Sales follow up process for teams that sell in DMs

A sales follow up process that runs inside direct messages is not a cadence of days, it is a cadence of windows. On Instagram, Messenger and WhatsApp the platform decides when your team is allowed to speak at all, and every step you write on a whiteboard has to fit inside that permission or it silently fails. Teams that design a follow up sequence in days first and discover the rules afterwards end up with steps that never send, steps that send but cannot say anything useful, and steps that quietly cost money nobody budgeted for.

This is written for the person who runs several closers or several client accounts and needs one process that behaves the same on every account: agency owner, head of sales, ops manager. The platform rules below come from Meta's own developer documentation, linked page by page and checked on the day of writing. The numbers about your own pipeline are yours to compute, so what follows gives you the model rather than a benchmark we cannot verify for your market.

TL;DR

• Your follow up process is capped by a messaging window, not by your discipline. On Instagram and Messenger the standard messaging window is 24 hours; on WhatsApp the customer service window is 24 hours.
• Inside that window you can write freely and include promotional content. Outside it, your options shrink to message tags, a manual human agent reply, or a paid approved template.
• The Human Agent tag lets a real person reply manually within a 7 day period. It is an escalation path, not a marketing channel: message tags may not carry promotional content.
• On WhatsApp, template messages are the only way to reach out after the window, and Meta charges per delivered template based on category and recipient country calling code.
• At team scale the hard part is ownership, not copy: who owns the thread when the window is still open and the shift has changed.

Table of contents

• Why a sales follow up process breaks at team scale
• The messaging window decides your follow up process
• Map the process to windows, not to days
• The follow up skeleton that fits inside the first window
• What to do once the window has closed
• Who owns the follow up when the shift changes
• No shows and dormant leads are two different processes
• The six numbers that tell you the process works
• Where SetScale fits
• FAQ
• Conclusion

Why a sales follow up process breaks at team scale

One person answering one inbox does not need a process. They remember the three conversations that matter, and the follow up happens because the same brain holds the whole pipeline. Put four setters, ten client accounts and two time zones on top of that and three failures appear immediately, always in the same order.

The first is the orphan thread. A lead replies at the end of a shift, the setter logs off, and nobody picks the conversation up because it does not appear anywhere except in an inbox someone else has to think to open. The second is the double touch, where two setters answer the same person within a few minutes, usually with contradictory qualification questions. The third is the expired step: a follow up written as "day 3" arrives after the platform has closed the window, so the message either fails outright or arrives stripped of everything that made it useful.

None of these is a copywriting problem. They are ownership and timing problems, which is why a follow up process for a team has to be built on the two things the platform actually enforces: who may write, and until when. The first half of that question, how fast the first answer has to arrive, is covered in our lead response time SLA rules. This article covers everything that happens after that first reply.

The messaging window decides your follow up process

Meta's Messenger Platform documentation defines the Standard Messaging Window as the 24 hour period in which you are allowed to send a message to a person. It opens when someone sends your Page or Instagram professional account a message, and also, which many teams miss, when they click a call to action button in a conversation, when they react to a message, when they comment on one of your posts, or when they click a click to Messenger ad and then send a message. Messages sent inside that window may contain promotional content.

Once it closes, the API returns error 1545041, "messaging window closed". At that point you have three narrow options rather than a free hand.

Message tags let you send outside the window, but only for an approved use case and only with non promotional content: no deals, offers, coupons or discounts, and the platform policy warns that misuse can restrict the account's ability to send messages at all. Tags also get withdrawn. From 27 April 2026, requests carrying the CONFIRMED_EVENT_UPDATE, ACCOUNT_UPDATE and POST_PURCHASE_UPDATE tags return error 100, so a commercial cadence built on tags is built on ground that moves.

The Human Agent tag is the one that matters for a sales team. It allows a business representative to manually respond to a person's message within a 7 day period, for flows that need an escalation path. Read that literally: a human, manually, answering something the person actually said. It is not a slot for an automated day 5 bump.

WhatsApp works on the same clock with a different economy. A user message opens a 24 hour customer service window. Inside it you send ordinary messages, and Meta has not charged for non template messages since 1 November 2024. Outside it, template messages are the only type that can be sent to reach out to a user, and since 1 July 2025 Meta charges per delivered template, with rates that vary by template category (marketing, utility, authentication) and by the recipient's country calling code. Utility templates sent in response to users inside an open window are not charged, and everything sent inside an open free entry point window is free for 72 hours. The same page flags pricing updates for service and utility messages launching on 1 August 2026 and 1 October 2026, so a cadence that depends on paid templates is a cadence with a moving cost line. The full channel setup sits in our WhatsApp Business API rules for teams.

Channel	What opens the window	Inside the window	Once it has closed
Instagram and Messenger	a message, a reaction, a comment on your post, a click on a call to action button, a click to Messenger ad followed by a message	24 hours, free form, promotional content allowed	message tags with non promotional content only, or a manual reply under the Human Agent tag within 7 days
WhatsApp Cloud API	a message from the user	24 hours, free form, non template messages not charged since 1 November 2024	approved templates only, charged per delivered template by category and recipient country calling code

Two mechanics belong in your playbook. A private reply to a public comment is a single message, it automatically carries a link to that post, and it has to be sent within seven days of publication. And the one time notification, which asks permission to send one follow up after the window, is not available on the Instagram Messaging API, so any process that assumes it will work on Instagram is wrong before it ships.

Map the process to windows, not to days

Rewrite your cadence with the window as the unit. In practice that means splitting every follow up sequence into three zones and giving each one its own rules, its own owner and its own copy.

Zone one is the open window. It runs from the moment the lead does something that opens it and lasts 24 hours. This is where almost all of your qualification, your objection handling and your booking should live, because it is the only zone where you can say anything you like.

Zone two is the escalation zone. The window has closed and there is a real, unanswered human question on the thread. A named person replies manually under the Human Agent tag, within 7 days, to the thing the lead actually asked. One message, written by a human, answering a specific point.

Zone three is re-entry. The window has closed and there is nothing left to answer. Your only honest moves are to bring the person back to a channel where a new window can open (an ad, a broadcast they opted into, an email if you have consent) or, on WhatsApp, to accept the cost of an approved template. Re-entry is a marketing decision with a price, not a follow up step, and it belongs in a different budget line.

The effect of this mapping is uncomfortable and useful: most five step, seven day DM sequences sold as templates online are three workable steps and four that will fail or misfire.

The follow up skeleton that fits inside the first window

Inside 24 hours the constraint is attention rather than permission. The skeleton below is deliberately boring, which is what makes it survive being run by four different people on ten different accounts.

1. The acknowledgement. Sent inside your response time SLA, it confirms a human is on the thread and asks exactly one qualifying question. One question, not three.
2. The qualification exchange. Two or three turns, one question each, in a fixed order the whole team uses. A fixed order is what makes a handoff possible later.
3. The offer of a slot. As soon as the qualification threshold is met, a concrete proposal with two options, not an open invitation to pick a time.
4. The soft nudge. If the thread has gone quiet for a few hours and the window is still open, a short message that adds one piece of information rather than repeating the ask.
5. The window close. A final message before the 24 hours run out that makes it easy for the person to reopen the conversation later, because their reply is what buys you a new window.

Step five is the one teams forget and the one that pays. Every reply from the lead reopens the clock, so the last message inside a window should be designed to earn a reply, not to close a sale. Who does which step, and where the setter hands over to the closer, is laid out in the setter closer model, and the ramp for teaching it to new hires is in our 30 day plan to train appointment setters.

Nothing here needs a tool. It needs a written order of operations that four people follow the same way at three in the afternoon and at eleven at night.

What to do once the window has closed

Treat the closed window as a branch, not as a failure. Three cases, three different answers, and the difference between them is whether the lead asked something.

If the lead asked a real question and nobody answered it, that is an escalation. A named human replies manually under the Human Agent tag inside the 7 day period, and an unanswered question at window close should be logged as a defect: it usually means the shift handoff failed rather than the lead going cold.

If the lead simply stopped replying, there is nothing to escalate. Do not dress a marketing bump as a service message. Move the person into your re-entry list and leave them there until you have a genuine reason to reopen the conversation.

If the conversation happened on WhatsApp and the lead is worth a paid touch, that is a template decision. Price it: a re-entry campaign across ten client accounts has a real invoice attached. The arithmetic of what a booked meeting is allowed to cost you sits in our breakdown of what appointment setting actually costs.

The product is not open yet, so there is nothing to buy today. If you want to hear when it is, Join the waitlist.

Who owns the follow up when the shift changes

The window does not pause when your team goes home, which makes ownership the highest leverage rule in the whole process.

Write three rules. One thread has exactly one owner at any moment, and that owner's name is visible to everyone who can open the inbox. An owner going off shift does not release the thread, they transfer it, naming the person who takes it. And any thread whose window closes while it is still open on your side gets reviewed rather than reassigned.

Two questions decide how you implement this. Do you know, per account, which threads have an open window right now? And does the person coming on shift see them without opening ten separate inboxes? If either answer is no, your follow up process depends on someone remembering, which is exactly what stops working at four setters. The multi account version of that problem is covered in the agency playbook for running many Instagram accounts, and the resale version, where you run this under a client's brand, in our guide to white label DM automation.

No shows and dormant leads are two different processes

They get lumped together because both look like silence, and they need opposite treatment.

A no show is a live lead. The person chose a slot, so there is intent on record and, usually, a recent interaction that opened a window. Give it a fixed branch: a same day message inside the window, a single named owner, and one honest re-booking offer rather than three. Since a reply reopens the clock, a no show handled inside the window costs you nothing beyond attention. Keeping those slots filled is the closer side of the same problem, covered in keeping high ticket closing calendars full.

A dormant lead is not live. The window closed weeks ago, there is no open question, and no tag legitimately covers reaching back out. Reactivation is an acquisition motion: an ad, an opt in broadcast, a paid template where the channel allows it. Deciding to spend on that is a different meeting than deciding how fast a setter answers, and the paid entry point mechanics are in our piece on click to WhatsApp ads.

Separating the two also protects your numbers. One shared "follow up" bucket hides the fact that your no show branch is working while your reactivation spend is not, or the reverse.

The six numbers that tell you the process works

Measure the process, not the sequence. Six numbers are enough, and you can pull all of them per account and per setter.

1. Time to first reply, in minutes, per account and per shift. It is the input to everything else.
2. Share of threads with an open window at any moment. This is your live workload, and it is the number that tells you whether staffing matches demand.
3. Windows closed with an unanswered question. Your defect count. Target zero, investigate every one.
4. Replies per thread inside the first window. A proxy for whether the conversation is a conversation or a broadcast.
5. Booking rate per qualified thread, split by setter and by account, so you can tell a copy problem from a traffic problem.
6. Cost of re-entry, meaning paid templates and reactivation spend divided by meetings recovered. If you do not price this, it hides inside the ad budget.

Do not import someone else's benchmark for these. Compute your own over four weeks, treat that as the baseline you are trying to beat, and pair it with the reporting your clients actually see. The wider argument for treating all of this as infrastructure rather than as a set of habits is in our view of setting infrastructure for teams and agencies.

Where SetScale fits

SetScale is an AI setter built for teams and agencies rather than for one person and one inbox: several client accounts, several closers, reporting per seat, with a white label option planned. It exists for the ownership problem described above, because that is what breaks first when the same process has to run on ten accounts at once.

The product is not open yet. There is no dashboard to show, no customer count to quote, and nothing in this article depends on using it: the window rules are Meta's, and the skeleton works on paper. When it opens, a free trial of 7 days is planned, with a card required.

If your team is holding this together with a shared inbox and a lot of memory, Join the waitlist.

FAQ

How many follow up messages should a sales follow up process have? Ask instead how many windows you get. Inside a single 24 hour window, four or five short messages is normal and none of them needs permission. Outside it, you get one manual human reply under the Human Agent tag within 7 days, then you are into paid re-entry. Counting messages without counting windows produces sequences that cannot run.

Can we automate a day 3 follow up in Instagram DMs? Not as a promotional message. Once the 24 hour standard messaging window has closed, only message tags apply, they carry non promotional content only, and the Human Agent tag is explicitly for a manual reply by a person. One time notifications, which would be the closest fit, are not available on the Instagram Messaging API.

Does a lead replying reopen the window? Yes, and this is the most useful mechanic in the whole process. A message from the person opens a fresh 24 hour window, which is why the last message you send inside a window should be designed to earn a reply rather than to push a booking.

What is the difference between the standard messaging window and the customer service window? The same 24 hour idea on two platforms, standard messaging window on Messenger and Instagram, customer service window on WhatsApp. The difference is what happens afterwards: WhatsApp lets you send an approved template for a fee, Messenger and Instagram give you tags and the human agent path.

Should the setter or the closer own the follow up after a no show? The setter, until a new slot is confirmed. The thread is a setting problem, not a selling one, and passing a no show straight to a closer is how threads end up with two owners and no answer.

Conclusion

Most DM follow up sequences underperform because they were written in days while the channel operates in windows, and half their steps were never allowed to land.

Rebuild the process around three zones, put one visible owner on every thread, log every window that closes with an unanswered question, and price your re-entry instead of hiding it in the ad budget. That much works with four setters and a spreadsheet. If you are building the version that has to run across ten client accounts without anyone holding it in their head, Join the waitlist.