What an assistant actually reads on your site

There is a moment when the difference suddenly becomes clear: when you read your own page the way a model receives it. Not as a layout, but as an excerpt of text, cut out of its context and placed beside excerpts from twenty other sites.

A great deal that works on a website stops working in that state.

The connection is missing

“We're happy to take care of that for you too.” On the page the sentence is perfectly clear, because two paragraphs above it says what “that” is. As an excerpt it is empty.

This is the most common finding and the cheapest to fix: paragraphs that name their own subject instead of inheriting it from the one before. It reads a fraction more laboriously and in return it carries on its own.

The answer is in the picture

Prices in a graphic, opening hours in an image, the service overview as a PDF download. For a person that reads perfectly well. In the text of the page the information isn't there — so for a model it doesn't exist.

The same goes for content that is only loaded after a click. Collapsible sections are convenient; whether their content is in the delivered document or arrives later decides whether it is read at all.

The most important page isn't a page

Often the best answer to a customer's question is scattered across a quote, an email template, the head of a colleague — everywhere except in a paragraph on the website.

That is rarely a technical problem. It is the observation that most websites describe what a company offers, rather than answering what a customer asks.

How to check it

You don't need a tool from us for this. Take the five questions customers ask you most often and look for the answer in the delivered text of your page. Not in the layout — in the text.

What you can't find there, an assistant won't find either.