Choose a keyword people can ask for naturally

A trigger keyword is the word a commenter types to set the whole sequence off. The best ones are single words a person can thumb-type without thinking: GUIDE, PRICE, LINK. Long or clever phrases lose people who mistype them, and the match only fires on the word you set.

What happens after the match, in order:

1. Someone comments the keyword on a post you selected in the campaign.
2. ReplyMagnet checks the comment against the campaign's keywords, ignoring case and matching whole words only. GUIDE fires on ‘send the guide’ but not on ‘guideline’.
3. A public reply goes under the comment, so the thread shows the request was answered.
4. A DM carries the resource link, or the capture page if you turned on email capture.
5. The download link is tracked, so you can see whether the person opened it.

If a single word feels limiting, two other modes cover the gaps. Any-comment mode responds to every comment on the post, which suits a giveaway or a post that exists only to hand something out. Smart matching, covered below, catches clear requests that never contain the word.

Platform rules that shape every keyword campaign:

• The connected account must be a Business or Creator account. Personal accounts cannot use the comment and messaging API.
• A private reply to a comment must be sent within 7 days of the comment. Once a DM conversation is open, further messages fall under Instagram's 24-hour messaging window.
• Overflow requests queue on every plan. A queued request that passes the window expires instead of sending late.
• Instagram does not allow messaging people who never interacted with you. Every DM in this flow answers a comment the person left themselves.
• ReplyMagnet runs on the Instagram Graph API that Meta reviewed through App Review, so nothing here relies on a browser extension or a shared password.

‘Comment RECIPE for the ingredient list’ makes sense without another explanation. ‘Comment YES’ depends much more on the surrounding conversation.

A specific word also helps later, when you review the campaign. RECIPE tells you what the person requested; YES tells you very little.

Try a keyword campaign with an existing post before changing every caption.

The keyword is yours to choose, but the ones that fire best describe what the commenter receives. Match the word to the post's job, then check that the attached resource actually does that job.

Keep it to a single word where you can. A commenter types it once, on a phone, often while half-watching a reel, so GUIDE beats SENDMETHEGUIDE every time.

These suggestions only work when the attached resource fulfils the corresponding promise.

The failure mode of a bad keyword is not that it never fires. It fires too often. A keyword like THIS or YES matches ordinary praise, so a post full of ‘love this’ comments triggers DMs to people who only meant to compliment you, and each one spends an engagement from your quota.

Run every candidate through this checklist:

• Does the word appear in ordinary praise? LOVE, THIS, YES, WOW, AMAZING and FIRE all do.
• Is it your own name or handle? People tag and mention you in comments that are not requests.
• Is it a reaction word such as ME or SAME? ME works on a giveaway post where every comment is an entry, and almost nowhere else.
• Could it appear inside a normal sentence about the post's topic? GUIDE under a post about tour guides is a poor fit.
• Is it more than one word? Phrases are mistyped more often, and a second variant is a better answer than a phrase.
• Does it name what the person gets? GUIDE, PRICE and LINK pass; CLICK and GO do not.

If a word fails two of these, choose another. On the free plan, with 1,000 engagements and one live campaign, an accidental match costs a larger share of the month than it does on a paid plan.

For a recipe sheet, RECIPE and INGREDIENTS might reasonably mean the same request. PRICE and FLOORPLAN probably don't, unless the promised document clearly includes both.

Don't turn one campaign into a bag of unrelated words. The system cannot infer a different attached file from each variant.

ReplyMagnet's campaign limit is up to 25 keyword variants per campaign. That is capacity, not a suggestion to use all of them. Start with the words the caption and actual requests justify.

Test three comments:

1. The keyword by itself: ‘GUIDE’.
2. A natural request: ‘Could you send the guide?’
3. An unrelated sentence containing similar text: ‘Those guidelines are helpful.’

Also inspect the selected posts. GUIDE might be appropriate on a tutorial and confusing under a discussion about tour guides. Post scope is part of keyword design. The setup guide explains that control.

Not everyone comments the word. Plenty type ‘can I get this please’ or ‘how do I download it’ instead, and an exact-keyword campaign never sees them. Smart matching is for exactly those comments.

An AI classifier runs only when every keyword check comes back empty, and only when a smart campaign is armed on the post. An exact keyword match never spends a check.

Which mode to arm, as a decision rule:

• Exact keyword: the caption tells people what to type and you want zero accidental matches.
• Keyword plus variants: real comments show two or three natural spellings of the same request.
• Smart matching: the post draws requests in many phrasings, and you would rather miss a few than DM the wrong people.
• Any-comment mode: every comment is a request by design, such as a giveaway entry post.

Smart matching is on every plan, with a daily cap by plan: 100 checks on the free plan, 300 on Basic, 750 on Pro and 2,000 on Agency. It fails closed, so an uncertain read sends nothing rather than DMing someone who did not ask.

The 30-day trial needs no card. That is long enough to arm a keyword campaign and a smart campaign on one post and watch both fire.

AI matching can help with requests that omit the exact word, but it has a daily allowance and should remain selective. An unrelated compliment is not a missed lead.

Keep a short note of the wording that caused confusion. Change one thing, then repeat the request and non-request tests.

Use campaign activity to inspect what happened rather than judging the keyword only by the total reply count.