Changelog

2026-09-20

• A rebuilt home page with a live article feed — the front page now carries a search-style Ask box, the headline catalog numbers, quick-start cards for every major tool, scenario shortcuts and a Latest articles feed that surfaces the ten most recent evidence-backed pages (answers, findings, comparisons, guides and brand reports) with their claim and link.
• Answers promoted to the main navigation — Answers now has its own top-level entry, and the navigation was regrouped into Insights / Tools / Buying / More menus so the site’s question-driven pages are reachable from every page. The home search box deep-links into Ask about cookware , running the question you type.

2026-09-19

• First-party demand signals from Reddit trend monitoring (internal) — the topic-ranking pipeline now also reads Reddit Pro’s trend monitoring (topic-level discussion volume and the month’s most-discussed threads) alongside autocomplete, Google Trends and Reddit search. It decides what we write about next; like the other demand signals it is never used as review evidence. Only on-topic threads count toward a keyword’s signal, so viral off-topic mentions don’t distort the ranking.
• Live community threads on the answer pages — the Answers pages now show the Reddit threads people are actually posting about the topic, next to our evidence-backed answer, and the Answers hub maps each tracked topic to the page that answers it (topics without a page become the next writing queue). These threads are labelled as a demand signal, not evidence — every number on the page still comes from our review corpus.

2026-09-18

• Sharper search-demand signals — the topic-ranking pipeline now combines Google Trends interest and Reddit discussion depth with autocomplete. Reddit data comes from the same-origin search API (up to 100 threads per question, with comment and vote totals), so questions are ranked by real discussion volume rather than a page of search results. Trends terms are measured on one shared scale, so “dutch oven” and “air fryer” can be compared directly instead of each being normalised to its own peak. Neither signal is used as evidence — they only decide what we write about next.

2026-09-17

• Every score now discloses its coverage — the scoring standards page states how many products (43,093 scored / 21,482 above the sample threshold) and brands (3,166 / 669) are actually measured, so you can judge what a score represents.
• Brand reliability pages rebuilt on real evidence — 60 brand pages now use the full, normalised brand rollup (e.g. Lodge: 186 products / 64,121 reviews). Previously almost every brand was marked insufficient data.
• Product pages now surface the most-reviewed products — the product insight pages are ranked by review volume and evidence, not by file order, so each page is backed by a meaningful sample instead of a handful of reviews.
• New finding engine (internal) — 8 classes of review-derived findings (price inversions, brand divergence, material effects, durability trajectories, claim-vs-measured gaps) are mined offline and every number is traceable to its metric and sample. The first 38 are published under Findings — A-vs-B classes (brand vs brand, material vs material, durability) appear only once, under Comparisons .
• Head-to-head comparisons — 39 new A vs B pages (brand vs brand, material vs material, durability trajectories), published only when the two sides differ by a statistically significant margin, with the metric, both sample sizes and the limitations shown.
• Thin-page cleanup — removed 527 legacy best-under pages built on a retired taxonomy and replaced them with 31 review-backed pages (Under budget ), each showing the price band’s top complaints and sample size. Consolidated 52 one-line glossary stubs into 7 data-backed glossary hubs . All retired URLs 301 to the new pages.
• One number, one source — every public figure now comes from a single generated file (data/site_facts.json), so pages can no longer disagree on how many reviews or products back the site. The pain-point shares were recomputed on the current corpus: each problem’s percentage is now its share of all labelled problem mentions (49,353 mentions across 28,556 products), not the share of negative reviews.
• Trust pages — new About , Contact , Privacy , Terms and Data Sources & Limitations pages explain who runs the site, how AI is used, the affiliate and advertising relationships, your privacy rights, and the ten limits you should keep in mind when reading any score or finding. All linked from the footer.
• Cleaner internal linking — links to pages retired in the cleanup now point straight at their live replacements instead of hopping through a redirect, the product lookup and comparison tools are linked from the footer, and every page is reachable from a hub (build now checks for orphan pages and dead internal links).
• Answers to the questions people search — 21 new answer pages target high-intent queries (does ceramic cookware last long?, dutch oven dishwasher safe, cast iron skillet safe for glass top stove, ceramic pan vs teflon and more). Each opens with a quantified answer and its sample size, then shows the metric, limitations and evidence chain.
• Download the data — the aggregates behind every number are now published as free CSV/JSON on Download the Data : brand scores, type×problem rates, material effects, durability trajectories and every mined finding, under CC BY 4.0, so anyone can verify or re-cut them.
• Press-ready data stories — a new Press & data stories page packages the most citable findings (price inversions, brand divergence, durability trends, claim-vs-measured gaps) with their metrics, samples and limitations.
• Durability Checker tool — an interactive Durability Checker shows any cookware type’s rating trajectory by review year and its most-reported problems, from the same corpus as every score.
• Monthly content review — all 432 indexed pages audited for thin and duplicate content (none found after the cleanup), and the next month’s topic shortlist refreshed from search-demand signals. The site also gained proper noindex support so weak pages can be retired without deleting them.
• Fixed legacy URL redirects — retired glossary and best-under URLs now 301 correctly to their live pages. (Cloudflare Pages only honours the first 100 wildcard redirects, so the map was rebuilt with one exact rule per old URL.)
• Search-demand signals now include Reddit — discussion counts for high-intent cookware questions are aggregated through the search index (Reddit blocks direct fetches) and folded into the topic-ranking formula, so next month’s pages follow real questions as well as search autocomplete.

2026-09-03

• Social sharing with highlighted excerpts — share buttons now lead with each page’s key takeaway (the evidence-backed claim or summary) instead of just the title, so posts are more eye-catching.
• How to Use guide — new guide page walking through every tool on the site.
• Changelog — this page, tracking site and data updates.