You probably don't need a chatbot

We build chatbots. So it's a slightly odd thing to say, but most of the small businesses that ask us for one don't need it — at least not first, and sometimes not at all. The chatbot has quietly become the default answer to the question "how do we use AI," and defaults are worth being suspicious of, especially when everyone selling them benefits from you not asking too hard.

The logic feels obvious. AI is good at conversation, customers ask questions, so put a conversational thing on the website to answer them. Fair enough on the surface. But underneath, "get a chatbot" is usually a solution that arrived before anyone defined the problem. And a solution in search of a problem tends to end up as a widget in the corner of your site that a few people poke at and most ignore.

Ask what it's actually replacing

The honest test for any chatbot is simple. What work does it take off a human, and how much of that work is there?

If you're a busy e-commerce shop fielding hundreds of "where's my order" and "what's your returns policy" messages a week, a chatbot has a real job. It's replacing genuine, repetitive volume, and it will pay for itself in hours saved and faster answers. That's a good fit, and we'll happily build it.

But a lot of small businesses don't have that volume. A specialist consultancy, a trades business, a B2B firm with a considered sales process — these get enquiries measured in the dozens per week, and each one is a bit different and quite valuable. Automating those conversations doesn't save much time, and it risks putting a robot between you and exactly the high-value conversations you most want to have personally. The chatbot isn't replacing drudgery there. It's replacing your best sales channel with something worse.

Before you buy a chatbot, count the conversations. If the repetitive volume isn't there, the chatbot isn't either.

The friction is usually somewhere else

Here's what we see when we actually look at where a small business loses time. It's rarely on the website answering the same customer question. It's in the back office. Chasing quotes and follow-ups. Copying information between systems that don't talk to each other. Drafting the same kinds of documents over and over. Assembling reports by hand. Getting an answer out of a pile of internal documents that only one person really understands.

None of that is customer-facing, which is exactly why it's invisible and exactly why it never gets automated. The chatbot is appealing partly because it's visible — it feels like doing AI, because visitors can see it. The boring internal automation that would actually give your team their afternoons back doesn't make a nice screenshot, so it gets skipped in favour of the thing that photographs well.

If you optimise for what looks like AI adoption, you get a chatbot. If you optimise for hours returned to your team, you almost always get something else.

When a chatbot genuinely earns its place

None of this is anti-chatbot. It's anti-defaulting. There are clear situations where a chatbot is the right first move, and they share a shape.

You have real, repetitive question volume — enough that a person spends meaningful hours each week answering things a well-briefed assistant could handle. The questions have knowable answers that live in your policies, product details, and FAQs, rather than requiring judgement on every reply. And you have a clean way to hand off to a human the moment a conversation goes beyond the routine, so the chatbot deflects the boring 70% and escalates the rest instead of trapping people in a loop.

Hit those three and a chatbot is a genuinely good investment, and one of the more satisfying ones because the benefit is immediate and obvious. Miss them, and you've bought a feature you'll quietly switch off in six months.

Start from the problem, not the product

The deeper point isn't about chatbots specifically. It's that "which AI product should we buy" is the wrong opening question. The right one is "where does our business lose the most time or money, and is any of it a good fit for AI." Answer that honestly and the product falls out of it. Sometimes that's a chatbot. Often it's an internal automation nobody would have thought to ask for, because it isn't the thing you see advertised.

This is why our first conversation with a client is a look at where the real friction is, not a pitch for a particular build. Occasionally the most useful outcome of that conversation is us saying a chatbot isn't your priority — that the money is better spent automating something in the back office, or that a much smaller change solves most of the problem. That's a strange thing for a build agency to say, and it's exactly the point. The businesses that get real value from AI start from their own problems, not from whatever product has become this year's default. If you want to talk through where yours actually is, that's what https://cal.com/aegeanpulse/ai-strategy-consultation is for.