SEO after the AI summaries
When search engines answer the question directly, the traffic model changes. What still works, and what stopped.
Sun 14 January 2024
For twenty years the implicit deal was straightforward: publish something useful, rank for the query, receive the click. Generated answer summaries have altered that arrangement, and a lot of content strategy built on the old assumptions is quietly underperforming.
What stopped working
Content that exists purely to answer a simple factual question is the most exposed. If the answer fits in two sentences, the summary will contain those two sentences and the click will not happen. Volumes of thin explanatory articles — the standard playbook for a decade — have lost most of their value.
What still works
Anything that cannot be usefully compressed. Original research, because there is no source to summarise from. Opinion and analysis, because a summary of an argument is not a substitute for the argument. Anything requiring trust, where the reader wants to know who is saying it. Tools and calculators, where the value is in the interaction rather than the text.
The practical shift
- Fewer, deeper pieces rather than high-volume thin coverage
- Distinctive point of view over comprehensive neutrality
- Original data, even in small amounts, as a genuine differentiator
- Brand search as a metric that matters more than it used to
Measure differently
If you judge content solely by organic sessions, you will conclude the whole channel is failing. Some of it is — and some of it is doing its job in ways your analytics were never set up to see. Track branded search volume, direct traffic, and how often prospects mention having read something. Those signals were always the point; the click was just a convenient proxy.
