Without really planning to, I’ve found myself spending more time researching and writing about AI and content.
This piece takes a slight detour from my recent posts on governance and retrieval. I’ve been exploring a question I keep coming back to: if someone can get what they need from our content without visiting our website, what still makes the visit worthwhile?
An organisation’s continuing value depends on what it can help someone accomplish after their question has been answered. That is where I believe content strategy needs to put more attention.
What are we asking visibility to deliver?
Clear structure and good metadata help an AI system extract information accurately. In regulated industries, that matters. If someone encounters our information through an AI answer, I want the answer to preserve its meaning.
The assumption I want to challenge is that making content easier to distribute will also sustain the relationship with its reader.
A useful explanation can do its entire job somewhere else. For public information, that may be an excellent outcome. The difficulty comes when we report that distribution as success against a plan that still depends on website visits.
Appearing in more AI answers tells us something about reach. It does not establish that more people completed a task, used a service or developed a reason to return.
Before investing, we should decide which outcome matters. Otherwise, we risk improving the distribution of an answer while leaving the purpose of the website unresolved.
A source can matter without receiving the visit
In my earlier articles, I explored how AI can reproduce an approved claim while dropping the qualification that made it valid. A statement about a treatment, for example, can change meaning when its patient population disappears.
That raises a related question about the source itself. What must remain connected to the information for someone to rely on it?
An approved product statement has a version and a market in which it applies. Its status can change. Reproducing its words accurately does not establish that it is still current.
Maintaining the governed record therefore has continuing value. That does not guarantee traffic to a company website.
The authoritative record may belong to a regulator. An intermediary may retrieve the current version and provide enough provenance for the user’s purpose. Someone can depend on a source without opening its website.
Maintaining a reliable record can deserve funding even when its value is delivered through another interface. We need to recognise that contribution without pretending it will necessarily generate a visit.
Give someone a reason to use the service
My first instinct was that interactive tools offered a clearer answer.
A calculator or assessment responds to a person’s circumstances. Its output is not simply a paragraph waiting to be extracted.
But a generic calculation can be performed elsewhere. A questionnaire can be recreated. An agent may access a useful service without taking the person through the interface we designed.
Adding interactivity is therefore too weak an investment criterion. The stronger question is what the tool enables and what it depends on.
Consider explaining eligibility versus confirming it for a particular person.
The explanation can travel widely. Confirmation may require access to their records and a decision from the organisation responsible. The value lies in getting a reliable determination that the person can act on.
The same distinction applies to explaining how to resolve a problem and actually resolving it. A good answer can tell someone how to correct an account error. The service has to let them make the correction.
This is where I expect more investment to move: towards the maintained information and working processes that let someone complete the next step. A better interface helps, but the organisation must be able to deliver the outcome behind it.
Give them the answer before they arrive
There is a practical way to test this on an existing content estate.
Choose a small set of important pages or journeys. For each one, ask:
If an AI gave the reader a faithful answer, what would they still need from us?
For some pages, the answer will be nothing. Treat those as information we want to distribute well. Check whether it is represented accurately and whether people can trace it to the appropriate source when they need to.
For a journey involving a service, name the remaining task. Perhaps someone needs to confirm eligibility, update a record or speak to a person who can resolve an exception. Measure whether they can complete that task and where they get stuck.
There will also be pages where nobody can clearly explain what remains for the visitor to do. Those deserve attention before another round of visibility optimisation.
Then test the journey with people. Give them the answer before they visit. See whether the website still helps them accomplish something worthwhile.
If it does, we have a concrete reason to improve it. If it doesn’t, we should reconsider its purpose. A page can remain useful as a published source without needing to become a destination.
Where I would put the investment
I still believe clear informational content deserves investment, particularly where accuracy matters and the organisation has a responsibility to publish it.
But a content strategy needs to distinguish between information that should travel freely and services that depend on a continuing connection to the organisation. Both can create value. They need different investment decisions and different measures of success.
That may mean funding something less visible than a new page. Keeping a record current takes ongoing work. So does fixing the process that prevents someone from getting a decision. Neither should be overlooked simply because its contribution is difficult to express as page views.
In the next content planning discussion, I’d start here:
If the answer is already available elsewhere, what can someone still accomplish with us?