Small-business chatbot marketing is not a 40-node tree. It is one offer people already ask for, delivered in the inbox they already use. We tested these plays the same way we scored the tool posts: write the brief, run it on a real account or a second phone, see what breaks.
Full disclosure: this is InstantDM's blog. InstantDM is the Meta row. If your questions arrive as Instagram comments, that is our product. If they arrive on a website bubble, buy Tidio. Weigh that.
TL;DR
- Local / services: Comment BOOK → city and time → calendar
- Shop: Comment LINK → product + code → one follow-up
- Coach / course: story reply JOIN → one qualifier → checkout or calendar
- Lead magnet: Comment GUIDE → PDF → email
- Instagram or Facebook: InstantDM. Website only: Tidio. WhatsApp required: ManyChat.
Free on InstantDM is 500 automations a month plus a 7-day full trial. Paid is $9.99 or $24.99 flat (instantdm.com/pricing, August 2026).
What chatbot marketing is (and isn't) for a small team
It is automated conversation used to sell or book. It is not a personality bot. It is not a replacement for a person on a $3,000 package. It is not an "AI agent" that runs the company while you sleep.
If last month's inbox was "how much?" and "link?", you already have a campaign. You just answer it slowly. The strategy is to write down those three questions, pick one offer, and let the first reply go out in seconds.
We keep saying "one offer" because small accounts die on option menus. Four buttons in the first DM looks busy. It also delays the thing they commented for. Send the thing. Then ask one question if you still need it.
How we tested these strategies
We did not invent a framework and call it research. We ran five briefs on Instagram plus a site widget, the same week we scored the tool posts (August 2026).
- Comment-to-DM with a single keyword and a link in message one.
- Story reply JOIN with a calendar in the first reply.
- A Click-to-Message test ad only after the organic flow worked.
- A product-page widget for size and shipping, separate from Instagram.
- One follow-up a few hours later, then stop.
What failed: two extra qualifier questions before the PDF. A Flow that asked the same thing the caption already answered. Putting the website widget on Instagram comments, which it cannot see.
What worked: caption names the word, first DM is the asset, one human-readable follow-up. That is the whole strategy list below, written as plays you can run this week.
Strategy 1: Comment-to-DM on the post you would have pinned anyway
This is the default small-business play. You already post. People already ask for the link or the price in comments. The caption tells them the word. The tool sends the asset in a DM.
Write the caption last. Write the DM first. If the first message is not useful without the rest of the flow, rewrite it. "Here's the menu" beats "Hi! I'm a bot, tap a button to begin."
InstantDM does this on Instagram and Facebook. Official Meta Business Partner, official Instagram Graph API. Keyword trigger or any-comment trigger. Collect email or phone in the thread if you need a list. Human handoff when the question is not the keyword.
Test from a second phone before you put the word on a Reel that already has reach. If the DM is empty or late, fix that, then scale the post.
Skip this if you have no offer you can send as a link, a PDF, or a calendar. A bot that says "thanks for commenting" is not a strategy.
Strategy 2: Story reply as the opt-in
A story reply is higher intent than a like. JOIN, MENU, or TIMES. First DM is the workshop link or the booking page. One qualifier only if the slot is scarce: "Weekday or Saturday?"
Stories die in 24 hours. That is a feature. You can change the offer tomorrow without editing a permanent post. We like this for coaches, salons, and anyone whose calendar is the product.
InstantDM story reply automation is the Meta way to do this. Do not scrape story viewers and cold-DM them. That is a different, worse play, and it is how accounts get restricted.
Strategy 3: Same offer on a Click-to-Message ad, only after organic works
Ads that open a chat are not a first-week strategy. Run the organic comment or story flow until the first message looks right and the handoff works. Then spend $20–50 a day on the same first message.
InstantDM ads automation sits on Trendsetter at $24.99/month. ManyChat can take the thread if the ad is wired to them. A website chatbot does not see the tap. Setup notes: Facebook Messenger advertising chat tools.
Judge the ad on conversations and booked calls, not landing-page CTR. You did not buy a landing page.
Strategy 4: Store widget for size, shipping, and "is this in stock"
If the question happens on the product page, the bot has to live there. Tidio or Shopify Inbox. Chatbase if the answers are already in a help doc and you do not staff a bubble. InstantDM will not put a bubble on your site. We have said that in every vs post. It is still true.
Do not buy Tidio because last month's leads were Reel comments. Do not skip the widget because Instagram is working. Two inboxes, two tools, if both are busy. That stack is still cheaper than one confused subscription. See Tidio vs ManyChat and live chat for leads.
Strategy 5: One follow-up, then stop
People get interrupted. One "still want the Tuesday slot?" a few hours later is polite. Five follow-ups is how they mute you. InstantDM follow-ups are on Trendsetter. Tidio can remind from the inbox. Write the follow-up as a human would. If you would be embarrassed to send it from your personal account, do not automate it.
Plays by business type
Local services (salon, clinic, tutor, gym). Keyword BOOK. First DM: two time windows and a map link. Qualifier: new or returning. Handoff if they want a custom slot. Do not ask for their life story before they see a time.
Product shop. Keyword LINK or the product name. First DM: the product, size chart, one code. Follow-up: "Need a size?" Widget on the site for the people who never open Instagram.
Coach or course. Story JOIN or comment APPLY. First DM: the syllabus or the application. One qualifier: budget range or start date. Calendar or checkout third.
Restaurant or venue. Keyword MENU or TABLE. First DM: today's menu or OpenTable. Do not build a full reservation bot in week one. The PDF or the booking link is the win.
Agency or freelancer. Keyword AUDIT. First DM: the three-question form or a calendar. You do not need ManyChat's six channels until a client asks for WhatsApp.
What to skip
- Multi-channel broadcasts on day one.
- A chatbot marketing agency for a single account with 80 comments a week.
- Cold DMs to people who never talked to you.
- Four qualifying questions before the PDF you promised.
- Calling InstantDM's AI comment replies an AI agent. They are comment replies.
- Buying Chatbase to run a giveaway. That is the wrong aisle. Chatbase vs ManyChat.
Tools, small-business edition
| If your questions arrive… | Start here | As of Aug 2026 |
|---|---|---|
| Instagram / Facebook comments and DMs | InstantDM | $9.99 flat after 500 free automations/mo |
| Your Shopify or website | Tidio or HubSpot Chat | Tidio pricing |
| WhatsApp as the default inbox | ManyChat or Chatfuel Light | Per contact on ManyChat; Light is free on Chatfuel |
| A help doc, no staff on chat | Chatbase | Hobby $32/mo annual, 700 credits |
Full scored aisle: top chatbot platforms for marketing campaigns.
A week of setup we would actually run
Monday: export or screenshot the last 30 questions. Circle the three repeats. Tuesday: pick one offer that answers the top repeat. Write the first DM on paper. Wednesday: connect InstantDM (or Tidio if the repeats are on the site). Thursday: test from a second phone. Read the DM out loud. Friday: put the keyword in the next caption. Do not add ads. Do not add WhatsApp. Do not hire an agency.
Week two: read transcripts. Cut any question the bot asked that a human would not. Add one follow-up if people go quiet after the link. Week three: only then think about a Click-to-Message ad or a second keyword.
Sequence shape: how to create a chat funnel. Safety: send automated DMs only as replies to comments, story replies, or inbound DMs. Stay inside Instagram's rate limits. InstantDM's DM Queue, Super Slow Mode, Rate Limit Controller, and Flood Control exist to keep you inside those lines, not around them.
What a small bill looks like
Organic Instagram only, under 500 automations: InstantDM free. Organic plus ads or follow-ups: Trendsetter $24.99. A site widget beside that: Tidio Starter $24.17 annual if volume is low. WhatsApp on the side: Chatfuel Light until you outgrow it. That is still less than one confused ManyChat contact invoice after a Reel works, if you did not need WhatsApp.
We tested and scored the tools in the vs posts. This post is the strategy those tools sit under. If you only buy a logo, you still have to write the first message.
Transcripts: the part nobody budgets
A strategy without transcripts is a caption idea. Once a week, read twenty threads. Mark three buckets: delivered and done, delivered and confused, never delivered. Confused means your first DM assumed they saw the caption. Never delivered means the keyword, the connection, or the rate limit failed.
Rewrite only the first message that caused confusion. Do not add a new flow. Small businesses grow the bot by subtraction. If a button is never tapped, delete it.
We tested this the boring way: the week we added a second qualifier, completion dropped. The week we put the link in sentence one, more people opened it. That is not a case study. It is the same pattern every small inbox shows.
Three worked weeks (illustrative, not promises)
Results depend on your offer, audience, and consistency. These are shapes, not guarantees.
Salon, 2,000 followers. Caption: comment BOOK. First DM: two windows this week. Twelve comments, nine DMs opened, four booked. Follow-up caught one more. InstantDM free tier covered it.
Shopify shop, 8,000 followers. Comment LINK on a restock Reel. First DM: product + 10% code. Widget on the site for the people who never comment. Instagram on InstantDM. Site on Tidio. Two tools. Comments did not appear in Tidio. That is expected.
Coach. Story JOIN for a workshop. First DM: syllabus and a Tuesday/Thursday choice. Ads the next week on the same first message, Trendsetter. Organic had to work first or the ad would have scaled a bad DM.
Common small-business mistakes we keep seeing
Buying ManyChat because a YouTube video said so, then never using WhatsApp. Buying Tidio because the Shopify app store ranked it, then wondering why Reels do not DM people. Building six keywords on day one. Asking for a phone number before sending the PDF you promised. Turning on any-comment triggers on a viral meme and blowing the rate limit.
Also: treating "AI" as the strategy. InstantDM AI comment replies can cover odd phrasing on comments. They do not replace a written offer. Chatbase can answer a help doc. It does not run BOOK.
Captions that make the strategy work
The bot cannot save a caption that hides the word. Put the keyword in the first two lines and again at the end. "Comment BOOK and I'll DM two times this week" is a strategy. "New availability, link in bio" is the old funnel you are trying to leave.
Do not use ten keywords on one post. People will guess wrong. One word. Whole-word match if your tool supports it, so "booking" does not fire BOOK by accident if you do not want that.
On Facebook, the same offer can sit under the post comments. InstantDM covers Facebook comments too. Same first DM. Do not write a second personality for Facebook.
When you should hire help
Hire an agency or a freelancer when you have more than one account, paid ads on, and nobody reading transcripts. Do not hire them to "set up AI." Hire them to write the first DM and watch the first hundred threads. If they will not show you the first message before they connect a tool, keep your money.
A $9.99 tool plus an hour of your own writing still beats a $2,000 bot build that asks four questions before the menu.
What we would run on our own small account
One comment keyword on the post that already gets "link?" in the comments. First DM is the link. One follow-up. InstantDM free until it is obviously working. No widget until the site has the same three questions. No ads until the organic DM looks like something we would send by hand.
ManyChat, Tidio, Chatbase, and Chatfuel are trademarks of their owners. InstantDM is not affiliated with them.
Start InstantDM free if the inbox is Instagram or Facebook. Wider map: what is chat marketing.
If you want the scored tool aisle after you have a strategy: top chatbot platforms, Tidio vs ManyChat, Chatbase vs Tidio. Strategy first. Logo second. That order is the whole point of this post. If you only remember one line: write the first DM before you connect anything. The tool cannot invent an offer you have not written.
We tested and scored the tools elsewhere so this page could stay on the plays. If a vendor blog leads with AI and never shows a first message, it is not a small-business strategy post. This one is. Use it, then pick InstantDM, Tidio, or Chatfuel based on the inbox you already have, not the homepage you liked.
First-message scripts you can copy and ruin
Ruin them on purpose. Make them sound like you. These are starting points we tested, not holy text.
BOOK: "Two openings this week: Tue 3pm or Sat 11am. Reply 1 or 2 and I'll hold it. If neither works, say a day."
LINK: "Here's the restock. Sizes in the chart on that page. Code HELLO10 if you check out today. Stuck on size? Reply your usual size."
GUIDE: "PDF is here. If you want the 12-minute walkthrough, reply VIDEO and I'll send that too."
JOIN: "Workshop is Thursday 7pm. Outline is in this note. Reply YES and I'll send the Zoom. If Thursday is dead, say another night."
None of those asks for an email first. Collect the email after they have the thing, or in the same thread with InstantDM's in-DM questions if you need a list. Asking for the phone before the PDF is how people bounce.
