CommunityDesk

Writing

Run the keyword play without losing your comment section

Comment a word, get a DM. It works, and it buries every real comment on the post underneath the entries. CommunityDesk marks the comments that are nothing but the keyword, hides them from the thread by default, and never hides them from your counts.

August 22, 2026

You know the play. "Comment RECIPE and I'll send it to you." It is the most reliable engagement mechanic on Instagram, it is the pattern most creators know ManyChat for, and it works.

It also does something nobody mentions. A post running that play fills with one word. RECIPE. recipe. RECIPE!! Hundreds of them, sometimes thousands, all identical, all from real people who did exactly what you asked. Your comment section is now a receipt printer.

And the thing about a receipt printer is that the actual comments are still in there. Someone asked whether it works for beginners. Someone said they have been following for two years and finally bought. Someone reported a bug. Those comments are sitting between entry number 340 and entry number 341, and you will never see them, because scrolling past nine hundred people typing one word is not something a human does twice.

So the campaign worked and the conversation died. That is the trade every tool in this category asks you to make, and it is not necessary.

What CommunityDesk does

When a campaign is running, CommunityDesk marks the comments that are nothing but the keyword. Those are entries. They are counted, they get your DM, they go in your list. What they are not is conversation, so the post's thread hides them by default.

What is left is your comment section, at its real size, with the questions and the praise and the bug report where you can actually see them.

One line at the top of the thread says what is missing, in the product's own words: "1,204 keyword comments hidden", with a Show beside it. Click it and they come back, and the line changes to "Showing keyword comments" with Hide them. The count is the disclosure. A creator looking at 180 comments on a post they know drew twenty-seven thousand has to be able to see where the rest went, and that CommunityDesk still has them.

The word "only" is the whole feature

A comment that says RECIPE is a lead.

A comment that says RECIPE, does this work with a stand mixer? is a person asking you a question.

CommunityDesk never hides the second one. The mark is set when the entry is captured, on comments that contain the keyword and nothing else, and it lives in the data rather than in a filter that guesses later. A tool that hid every comment containing the word would hide that question. It would also hide thirty people saying "happy birthday" from a campaign whose keyword happened to be BIRTHDAY.

That precision is what makes hiding them safe enough to do by default. Without it the safe choice would be to show everything, which is where you started.

Hidden from the reading, never from the counting

This matters and it is worth being exact.

Hiding is a reading decision. Every entry is still stored, still counted, still in your campaign list with the person attached, still in your export. The number of people who entered is the number of people who entered. Nothing about the filter touches what you got.

You are not choosing between seeing your leads and seeing your comments. You get both, in the place each one belongs: entries in the campaign, conversation in the thread.

What it costs

Nothing, and that is deliberate.

Campaigns are not metered on any CommunityDesk plan. There is no per-contact charge, no campaign tier, and no allowance to run out of on the day a post takes off. A comment that is only your keyword does not count against your plan's interactions either, because reading one word is not the work you are paying for.

Tools that meter contacts charge you most on your best day. That is the wrong shape for a mechanic whose entire purpose is volume.

What this is not

CommunityDesk sends one message per person, to people who asked for it. There is no flow builder, no branching, and no drip sequence. If the DM needs to hold a conversation after the first message, that is a different product and CommunityDesk does not pretend otherwise. The comparison with ManyChat says the same thing at more length, including when they are the right pick.

The keyword play is not the interesting part of your comment section. It is the part that makes the interesting part unreadable. CommunityDesk runs it, then hands the comment section back.