# Instagram DM keyword triggers

> An Instagram DM keyword trigger sends an automatic reply when an incoming direct message contains a word you have chosen, such as PRICE or LINK. It runs on the official Instagram API and can only reply within 24 hours of the person's last message.

Updated 2026-10-09. Source: https://dmpilot.pro/use-cases/dm-keyword-trigger

## Steps

1. **List the questions you already answer** — Scroll your last two hundred DMs and write down every question you have answered more than three times. That list is your keyword set, derived from real messages rather than guesses.
2. **Assign one keyword per answer** — Map each recurring question to a single trigger word. One word, no punctuation, no accents. Resist adding six synonyms per rule on day one, since overlap causes confusing behaviour later.
3. **Order the rules from specific to general** — Most tools evaluate rules top to bottom and stop at the first match. Put narrow keywords above broad ones so the specific answer wins. A rule matching "price list" must sit above the rule matching "price".
4. **Write the fallback** — Decide what happens when a message matches nothing at all. This is the majority case for most accounts, and leaving it empty is the most common setup mistake.
5. **Test the near-misses, not the exact hits** — Send yourself the messages you expect to fail: plurals, typos, the keyword inside a longer sentence, the keyword in the middle of an unrelated word. The exact match will always work. The edges are where the setup breaks.

## What a keyword trigger does

Someone sends you a DM. The message text is checked against a list of keywords you have set. If one matches, a pre-written reply goes out immediately. If none match, a fallback runs or nothing happens, depending on how you configured it. That is the entire mechanism, and its simplicity is both its strength and where most of the problems come from.

This is different from [comment to DM](https://dmpilot.pro/use-cases/comment-to-dm), which starts from a public comment on your content. Keyword triggers work on the inbox itself, on people who have already chosen to message you. That makes them a support and qualification tool as much as a marketing one.

Because the person messaged you first, the 24-hour window is already open when the trigger fires. Once that day passes without another message from them, automated replies stop until they write again.

## Choosing keywords people will actually type

The best keywords are words that only appear when someone wants the specific thing. "PRICE" is good because it is rarely typed casually. "INFO" is worse, because it shows up in half of all vague messages and will fire the same canned reply at people asking about completely different things.

Short beats clever. A keyword you have to explain is a keyword people will mistype. Single words survive autocorrect better than phrases, and a word with no accents and no apostrophes will match the same way regardless of which keyboard someone is using.

Think about where the keyword is taught. If you tell people in a story to "reply BOOK", they will type it exactly. A keyword that only exists in your settings and nowhere in your content will never be typed, and your fallback becomes the only rule that matters.

- Prefer nouns tied to one intent: PRICE, BOOK, GUIDE, SIZE, REFUND
- Avoid generic connectors: INFO, HELP, HI, YES, MORE
- Avoid words that appear inside other words you care about
- Teach every keyword somewhere public, in a story, caption, or pinned comment

## Where matching goes wrong

Almost every complaint about keyword automation traces back to a matching assumption that turned out to be false. Most tools do a case-insensitive substring check: they lowercase the message and look for your keyword anywhere inside it. That behaviour is convenient right up until it is not. DMPilot matches whole words rather than substrings, so "art" never fires on "start", but it also means "price" does not catch "prices": a second word needs its own automation.

### Partial words

A substring match on "art" fires on "start", "party", and "smart". A trigger on "buy" fires on "buying" but also on the name of anyone called Buys. If a keyword is short or is a common letter sequence, assume it will match somewhere you did not intend and pick a longer word instead.

### Plurals and variants

Setting "price" catches "prices" because the singular sits inside the plural. Setting "prices" does not catch "price". Always register the shorter form, and add spellings that genuinely differ, such as color and colour, as separate keywords rather than hoping one covers both.

### Case and punctuation

Matching is normally case-insensitive, so BOOK, Book and book behave identically, and you do not need to enter all three. Punctuation and emoji attached to the word are usually fine because the substring is still present, but leading or trailing spaces stored in the keyword itself can silently break every match. Trim them.

### The negation problem

Keyword matching has no understanding of meaning. "I do not want the price list, I already bought" contains the word price and will trigger your pricing reply. There is no configuration that fixes this. It is an argument for keeping the automated reply useful rather than pushy, so that a wrong fire is mildly redundant instead of actively annoying.

## Ordering rules when keywords overlap

Once you have more than three or four rules, overlap is inevitable, and the order they are evaluated in becomes the thing that determines your automation's behaviour. The standard model is first match wins: the tool walks the list from top to bottom and stops at the first rule whose keyword appears in the message.

The rule that follows is simple. Specific goes above general. If you have a rule for "refund policy" and a separate rule for "policy", the refund rule must sit higher or it will never fire, because "policy" will always match first. Sort your list roughly by how narrow each keyword is and re-check the order every time you add a rule.

Watch out for the broad keyword near the top that quietly absorbs messages meant for rules below it. If a rule stops firing after you added a new one, the new one is almost certainly sitting above it.

## What to do when nothing matches

For most accounts the majority of incoming DMs will not contain any of your keywords. People write in their own words, they say hello first, they send a voice note or a photo with no text at all. Your fallback is therefore not an edge case. It is the most-used message in the whole setup.

A good fallback does three things: it confirms the message arrived, it sets an honest expectation about when a human will reply, and it offers the keywords as a shortcut for the common questions. That last part is what makes the whole system improve over time, because it teaches people the vocabulary while they are already in the inbox.

The alternative, silence, is a defensible choice for accounts with low volume where you answer everything personally within the hour. What does not work is a fallback that pretends to be a person or that asks the sender to rephrase. Both waste the sender's time and neither gets you closer to an answer. If you use a fallback at all, apply a cooldown so someone sending four messages in a row does not receive four identical auto-replies. See [Instagram auto reply](https://dmpilot.pro/use-cases/instagram-auto-reply) for more patterns.

- Acknowledge, do not apologise: "Got your message, I usually reply within a day"
- Offer the shortcuts: "For pricing reply PRICE, to book reply BOOK"
- Send it once per person per cooldown period, never per message
- Never imply a human is typing when one is not

## Knowing when to hand off to a human

Keyword automation is good at the first message and bad at the second. It answers the question that has one answer. The moment a conversation has context, exceptions, or emotion in it, an automated reply stops helping and starts costing you something.

The clearest handoff signals are frustration, complaint language, a question about someone's specific order or booking, and anything involving money that has already changed hands. Automation should also step back when the same person has triggered two or three replies in a row, because that pattern usually means the answers are not landing and the person is trying to reach you rather than a menu.

In practice, that means using keyword triggers to buy time and deliver the easy answers, then reading the inbox yourself daily. Automation that reduces your DM load by half is a success. Automation that answers everything is a support experience nobody enjoys, and it is not what these tools are for. Support-specific patterns are covered in [customer support deflection](https://dmpilot.pro/use-cases/customer-support-deflection).

## Limits and cost

Two constraints shape what you can build. The 24-hour window means an automated reply is only possible within a day of the person's last message, so no keyword rule can be used to revive an old conversation. And Instagram's hourly API allowance, scaled to your audience size, means a very busy hour can run into the limit.

DMPilot enforces the window for you, backs off and retries a send that hits the allowance, and applies a 24-hour per-contact cooldown. It is free for up to 100 people a month, then $9 a month flat regardless of how many people are in your inbox.

## FAQ

**Are Instagram DM keywords case sensitive?**

No, matching is case-insensitive, so BOOK, Book and book all trigger the same rule. You only need to enter the keyword once, in whichever case you find easiest to read.

**Does the keyword have to be the whole message?**

It depends on the match you pick. In DMPilot, exact-word matching needs the message to be just the keyword, ignoring case, punctuation and emoji, while anywhere-in-the-text matching also fires on "hey, whats the PRICE for this?". DMPilot matches whole words, so price never fires inside priceless; tools that match raw substrings can.

**What happens if a DM matches two keywords?**

The first matching rule in your list wins and the rest are skipped. Order your rules from most specific to most general so the narrower answer is the one that fires.

**Can I auto-reply to DMs from people who never followed me?**

Yes, as long as they messaged you first. Instagram permits replies to anyone who has contacted you, but you can never start a conversation with someone who has not.

**How many DMs can you send a day?**

There is no single published number. The API caps calls per hour scaled to your audience size, so a small account has a smaller allowance, and tools apply their own daily ceilings under that. In practice pacing matters more than any single number.

**Should every account use a fallback message?**

Not necessarily. High-volume accounts benefit from one because most messages will not match a keyword, but if you personally answer every DM within a few hours, silence can read better than an automated holding message.

**Can I use emoji as a keyword?**

Not in DMPilot: a keyword is letters and digits, and emoji are ignored when a message is matched. Emoji make unreliable triggers anyway, because people rarely send exactly the emoji you expected.

## Related

- https://dmpilot.pro/use-cases/instagram-auto-reply
- https://dmpilot.pro/use-cases/comment-to-dm
- https://dmpilot.pro/use-cases/customer-support-deflection
- https://dmpilot.pro/how-to/automate-instagram-dms
- https://dmpilot.pro/guides/instagram-dm-limits