Synonyms with Google's forward/back navigation — walking the related-search graph.

Most Google-based synonym work stops at the first SERP — autocomplete, the visible results, the "People also ask" block. The related-search graph goes further. Click into a SERP listing, hit back, and Google surfaces a "People also search for" panel that's tuned to the page you bounced from. Each new node opens new vocabulary; chained together, the graph maps how Google sees the product category's intent.

The forward-walk pattern

1. Run the main query on Google. Capture the autocomplete suggestions.
2. Click into the top organic result that points to Amazon (or to a category-review article).
3. Hit the back button. Google now shows a "People also search for" overlay or a refinement-chip row beneath the result. Capture every chip.
4. Click each chip in turn — each is a new query. Repeat the back-walk on each.
5. After 3–4 hops, the chip set stabilises; new queries stop appearing. That's the boundary of the cluster.

What the graph reveals that other sources miss

• Cross-category synonyms. A walk that starts at "loaf pan" might surface "proofing basket" or "dough scraper" — adjacent-category products that shoppers research in the same session. Useful for cross-listing strategy and for "Frequently bought together" optimisation.
• Intent transitions. The chip row often progresses from generic ("loaf pan") to specific ("non-stick loaf pan 9 inch") to comparative ("Pyrex vs USA Pan"). The chain of progression is a map of how shoppers refine — and refinement vocabulary is bullet-copy vocabulary.
• Geographic vocabulary. Different walks in different markets reveal local terms. A UK walk surfaces "tin" where a US walk surfaces "pan"; an Australian walk might surface neither.

The back-walk pattern

On the SERP itself, after clicking into a result, Google sometimes shows a different chip set when you return to the SERP than the one you started with. The post-click chips are personalisation-adjusted by Google's signal that you "engaged" with the result. Reading the difference between the pre-click and post-click chips exposes the queries Google associates with high-engagement intent — usually a tighter, more transactional set.

The discipline of stopping

• Stop after 3 hops. Hop 4 and beyond drift into adjacent categories your listing has no business in.
• Discard chips with branded competitor names — same rule as every other source.
• Discard chips that include city names or regional store names — local-intent queries Amazon doesn't reward.
• Keep chips that include the product type plus a use-case word — those translate directly into bullets.

How this fits the synonym sources stack

The Google related-search graph contributes the smallest harvest of all the Episode 03–13 sources — typically 10–25 candidates per ASIN. But the unique value is the cluster structure: most other sources hand you a flat list, while this source hands you a labelled cluster (refinement chips already group related variants). The cluster labels often become the bullet headings in Module 8.

Where to spend the time

One full walk per category, refreshed when major brand or category news breaks. The decay rate of Google's related-search graph is slow — last year's walk is mostly still right. Treat it as the cross-check, not the foundation: the rest of the synonym sources do the heavy lifting; this one catches the clusters they missed.