The original article collected my takeaways from reporting on the State of Search conference ahead of 2016. It mixed practical website work with forecasts about structured data and assistants.

Those forecasts belong to their time. The useful work can be assessed without repeating the old statistics as facts about your business.

Review duplicate URLs by purpose

Identify alternate versions of the same content and decide which page should be the main destination. Check canonical signals, redirects and internal links. Don’t assume every archive or repeated element is a penalty, or remove useful pages to make two reporting totals match.

Measure performance rather than quoting a universal loss

The old article turned a page-abandonment statistic into a claim about lost business. That inference was not justified. A visitor leaving is not automatically a lost sale, and one loading threshold does not describe every audience.

Test representative pages and the steps customers need to complete. Prioritise faults that obstruct those steps, then measure the effect of the fix. Our conversion work connects that investigation to the customer journey.

Keep structured data accurate

Structured data should describe the content and offer actually present on the page. It is not a substitute for useful visible information or a guarantee of enhanced search presentation.

Use our structured data explainer to separate implementation checks from ranking expectations.

Answer the question clearly

One useful distinction in the original discussion was between structured data and the answer a page provides. Write a direct, useful answer to the customer’s question. Don’t assume a markup change or a question heading will secure a featured result.

Treat assistant predictions as history

The show discussed the prospect of assistants helping with inboxes and schedules. It did not demonstrate a working system or establish that autonomous actions were safe. For present-day business use, start with the task, data access and approval boundaries rather than a film analogy. See our AI consulting work.

Historical note: this is a refreshed summary of the original conference recap and transcript at their original article URL. Publication date retained. Old numerical claims, expired events, product predictions and unsupported ranking implications are not presented as current evidence.