In many marketing meetings at the moment, one question dominates: how do we get into ChatGPT's answers?It is understandable that this question is being asked.It becomes problematic when budget and attention are subsequently withdrawn from what has so far been the most important access channel.

A current study provides a sober reference point for this.According to the study, AI assistants are sending more visitors to websites than ever before.At the same time, Google still delivers around fifty times that volume.Both statements are true at the same time – and it is precisely this simultaneity that is often overlooked in practice.

Growth rate is not the same as volume

The most common thinking error in this debate is confusing relative growth with absolute reach.A channel that starts from a very low base will inevitably show impressive growth rates.In the analytics dashboard this looks like a trend break, but in absolute visitor numbers it remains a marginal quantity for the time being.

For budget planning, this means: percentages are poorly suited as the sole basis for decisions.More relevant is the question of how many qualified contacts a channel actually contributes and how that number develops over several quarters.Anyone who does not measure this properly is optimising based on assumptions.

This is expressly not an argument for ignoring AI assistants.It is an argument against reallocation without a data basis.The study suggests investing in addition rather than replacing – and at a scale that corresponds to the actual contribution.

The good news: the work overlaps

There is considerable overlap between classic search engine optimisation and discoverability in AI systems.Language models and answer systems draw on content that is machine-accessible, clearly structured and unambiguous in topic.These are the same properties that have mattered in organic search for years.

Specifically, both channels benefit from the same measures:

  • Clear page structure: a clear hierarchy of headings, one topic per page, no duplicates across multiple URLs.
  • Content that answers questions: precise definitions, concrete figures, comprehensible examples instead of general advertising language.
  • Structured data: machine-readable markup for products, events, organisations or FAQ content that makes the context of a page explicit.
  • Technical accessibility: clean load times, server-delivered content and no central information that only comes into being through JavaScript in the browser.
  • Traceable origin: author details, update dates, reliable information about the company.

Anyone who works through these points improves their position in classic search and at the same time increases the likelihood of being cited in AI answers.In most cases, a separate project with its own budget and its own agency is not necessary for this.

Where the channels differ

There are still differences, and they mainly concern user behaviour.In classic search, the results list determines whether someone clicks.With an AI assistant, the answer is created directly in the chat window; a visit to the website is then optional and tends to happen when a concrete next step is due – a price, a form, documentation, a contact.

This results in a changed set of expectations among the visitors who actually arrive.They are often already informed and are looking for confirmation or details.Landing pages that start by explaining everything from scratch again waste potential at this point.Pages that get to the point quickly and offer clear options for action are more sensible.

Second difference: measurability is worse.Referrals from chat interfaces cannot always be clearly attributed.It is therefore worth evaluating referrer data specifically and, if in doubt, openly asking about the first point of contact, for example in the quotation process.

What the CMS must deliver

Many of the requirements mentioned are less a question of copywriting than a question of system architecture.If structured data can only be maintained manually per page, this creates effort that gets left behind in day-to-day work.If content is stuck in rigid text fields, it cannot be sensibly reused as a data source.

In a system such as TYPO3 – an open-source CMS that is widely used in mid-sized environments – these points can be solved structurally:

  • Content elements with clearly defined fields instead of unstructured text blocks, so that information remains machine-interpretable.
  • Automatically generated structured data from existing fields, instead of manually maintained code snippets in individual pages.
  • Consistent URL and navigation logic that maps topic clusters and avoids multiple paths to identical content.
  • A clean interface layer in case content is also to be played out in other channels or applications.

The advantage of this approach: it is independent of which assistant or which search interface dominates in two years.Well-structured content survives channel shifts; tricks tailored to a single platform rarely do.

A pragmatic sequence

For companies that have to decide now, the following prioritisation has proven effective.First: keep the existing organic channel stable and eliminate legacy technical issues.Second: revise content so that it answers concrete questions precisely.

Thirdly: anchor structured data systematically in the CMS.Fourthly: set up measurement points for AI referrals and monitor them over several months.

Only once this foundation is in place and the figures show their own momentum does a separate budget line for AI visibility make sense.Until then, the growth is a by-product of good foundational work – and no reason to neglect the channel that continues to deliver the majority of visitors.

Classification for web and software projects

For ongoing projects, this above all means: the investment in clean information architecture, structured content and technical quality currently pays off twice – in classic search and with AI assistants.Anyone who instead starts a separate optimisation project for a channel that currently delivers a fraction of the volume increases complexity without a secured return.The more robust strategy consists of setting up the CMS and the content models in such a way that new output channels can be connected without rebuilding.

Sources