If your business takes enquiries through Instagram or Facebook messages and you want ManyChat set up properly, or you already pay for it and are not sure it is doing its job, this is the service for you. I build ManyChat flows that answer the common questions straight away, collect what you need to quote or book, and pass the conversation to a person when it needs one.
ManyChat is a chat automation tool that connects to Instagram and Facebook Messenger through Meta’s official messaging tools. It reacts to something a person does, such as sending a keyword, commenting on a post or replying to a Story, and runs a scripted conversation in your direct messages. It is not a website chatbot and it is not a CRM. A good build treats it as the front door to those systems, not a replacement for them.
What a good ManyChat build changes
The first change is speed. A message sent at 10pm on a Sunday gets an answer and a clear next step, instead of sitting unread until Monday and going cold. The second is consistency: every enquirer is asked the same qualifying questions, so whoever picks up the conversation has the details in front of them rather than starting from scratch.
The third change is that conversations become data. A comment on a Reel asking for your price list can turn into a contact with a tag showing what they asked about, their email address if they chose to give it, and a note of the post that brought them in. That is what lets you judge whether Instagram produces customers or only likes.
Who it suits, and who it does not
It suits UK businesses whose customers already message them: salons, aesthetics and beauty clinics, personal trainers and gyms, restaurants taking group bookings, event venues, course creators, independent shops, and trades who show their work on Instagram. It also suits anyone running click-to-Messenger and Instagram message ads who needs replies faster than a person can manage.
It is a poorer fit when:
- few enquiries arrive by message, so there is little to automate;
- most questions need a person’s judgement, such as a quote that depends on photos and a site visit, where a short acknowledgement and a fast handover beat a long scripted flow;
- you want a bot that answers anything using your website and documents, which is a job for an AI chatbot rather than a scripted flow;
- you hope to message people who have never contacted you. Meta does not allow businesses to start conversations on Instagram or Messenger, and no tool gets around that.
What ManyChat does well, and where it stops
ManyChat is good at triggers and branching. A conversation can start from a keyword sent by DM, a comment on a chosen post, a Story reply or mention, a link or QR code, or a click-to-message ad. Buttons and quick replies let people tap instead of type, answers are saved to custom fields, contacts are tagged by interest, and a live chat inbox lets your team take over any conversation.
Comment-to-DM is the feature most businesses come for. Someone comments a word such as “menu” under a post, and ManyChat sends them the link privately, which saves them hunting through your bio for it. Pair it with a genuinely useful lead magnet, such as a price guide or a booking link, and your existing posts start real conversations instead of collecting comments.
Where it stops: ManyChat is not where your customer records should live. Its contact list only holds people who have interacted with you through Messenger or Instagram, and it has no sales pipeline, quoting or invoicing. Its AI features can recognise what someone means and draft replies, but they need tight limits and testing before they speak to customers on your behalf.
Meta’s messaging rules shape every build
Most disappointing ManyChat setups were planned without these rules in mind. Meta sets them, not ManyChat, so they apply whichever tool you use.
- The customer opens the conversation. Every flow begins with something the person did: a message, a comment, a Story reply or a tap on an ad.
- The 24-hour window. For 24 hours after a person’s last message you can send them messages, including promotional ones. After that, what you can send automatically is tightly restricted.
- Human replies get longer. Meta’s human agent tag lets a member of your team reply up to seven days after the customer’s last message, for genuine one-to-one support only, never for automated promotions.
- One private reply per comment. Comment-to-DM sends a single private reply. The conversation only continues if the person answers.
- Marketing outside the window needs explicit opt-in. Messenger offers a route for recurring marketing messages that people choose to subscribe to, and what is available on Instagram is narrower. Meta revises these features often, so I check the current position before planning any campaign that depends on them.
Ignoring the rules leads to restricted messaging or a restricted account, which costs far more than the automation saves. A build designed around the window does its useful work while the person is engaged: answer, qualify, ask permission for an email address or phone number, and hand over.
UK data protection is part of the build
The names, Instagram handles, phone numbers and email addresses ManyChat collects are personal data under UK GDPR. Your privacy notice should say that you use ManyChat for messaging, what you collect through it and why. ManyChat is based in the United States, so you need its data processing agreement in place, and your records should cover the transfer of personal data outside the UK.
Two further points catch UK businesses out. If you plan to send marketing emails or texts to people you met in DMs, PECR applies: you need their consent or, for existing customers, the soft opt-in. The flow should ask clearly, with a separate opt-in button rather than consent buried in another answer. And clinics, pharmacies and anyone in health should keep their flows away from health details, which are special category data and need far stronger justification than convenience.
What a build covers
Entry points
Which comments, keywords, Story replies, ads, links and QR codes start a conversation, and what each one promises. Every entry point should deliver what its caption said within one message.
Conversation flows
A welcome, answers to the questions you get every week (opening hours, prices, location, delivery), qualifying questions, a booking or quote request, and an out-of-hours reply. Written in your voice, kept short, with a visible way to reach a person at every step.
Fields and tags
Custom fields for the details your sales process needs, and tags for interest and source, so you can see which posts and ads start conversations that go somewhere.
Handover to your team
Rules for when a person takes over, notifications to whoever is on duty, and automation paused while a person is talking so the bot never interrupts. Keeping a human in the loop is the safer design for anything involving money, health or complaints.
Connections to your other tools
Contacts sent to your CRM, a Google Sheet, Shopify or your email platform through ManyChat’s own integrations, or through a webhook to Zapier, Make or your own system where no direct connection exists. Consent and source travel with each contact.
AI replies, where they fit
If you want free-text questions answered, I set up ManyChat’s AI features with a narrow brief, approved answers to draw on and a fallback to a person. This is optional, not the default.
How I run it
- Discovery call. We look at where messages come from now, what people ask, what a good outcome is (a booking, a quote request, a sale) and who answers today.
- Flow map. I map every entry point and branch on one page before anything is built, and we agree the wording.
- Build in your account. ManyChat is connected to your own Facebook Page and Instagram professional account, on your own subscription, so you own everything.
- Test. I test each trigger from a separate account on a phone, including the out-of-hours path, the handover and every integration, and check the data arrives where it should.
- Launch and adjust. After launch I review conversations where people dropped out or asked something the flow did not expect, and refine it.
Problems that come up most often
- Comment triggers set to fire on any comment, so a friend’s “congratulations” gets a sales message.
- Flows ten questions deep before a person is offered, which most people abandon long before the end.
- Bots that keep messaging while a member of staff is mid-conversation.
- Email addresses that sit in ManyChat and never reach the CRM or email platform, or arrive with no record of consent.
- Tags created ad hoc by several people until nobody knows what “lead2” means.
- Accounts built on a freelancer’s own ManyChat login, so the business cannot see or change its own flows.
What you receive
- The flow map and the agreed copy for every flow.
- The working build in your own ManyChat account, tested on Instagram, Messenger or both.
- A list of tags and fields explaining what each one means and what sets it.
- Documented integration settings, so another developer could pick them up.
- Suggested wording for your privacy notice covering ManyChat, for you or your legal adviser to approve.
- A short recorded walkthrough of the live chat inbox for your team.
What it costs
Your ManyChat subscription is billed to your own account in US dollars and scales with the number of contacts, so the cost in pounds moves with the exchange rate and the size of your list. At the time of writing, the AI features are a separate paid add-on, so check ManyChat’s current pricing before you budget. My fee depends on the number of entry points, flows and integrations, and on whether I am building from scratch or repairing an existing account. I write each quote in GBP after a free first call, and we agree it in writing, including any support after handover, before work starts.
Next step
Tell me where your messages come from, the questions people ask most and whether you already have a ManyChat account. I will tell you what I would build first, or whether you need ManyChat at all. Get in touch through the contact page or on +44 7352129369.