failed, but because the first visit never produced a return month for 17 of them. The binding constraint on this shop's whole follow-up program is not the sequence design. It is a field the techs are filling less than half the time.

Estimate the size of that gap honestly. If capture went from 13 to 26 of 30 and the 13 newly-captured records converted at half of touch 4's observed rate, about 21 percent, that is roughly 3 additional bookings, moving the cohort from 11 of 30 to about 14 of 30, from 37 percent to about 47 percent. Assume instead that the added reasons are weak enough to convert at touch 3's 14 percent, and you get roughly 2 additional bookings, landing near 13 of 30, about 43 percent. Either way the answer is the same: the next quarter's effort belongs on the return-reason field at the truck, not on rewriting touch 5.

The wrong read, and it is the common one, is to look at touch 5's 11 percent and cut it. Touch 5 cost two phone calls and produced two jobs from customers who had gone eleven months without contact. It is the cheapest touch in the sequence per booking. Cut it and you have not fixed anything; you have removed the last chance at 19 customers.

How to verify you got this right

Every touch has a job, a trigger, and a prohibition written down. If you cannot state a touch's prohibition, that touch will drift into being a second copy of the touch before it.

The daily "what is due today" list gets read every working day. Check by sampling: pick a week two months back and confirm the touches dated in that week actually went out. A sequence that runs at 60 percent completion is not a sequence, and its measured conversion rate is meaningless because you do not know which touches the customers actually received.

Conversion is tracked per touch, not for the sequence as a whole. A blended sequence conversion rate cannot tell you which touch to fix. Per-touch numbers pointed straight at the capture problem in the example above; a single blended number would have suggested rewriting messages.

Stop rules have fired at least once. If no customer has ever exited early for a stated reason, either the sequence is not reaching anyone or nobody is recording exits, and both are worth finding out before you scale it.

References

• U.S. Small Business Administration (SBA), small business customer communication and retention guidance
• Trade-standard practice for service follow-up, callbacks, and workmanship confirmation
• See related: How to Turn a First-Time Call Into a Second One, How to Follow Up After a Completed Job, The Follow-Up Automation Worth Setting Up Once, The Customer Review Request SOP