The comment-to-DM funnel as a product: set it up for a creator and charge monthly

Start with the number that decides your whole pricing conversation.

Across 39 Fiverr gigs advertising ManyChat build work, read on 12 September 2026, the starting prices ran $5 to $95. The single most common price was $30, on 14 of the 39. Of the 13 whose titles name Instagram DM automation specifically, the band was $10 to $50, clustered at $20 to $30.

Not one of the 39 offered a monthly plan.

So the build is a commodity and everyone knows it. If you want a retainer, you are not selling the build. You are selling everything that happens after it, and you will be better at that than almost everyone if you get two things right that this niche routinely states wrong: how Meta's private reply window actually works, and what happens when you point a funnel at an account with no reach.

I am going to be direct about the second one, because it is the thing that makes the rest of this credible. My own funnel is built properly and it has barely ever fired. The numbers are further down.

What you'll have when you're done

• A funnel built the way Meta's rules actually allow, not the way the tutorials draw it
• The message sequence that opens the 24-hour window instead of wasting the one message you get
• A cost to serve per client, in dollars, from published pricing
• A qualification test that tells you when to turn a client down
• A monthly report that justifies a monthly invoice
• [SCREENSHOT: a live flow canvas showing the trigger, the reply, and the opt-in branch]

Before you start

• An Instagram professional account with a linked Facebook Page. Meta requires the Page for comment-to-DM.
• A tool. ManyChat and Metricool Flows both do this. Pricing is compared below; pick after you read it.
• A client with comment volume. Not followers. Comments. This is the qualification, and it is covered in its own section for a reason.
• Two hours for the build. The rest of the week is pricing and the pitch.

The rule everyone gets wrong

Almost every guide on this draws the flow as: someone comments the keyword, you DM them the link, done.

That is not what Meta allows, and building it that way is why flows underperform for reasons the operator never diagnoses.

Here is what Meta's private replies documentation actually says. Private replies let you send a single message to someone who commented on your post, ads post, reel or live story. Only one message can be sent. It must be sent within 7 days from when the comment was created, for comments on a post, ads post or reel. And only when the person responds to that private message can you continue the conversation, within the 24-hour messaging window.

Read that again with a commercial eye:

So a flow that spends its one message on a bare link has nothing left and no window. A flow that spends it on a question the person will answer has just bought 24 hours in which links and follow-ups are allowed.

Metricool's own documentation describes the same mechanism from the other side: asking the user to react or respond acts as a double opt-in and opens Meta's 24-hour window to send additional content and links.

That is the entire reason working flows open with "reply YES" rather than dumping the link. It is not a growth-hack. It is the API.

Step 1: Build the four-step flow

Four steps, in this order. Every one of them exists because of a rule above.

1. The public comment reply. A short visible reply to the comment itself. This does two jobs: it confirms to that person that something happened, and it shows everyone else scrolling the comments that the keyword works. Keep it under the character limit your tool enforces and never put a link in it, because links in public replies are not clickable anyway.

Sent! Check your DMs.

2. The one private message, which asks rather than tells. This is the single message Meta gives you. It must earn a reply.

Hey! You commented FUNNEL so I've got the guide ready to send.

Quick thing first: reply YES and I'll drop the link straight in here.

The reason it works is not psychology. It is that the reply is the thing that opens the window.

3. The delivery, inside the 24 hours. Now links are allowed. Send it immediately and keep the promise before you ask for anything.

Here it is: [your link]

That's the full build, free, no signup. Takes about 20 minutes.

4. The ask, last. Only after the promise is kept. Email capture, a tag, a next step. Tag by keyword so the contact list becomes a per-post conversion report for free.

Step 2: Know the limits before a client asks

Meta publishes the ceilings, so you can answer this from documentation rather than vibes.

Seven hundred and fifty an hour is the real operational ceiling for comment-to-DM, and it is the only number in this space with a genuine published unit. You will not hit it on creator work. Knowing it means you can stop worrying about throughput and start worrying about the thing that actually limits this business.

Two more from Meta's policy documentation, both worth a line in your client contract. One-time notifications and sponsored messages are not available for the Instagram messaging API, so do not promise broadcast re-engagement you cannot deliver. And where applicable law requires it, automated chat experiences must disclose that the person is interacting with an automated service, at the beginning of a conversation, after a significant lapse of time, or when a chat moves from human to automated.

Step 3: Work out your cost to serve

This is what makes a monthly price honest rather than made up. All figures from each vendor's own pricing page, read 12 September 2026.

ManyChat

Note the page defaults to annual billing, so the number you remember is probably the discounted one.

The billing unit is the active contact, not the message. ManyChat defines it as a real person you interact with during a monthly billing period, whether through messages or automations, and says it does not matter how many messages you exchange: one person equals one active contact for the month. Going over does not switch automations off; it charges an overage per extra contact.

Two things that catch people. The free plan caps at 4 active automations, which is one keyword per post for four posts, so it is a demo tier and not a client tier. And API access is a Pro capability according to ManyChat's own help centre articles; it is not mentioned on the Essential plan page at all.

Metricool

Flows is included from Starter upward. In Metricool's own comparison table every Flows row reads the same thing in the Starter, Advanced and Custom columns: no extra cost for a limited time. The Free column is blank.

That wording is the whole truth about what Flows costs. Metricool has not published a price for it. Do not tell a client it is free forever, and do not build a margin on the assumption.

Metricool's Flows prerequisites are worth copying into your onboarding: a connected professional Instagram account, the Edit Flows permission, and for comment-to-DM a Facebook Page linked to the Instagram account. Flows also does not fire on Close Friends replies or story likes; only text and emojis trigger it.

The others, for comparison

What that means for pricing

Your tool cost per client sits somewhere between $20 and $39 a month for a creator-sized account, and the structure is per active contact, not per send. Put the subscription on the client's own card. Then your cost to serve is your time, your margin is not eroded by their success, and you never have to explain an overage line to someone who does not know what an active contact is.

The honest part: what happens on a small account

Here is my own data, and it is not flattering.

I have 15 Instagram keywords and 15 TikTok flows live on my own account. They are built correctly, scoped to any post rather than pinned to one, and tested.

Exactly two automations have ever fired. One send each, on two keywords. Both of those sends were clicked, which tells you the copy is fine. TikTok is zero across the board.

The flows are not broken. There is almost no comment volume to trigger them. A comment-to-DM funnel is a multiplier on reach, and it cannot create reach. Multiply a small number and you get a small number.

I am telling you this because it is the single most useful thing in this guide for anyone trying to sell it. Every competitor pitching this product is implying it generates leads. It does not. It converts attention that already exists, very efficiently, into something you own. Say that out loud in the sales conversation and you will be the only person in the room the client believes.

It also gives you a qualification test, which is the next section, and turning down the wrong client is worth more to a service business than winning them.

Step 4: Qualify on comments, not followers

Follower count is the wrong number and it is the one every prospect will lead with.

A rough working line: if the last ten posts averaged fewer than about ten comments each, the funnel will fire a handful of times a month and the client will cancel in month three, correctly. Tell them so, and tell them what to fix first. That conversation loses you a bad month and wins you a client later.

Step 5: Price the part that is not the build

The build is $30 on Fiverr. Do not compete there.

What justifies a monthly invoice is the work that recurs, and it is real:

• One keyword per post, chosen per post. This is the difference between a funnel and a novelty, and it is a weekly decision about what each piece of content should route to.
• Copy that gets rewritten when it underperforms. The opt-in message is the whole conversion point. It is worth testing every month.
• Window compliance. Somebody has to make sure the flow is not trying to send outside the 7-day and 24-hour rules, and that the disclosure line is where the law needs it.
• Contact-count management. Watching the active-contact number against the plan and moving tiers before an overage instead of after.
• The monthly report. Comments in, opt-ins, delivered links, clicks, emails captured, per keyword. This is what a retainer buys and it is what makes renewal a non-conversation.

What the market says, honestly

I could not verify a single current retainer figure for this service, and I looked. The bands circulating on the web all trace back to tool vendors' own blog posts with no source behind them, which is exactly how invented numbers get into circulation.

The two ends I can cite:

There is nothing verified in between, and I am not going to fill the gap with a number I made up. What that actually means is good news for you: the commodity floor is a one-off build at $30, nobody is selling this as a service on the open marketplaces, and the only people ever documented charging monthly were charging four figures. The middle of that market is empty because it is a positioning question, not a price-discovery one.

So build your number from the bottom up: your tool cost per client, your hours per month on the list above, your rate. That number is defensible in a meeting. A number copied off a blog is not.

What I would do differently

Sell the report before the build. The build is the thing they think they are buying and the report is the thing that makes them keep paying. Show a sample report in the pitch, filled in with plausible numbers labelled as a sample, and the monthly price stops sounding like a subscription to a tool.

Put the tool on their card, always. It removes the overage conversation, it removes the awkward moment when they succeed and your cost goes up, and it means they own their contact list, which is the right answer ethically as well as commercially.

Say the multiplier line in the first meeting. A funnel converts attention you already have. If they do not have it yet, tell them. You will lose the deal and gain the reputation, and in a market where every competitor is implying lead generation, the reputation is the product.

Write the four-step flow once and reuse it. The copy above works because it is shaped by Meta's rules rather than by a template. It does not need reinventing per client; only the keyword, the offer and the opt-in line change.