I folded twenty thin FAQ posts into one hub page: query augmentation, entity chips and canonical tags instead of a link list

My own site had a page whose entire job was to link to other pages. Twenty-eight of them, in a bare list, under a heading that promised to explain hiring a fractional executive. Underneath it sat roughly thirty short posts, most answering a single question in a few hundred words, most of them sitting in Search Console as "crawled — currently not indexed".

That is not a problem you fix by writing more posts. It is a shape problem. Google had looked at the cluster and decided it wasn't worth keeping, and honestly, it was right.

Last week I rebuilt the whole thing. I also used the rebuild to test a technique on something I own, before I let it anywhere near a client's site.

Let the search results tell you the headings

The technique is two steps and costs nothing but attention. First: take the People Also Ask questions Google shows for your target query and use those as your H2s — the literal question, in the literal phrasing people type, not your internal way of naming the topic. Second: search each of those questions, switch to the Images tab, and read the filter chips along the top. Those chips are the entities Google associates with that exact question. Weave them into the answer that sits under that heading.

It is a cheap, observable proxy for a slice of Google's entity graph, per question rather than per topic. You are not guessing which related concepts belong in a section. You are being shown them.

For my cluster I ran six queries through it — what a fractional executive is, what one costs, fractional versus interim, when to hire, what they actually do, how to choose. The consolidated entity set that came back was blunt and useful: fractional CFO, CMO, COO, CTO, C-suite, interim leadership, executive search, retainer, executive compensation, company size. Half of those I would have written anyway. The other half I would have left out because they felt too obvious to bother with, which is exactly the instinct that gets you a page nobody finds.

As Renée Mauborgne writes in Blue Ocean Shift, companies "define the scope of their offering as the industry has long defined it." The same trap sits in your headings. You describe the work the way your industry describes it, and your buyer types something else entirely.

Canonicals, not a graveyard

The rebuild turned that 28-item link list into thirteen question-shaped sections, each one folding in the substance of the thin posts it replaced, each carrying the entities for its own question. Twenty-nine internal links now sit inside prose instead of in a pile.

The part I want to flag is what happened to the old posts. I did not delete them and I did not unpublish them. Twenty of them got a canonical_url pointing at the hub. They stay live, they still resolve, and Google is told plainly which URL is the one that matters. Google's own guidance on consolidating duplicate URLs treats this as the normal move, and it has two properties I like: zero 404 risk, and it is reversible on an afternoon's notice if the measurement says I was wrong.

Deleting would have been faster and it would have felt more decisive. It also would have been irreversible, which is a bad trade for a change whose effect I cannot see for six weeks. I have written before about the work being in what you decide not to index — this is the same call from the other direction.

The schema part, honestly

I regenerated the FAQPage JSON-LD to match the twelve on-page question-and-answer pairs, and I want to be precise about why, because there is a lot of stale advice here. Google cut FAQ rich results back to a small set of authoritative sites in August 2023 and has narrowed them since. If you are adding FAQ markup for the star-and-accordion look in the results page, you are working from a 2022 playbook.

I still ship it, for a different reason. The markup is an unambiguous machine-readable statement of "this page answers this question with this answer", and the things reading pages now are not only the blue-links index. Expect nothing in the SERP and you will not be disappointed; the structure is doing its work elsewhere.

What actually earns the ranking is the same thing it has always been: a page that answers the question better than the alternatives, which is all Google's helpful content guidance has ever asked for. The entity chips don't replace that. They stop you writing a good answer that omits the three concepts the reader assumed you would cover.

Where I will not use it

I rolled the same technique into two other properties the same day — a set of service pages that hadn't been written yet, which is the cheapest possible moment to apply it, and ninety-seven town pages in a directory, where it went into the generator rather than the pages, so it survives the domain migration that project is heading for.

I am not putting it into the essays. Not these ones. The house voice here is the asset, and retrofitting an honest field note into a question-and-answer layout would strip the one thing that makes it worth reading. Query augmentation belongs on surfaces that are already structured: hubs, reference pages, service pages, FAQ blocks. It does not belong anywhere someone is reading for the argument. The same discipline applies to any tool that promises reach — the leverage is in knowing where it does and doesn't apply, not in applying it everywhere.

Measurement is the honest part of this post that I can't give you yet. The baseline is thirty posts largely not indexed and a hub ranking for almost nothing. In four to six weeks I check one number: how many distinct queries the hub URL picks up, and whether the folded URLs drop out as the canonicals are honoured. If it doesn't move, I'll say so here.

Either way, I stopped writing the headings I wanted to write and started writing the ones people type. That was the cheap part, and it was overdue.

Sources & further reading

External: Google Search Central — consolidate duplicate URLs · Google Search Central Blog — changes to HowTo and FAQ rich results · Google Search Central — FAQPage structured data · Google Search Central — creating helpful, reliable, people-first content

Related posts: The work was deciding what not to index · Ten minutes of querying killed the content sprint · 2,390 ways to say 29 things · The death of generic AI