Why customers hate chatbots, specifically

The complaint is never "this is a bot". It is always one of four specific things, and every one of them is a decision somebody made.

· 1 min read

It hides the person

The single biggest cause. A bot placed in front of your team as a filter is experienced as a wall, and people learn to type "agent" before they type anything else.

It makes them repeat themselves

Explaining the problem to a bot, then explaining it again to the human it hands you to, is the moment goodwill runs out. The transcript should travel with the customer.

It answers a question they did not ask

Keyword matching returns the nearest article, not the answer. Being handed a returns policy when you asked about delivery reads as being ignored by something that is not even listening.

It is confidently wrong

Worse than useless, because they act on it. A wrong delivery date is a complaint, a refund and a review — and the shop finds out days later.

What actually fixes it

Answer the question that was asked, from real data. Hand over the instant a person is wanted, without being asked twice. Carry the conversation across with them. None of that is a model capability — it is configuration, and it is yours to get right.

The pattern underneath all four

Every one of them is the shop optimising for itself. Hiding the person saves support cost. Keyword matching is cheaper than understanding. Always producing an answer looks better in a deflection report. Customers can feel which side a chat window was built for, and they respond accordingly.

What people actually say they want