The future of Google Search
If Google sends fewer visitors, what brings the next customer?
Plan for a changed journey, not a perfect prediction.
The old question was what Google would do next. AI Overviews and AI Mode are already part of Search, alongside familiar web links and other result types. The useful question for a business is which customer decisions still need its website, and what happens if fewer people reach it.
There is no reliable forecast for your future click rate just because a search shows an AI answer. Treat the possible loss as something to investigate and prepare for, rather than a fact about every page.
A question may end on the results page.
A brief definition or opening-time lookup may give a person everything they need. A page that repeats the same short answer has a limited reason to win a visit. That does not make accurate basic information worthless; customers still need it.
A decision may need more evidence.
A customer may still need exact availability, a detailed comparison, photographs, service limits or a conversation. Build the page around that decision. Complex topics are also handled by AI features, so complexity alone does not protect your traffic.
Use the rules that exist now.
Google’s AI Search guidance says ordinary SEO practices remain relevant. A supporting page must be indexed and eligible to appear with a snippet. Special AI files or special schema markup are not required, and meeting the requirements does not guarantee a link.
Keep the useful material accessible.
Provide important information as text, connect relevant pages with internal links and make images and video useful additions. Ensure structured data agrees with the visible content. Check accurate business and product details where those apply.
Make participation a deliberate choice.
Search Console now has a Search generative AI control under Settings. It can include or exclude a property’s links and content from covered AI features, or inherit a parent property’s setting. Check which property and inherited choice apply before changing it.
Google says this control is not a ranking or inclusion signal for other parts of Search. It does not control AI training. Decide whether to participate from your publishing needs and evidence, with those separate effects understood.
A scenario to test
How much demand depends on Google?
Use qualified enquiries or another meaningful outcome to make the dependency concrete. The figures below are invented. They show a planning calculation, not a prediction of what AI will do.
100 enquiries
The business receives 100 qualified enquiries in a month. Its available records attribute 60 to organic Google search.
30% fewer from Google
Assume those 60 enquiries fall by 30%, while every other source stays unchanged. That removes 18 enquiries.
82 remain
The total becomes 82. Maintaining 100 would require 18 additional qualified enquiries, or a different route to the business’s actual target.
Your figures may be uncertain: customers use several channels, attribution can be incomplete and enquiry quality can change. Run more than one scenario. Then ask whether another source can realistically supply demand of comparable value, and whether the business can serve it profitably.
A smaller traffic total with stronger enquiries can be useful. A larger total with few suitable customers can conceal a problem. Keep both the volume and the quality visible.
Give the next visitor something a summary cannot finish.
Consider an invented storage business. A short explanation of storage sizes may answer a general question. Someone choosing a unit still needs to know whether their furniture fits, how access works and what is available for their dates.
Turn general advice into a decision.
Show accurate dimensions and useful photographs. Explain access restrictions, what the price includes and what information is needed to check availability. Keep the enquiry route working. Those details help a customer choose without pretending the page guarantees an AI citation.
Make the evidence accountable.
Name the person responsible for specialist advice where that helps the reader judge it. Explain relevant experience, link important claims to primary sources and review details that can change. A byline or review date should describe real work, not decorate unsupported copy.
Apply this test to an existing page before commissioning more: after reading a short summary, what would the customer still need to see, compare, verify or do? Supply that missing value directly. Adding length without an answer does not solve the problem.
Separate AI visibility, website visits and customer outcomes.
Google announced dedicated Search Generative AI performance reports in June 2026 and confirmed worldwide rollout on 31 August. These provide an impressions view for AI features, with page, country, date and, for Search, device breakdowns. The data also remains in overall performance reporting.
Use visibility to find pages worth inspecting.
Check which URLs appear in AI features and how their impressions change. The dedicated report measures visibility; it does not by itself establish clicks, revenue or enquiry quality. An impression is not a customer visit.
Use other evidence to judge the business effect.
Compare relevant Search Console traffic with website analytics and actual enquiry or order records. Keep periods and filters comparable. Note changes in demand, tracking and campaigns. A manual search result check is a sample, not a measurement of everyone’s experience.
Build one more route to the same customer.
Choose a channel that fits how customers already find and trust the business. For the storage example, a relevant local partnership or a useful resource people return to may be worth testing. A consent-based email relationship can help appropriate repeat purchases; it is not automatically useful for every service.
Keep your website, source material and customer records usable outside one platform. Test a small, specific route with a named owner and a clear measure of suitable enquiries. Expanding to every social network adds work before it proves demand.
Start with one important page and one additional route to its audience. Improve the decision the page helps with, record the baseline and review what happened. That work can remain useful through several changes to Google’s search interface.