Chess.com's sitemap index is a short file. It points at 46 others, and that is the entire tree — no nested indexes, no layering. Two levels down and you have every URL the site wants a search engine to know about.
We downloaded all 46 leaf files and parsed every entry. The total came to 832,773 unique URLs — a large site, though not the largest in this series.
What makes it unusual is the composition. 15 of those 46 files contain nothing except profiles of chess players, 50,000 in each. The lessons, the videos, the news archive, the articles, the glossary, the marketing pages — everything the brand is known for — fits in the space left over.
For every lesson page teaching someone how to play, chess.com publishes 148 profile pages about people who already can.
Results at a glance
Chess.com runs a programmatic directory of competitive chess players that is roughly 26 times larger than everything else the site publishes combined. It is not a list of members but a reference set of named players, one page each, generated from a template.
The link profile does not scale with it. 66,476 referring domains is a real asset, but divide it across 832,773 pages and chess.com lands in the crowded lower half of this series — just under PyPI, just above Meetup. Directories grow faster than the links pointing at them. That is the trade.
A directory, not a membership list
The first instinct on seeing 801,793 URLs under /players is that chess.com published its user base. It has not. Member accounts live at /member, and that path appears in the sitemap exactly twice.
The /players pages are something else: a structured directory of competitive players, built the way a governing body's rating list is. The slugs give it away: almost every one contains a hyphen and none contain uppercase characters, because they derive from real names rather than usernames. We pulled 11 at random from across the full set, not the head of the list, and every one returned a live page titled in the form "<Name> | Top Chess Players".
That distinction matters. A site publishing its own users publishes something it created. A site publishing a directory of competitive players claims the search results for something the sport created.
Where the real content actually hides
The strange part is not the directory's size. It is where the rest of the site is filed.
The main numbered sitemap files contain the player directory, the lessons and one hub page for news. The news archive is not there. It lives in files named as image sitemaps. So do the articles. The videos exist only in their own video sitemap family. Together, roughly 25,000 genuine content pages sit inside files a crawler might reasonably skip as media.
There is a second trap beside it. One file is named like a sitemap index but is a plain list of glossary URLs, and another named as though it holds the news archive contains just 4 entries — a Google News recency window. Read either by its filename and you draw the wrong conclusion.
None of this is broken. Every file returned a valid response, every closing tag was intact, and the pages inside are real. But filenames are not documentation. To know what a site publishes, open everything.
What the links look like
Chess.com's link profile is substantial in absolute terms. Our tool reports 66,476 referring domains and a domain rank of 90, with just under 90% of backlinks marked dofollow. For most sites, that is strong.
Divide by the page count and the picture flips. At 0.0798 referring domains per page, chess.com sits almost exactly where PyPI does, a little above Meetup and Rotten Tomatoes, and at roughly half of Eventbrite's rate. The comparison that puts it in perspective is calculator.net, measured earlier at 143 referring domains per page across a few hundred URLs.
Why the ratio is not a mistake
A 26 to 1 split between directory and editorial looks lopsided until you consider what each page type is for.
Editorial pages answer questions people phrase as questions. The supply is finite and each one costs more to write. Directory pages answer a different shape of query: somebody types a name. The supply of names is set by the world, not the publisher, and covering one more costs almost nothing once the template exists.
So the ratio is not really a content decision. It is what happens when you point a template at a list longer than any editorial calendar could match. The lessons and videos were made one at a time. The directory was made once.
The risk is equally structural. A directory growing faster than the interest in it eventually contains a great deal of nothing much, and we cannot tell from outside which side of that line chess.com sits on.
How the site is built
Our audit gave chess.com a score of 83, with 111 checks passing, 3 critical issues, 3 warnings and 5 notices. That is level with Meetup and levels.fyi, a few points behind GOV.UK and Eventbrite, and well behind PyPI.
Read as a report card, 83 is unremarkable. Read as a description of how the site is assembled, it is more interesting. A site with this composition has very few distinct page types — a handful of templates covering more than 800,000 URLs. On-page markup scored 24 out of 25 and mobile passed cleanly, which is what you expect when almost every page comes from the same generator. The lost points sit in the technical category, at 59 out of 67.
The corollary is worth internalising. On a site like this, a template flaw is not one problem. It is the same problem 800,000 times, and fixing it once fixes all of them.
The pattern other sites keep repeating
By this point in the series the shape is familiar. itch.io, Meetup, PyPI and Rotten Tomatoes were all dominated by one entity type, the human-written material a rounding error of the sitemap. Chess.com is a cleaner example, because its dominant type is not user-generated at all. It is a reference set.
What separates the sites that make this work from the ones that merely make it big is whether the entity list was worth having. The template is cheap. The list is the thing.
Copy this in an afternoon
You do not need 800,000 pages to run this on your own site. You need an hour and a willingness to be disappointed.
1. Count your sitemap by first path segment. Fetch the index, fetch every leaf, group every URL by its first path segment. Do not sample and do not guess from filenames. Most teams are wrong about their composition.
2. Open the files you would normally skip. Chess.com keeps its news archive, articles and videos in files named as image and video sitemaps. Anything saying "image" still needs opening, and anything saying "index" still needs checking, because it may not be one.
3. Divide referring domains by total pages. Not backlinks — referring domains. One number telling you whether your link profile kept up with your page count. If not, publishing more pages will not fix it.
4. Audit a template, not a page. Pick one URL from each distinct page type. Auditing 6 templates tells you more about a large site than auditing 60 random pages, and every fix lands everywhere at once.
What you cannot copy
The unfair advantage is the list, and chess.com did not have to invent it. Competitive chess produces named players with public, ordered results as a by-product of being a sport. That roster exists whether or not anyone builds a site around it. Most companies attempting this pattern must manufacture their entity list first, and a manufactured list of 800,000 things nobody has heard of is not the same asset. You can copy the template. You cannot copy having somewhere real to point it.
And here is what we could not check:
We counted URLs, not value. A page in a sitemap is a request for indexing, not evidence of indexing, ranking or traffic. We have no analytics access to chess.com. A large share of the 801,793 player URLs may attract nothing, and nothing in our data would reveal that.
We opened 11 profiles out of 801,793. All 11 returned a live page with a real title and no noindex directive, drawn from across the full list rather than the top of it. That is a reasonable spot check, not proof that every profile is substantial. We did not assess content depth.
Our backlink tool does not fully reconcile with itself. Its dofollow and nofollow figures add up to roughly 5% more than the total backlinks figure on the same screen, which also means the dofollow share depends on which denominator you take. That is our tool's defect, not chess.com's. Treat the total as approximate. Referring domains, the number we compare across the series, is unaffected.
The composition has a small unclassified tail. 124 URLs, or 0.015% of the sitemap, fall into no named bucket — opening pages, forum categories and campaign landing pages. Strictly, one of the 801,793 player URLs is the directory's own landing page rather than a profile. Neither changes anything, but know they are there if you recount.
The raw entry count is not the unique count. The leaf files contain 832,788 entries, of which 15 are duplicates listed in two places. We report the unique figure. Anyone arriving at 832,788 has not made an error — they have simply not deduplicated.
Chess.com's whole site runs on a handful of templates, so one technical fault repeats across hundreds of thousands of URLs. Run the UNmiss Website Audit on one representative page from each page type and you will find the faults duplicated everywhere.
- Technical, on-page, speed and mobile checks in one pass
- Issues split into critical, warning and notice for triage
- The same audit we ran on chess.com and every site in this series
Frequently asked questions
How many pages does chess.com list in its sitemap?
832,773 unique URLs, measured on 29 August 2026. This is a full count, not an estimate: all 46 leaf files were downloaded and every page-level entry parsed, with two independent parsers agreeing. The raw entry count is 832,788, of which 15 are duplicates listed in more than one file.
What share of the site is player profiles?
801,793 URLs sit under the players path, 96.28% of the sitemap. Everything else combined — news, articles, lessons, videos, glossary terms and every product and marketing page — comes to 30,980 URLs, a ratio of 25.9 to 1.
Are these chess.com member accounts?
No. Member accounts live under a separate path that appears in the sitemap only twice. The player pages are a programmatic directory of competitive players with real-name slugs and titles in the form "<Name> | Top Chess Players". Reading this as chess.com indexing its users would be wrong.
What makes up the rest of the sitemap?
News accounts for 14,206 URLs, articles 6,927, lessons 5,403, videos 4,087 and glossary terms 233. A further 124 cover openings, forum categories and assorted landing pages. There are 64 distinct first path segments, and 52 of those have exactly one URL each.
Why is content stored in image sitemaps?
We do not know the reason, only the effect. The news archive, the article set and the video pages are listed as page URLs inside files named as image and video sitemaps, and appear nowhere else in the tree. A crawler treating those files as media-only would miss roughly 25,000 real content pages.
What does the backlink profile look like?
Our tool reports a domain rank of 90 and 66,476 referring domains, with 7,689,049 total backlinks split 7,268,797 dofollow to 819,234 nofollow, or 89.9% dofollow. The split sums to about 5% more than the reported total, so treat the total as approximate.
How does that compare with the rest of the series?
At 0.0798 referring domains per page, chess.com sits just below PyPI at 0.082 and just above Meetup at 0.069. The top of the range is calculator.net at 143 across a few hundred pages, the bottom levels.fyi at 0.026. All figures were measured with the same tool.
What did the site audit find?
A score of 83 with 111 checks passed, 3 critical issues, 3 warnings and 5 notices. On-page scored 24 out of 25, technical 59 out of 67, speed 27 out of 29, mobile 2 out of 2. For comparison, PyPI scored 91, MDN and itch.io 88, Healthline 74.
The interesting thing about chess.com is not that it publishes many pages. Plenty of sites do. It is that the pages almost nobody discusses outnumber the ones the brand is famous for by 26 to 1, and the imbalance is not neglect. It is what a template does when you give it a long enough list.
The lesson is that the list comes first: point the same template at a list nobody was searching for and you get the page count without the demand. And whichever list you point it at, the fault you build into the template travels with every page. Pick one URL from each of your page types, run the UNmiss Website Audit over them, and fix the problems where they multiply rather than where you noticed them.
Measured on 29 August 2026. Page counts come from a complete parse of chess.com's sitemap index and all 46 leaf files, counting page-level entries only and excluding image and video media URLs; the count was reproduced by a second parser with no discrepancy.
Backlink and audit figures come from the UNmiss Backlink Analyzer and UNmiss Website Audit, run on the same date with the same settings used for every other case study in this series.