Perplexity

The engine that shows its working. Perplexity cites sources as a matter of course, numbered and visible, which makes it the one place in this subject where you can see the inputs rather than infer them.

Why that matters more than its market share

Perplexity has fewer users than ChatGPT and far fewer than Google. It is still the most useful engine to measure first, because when it names a company it usually shows you where that came from. You can open the sources, see which ones mention your competitor and not you, and work out what a model assembling an answer about your trade is actually reading.

That is a genuine diagnostic, and it is free. No other engine in this set hands it to you as reliably.

The limit, stated plainly

What you learn here does not automatically transfer. Perplexity is retrieval-first by design, and the others weight their training and their own indexes differently. A source list from Perplexity tells you what Perplexity read. Treating it as a map of what ChatGPT or Gemini read is an assumption, not a finding, and an agency presenting one engine's sources as the answer for all four has overstated its evidence.

PerplexityBot and what the source list does not tell you

Perplexity operates its own crawler, so the same robots.txt question applies here as everywhere. Worth knowing what the visible source list leaves out: it shows what was cited for that answer. What got considered and discarded stays hidden, and so does the reason one source won. The transparency is real and it is partial.

A sensible first measurement

Take ten questions a homeowner in your market would actually type, run them here, and write down every source the answers cite. You will usually find the same handful of directories and review platforms doing most of the work, which tells you where your absence is costing you before you spend anything on content.

In more detail

The crawler and the source list

PerplexityBot, and the three things the visible sources do not tell you.