Case Study: How Indeed’s Job-Title-Plus-City Pages Built a Search Giant

Indeed's job-title-plus-city pages are one of the cleanest programmatic SEO patterns ever run at scale, and the public record on them is unusually detailed because the company's own SEO team has described how the system works. Indeed launched in November 2004 as a job-listing search engine co-founded by Paul Forster and Rony Kahan in Austin, Texas and Stamford, Connecticut (Wikipedia (2026)). Instead of hosting jobs itself, it aggregated listings from job boards, staffing firms, associations and company career pages, then served a dedicated page for nearly every pairing of a job title and a location, so a search like nurse jobs in Austin resolved to a matching Indeed URL. That structure carried the site past Monster.com to become the highest-traffic job website in the United States by October 2010, a dated milestone, and into an October 2012 acquisition by Japan's Recruit Holdings (Wikipedia (2026)). This teardown separates what Indeed built from what third parties estimate about its traffic, and it ends with the parts of the pattern a directory or marketplace can still copy today.

The numbers

The most reliable part of the record concerns Indeed's origin and scale, because those facts come from company history and dated reporting rather than traffic estimators. Indeed was founded in November 2004 by Paul Forster and Rony Kahan, with bases in Austin, Texas and Stamford, Connecticut, and the company is available in more than 60 countries and 28 languages (Wikipedia (2026)). The company aggregates job listings from thousands of websites, including job boards, staffing firms, associations and company career pages, which is the supply layer the title-plus-city pages were built on top of (Wikipedia (2026)). In 2005 the young company took a $5 million investment round from Union Square Ventures, The New York Times Company and Allen & Company (CNET (2005)).

The competitive milestones are dated snapshots rather than current rankings, and they should be read that way. Wikipedia records that in October 2010 Indeed surpassed Monster.com to become the highest-traffic job website in the United States, and that on October 1, 2012 the company was acquired by Japan-based Recruit Co. Ltd. and became an independent operating unit (Wikipedia (2026)). Recruit later acquired Simply Hired in 2016 and Glassdoor in 2018, which made Glassdoor a sister company to Indeed (Wikipedia (2026)). The description of Indeed as the world's number one job site is a claim that appears in a vendor case study, not in Indeed's own filings, so it should carry that attribution rather than being stated as an objective ranking (Indexing Insight (2026)).

The scale of the underlying pages is best conveyed by the company's own engineering record. Indeed's engineering blog documented the work of scaling its document-serving system from 1 to 1 billion job-related documents, a figure that reflects the size of the job-listing corpus the title-plus-city templates render (Indeed engineering blog (2013)). That is a company-reported figure describing documents served rather than unique landing pages, so it measures supply rather than indexed URLs.

Two further figures are third-party estimates, not company numbers. Similarweb and Semrush data reported by AIM Group showed Indeed's total visits falling 10% globally and 9.4% in the United States between January and May 2024, and AIM Group itself framed those as third-party estimates rather than Indeed-reported figures (AIM Group (2024)). Separately, Google Trends data analyzed by Startups.co.uk found Indeed was the most-searched job site among UK workers, with the search "Indeed jobs" reaching peak popularity, a score of 100, at the start of January 2022 (Startups.co.uk (2024)). A score of 100 in Google Trends is a relative popularity index for UK searches only, not an absolute count of visits, so it should not be read as a traffic total.

The index-coverage figures deserve a specific caveat, because the source itself is internally inconsistent. The Indexing Insight case study carries a headline of 88% index coverage while its own description says the team got 80% of key pages indexed, so either number should be cited with its attribution and the other noted (Indexing Insight (2026)). The more useful detail from that study is the shape of the coverage curve rather than any single percentage: near-total coverage is easy at low page volume, and a stubborn 5 to 10% goes unindexed once volume passes roughly 10,000 pages.

What they built

Indeed's core asset is a templated page for a job title in a location. Take a structured record for a job, render a page for the pairing of its job title and its city, and repeat that for every title and every city where listings exist. A search for nurse jobs in Austin resolves to a page that aggregates the matching listings for that title and place. The pages exist because the data exists, not because an editor chose to write them, which is what makes the system programmatic rather than editorial.

Aggregation as the data layer

The supply of pages depends on the supply of listings, and Indeed's supply comes from aggregation rather than direct employer input. The company aggregates job listings from thousands of websites, including job boards, staffing firms, associations and company career pages (Wikipedia (2026)). That model means Indeed inherits both the breadth and the noise of the wider job market: it can build a title-plus-city page for any market where a partner site posts a listing, but it also has to deduplicate and normalize the same job posted across several boards. The aggregation layer is the reason the template could scale to the long tail without a sales team signing up every employer in every city. The same catalog logic runs the education vertical, which the Coursera case study traces through course pages.

The title-plus-city template

The page itself is a template with three variable parts: the job title, the location, and the set of listings that match both. The title and location fill the headline and the URL, and the listings fill the body, so the page is specific where specificity matters to a searcher and a search engine. A nurse jobs in Austin page and a nurse jobs in Dallas page are nearly identical in structure but differ in the listings they surface, which is enough difference to be useful when the underlying data is fresh. The risk, as with every templated system, is that structure alone does not make two pages meaningfully different, and the quality of the data is what decides whether the template reads as useful or as filler.

A new section and a new URL path

The hyper-local pages were not a restyling of existing pages; they were an entirely new section and URL path on Indeed, which meant Google needed time to establish trust in that part of the site (Indexing Insight (2026)). This is a recurring theme in programmatic SEO: a new path starts with no accumulated authority, and its pages compete for crawl and indexation against the rest of the domain. Indeed's team treated that ramp as expected rather than a failure, a useful correction for anyone who expects a new section to index immediately at scale.

Machine learning at scale

At a certain size, hand-authoring the variables for a huge number of pages stops being practical, and Indeed moved toward machine learning to power its programmatic SEO. Indeed's SEO lead Chris Reynolds discussed using machine learning to power programmatic SEO at scale in a Page 2 Podcast episode (Page 2 Podcast episode listing (undated)). The episode listing does not carry a visible publication date, so this detail should not be tied to a specific year. The point stands regardless of timing: the long tail of title-plus-location pages is too large to write and too repetitive to template by hand, so the practical version leans on models to generate pages a small editorial team could never cover.

How the SEO team validated the pattern

What separates Indeed's approach from a typical content farm is that the company ran the pages as an SEO product discipline with designed experiments, rather than as a fire-and-forget page dump. Indeed is described in a vendor case study as the world's number one job site employing multiple teams of SEO specialists dedicated to ranking the site in Google Search (Indexing Insight (2026)). The number one framing is the vendor's, but the description of an in-house SEO product team running experiments matches the rest of the public record.

Hyper-local experiments on zero-volume keywords

The team ran designed experiments to test whether programmatic hyper-local pages targeting zero-search-volume, mid-funnel keywords could drive organic traffic (Indexing Insight (2026)). The phrase zero-search-volume is the interesting part, because it is counterintuitive: standard keyword research discards zero-volume terms. Indeed's team treated those terms differently, reasoning that each individual title-plus-city query may show no measurable volume, but the aggregate across millions of such queries is large. This is a testable claim about the long tail, and it is the opposite of chasing a handful of head terms.

Internal linking as a hard requirement

The experiments produced one finding that became a piece of internal evidence rather than just a ranking tactic. Indeed's SEO team found that pages without internal links were not being indexed, and they used that observation as evidence to persuade the engineering team of the importance of internal linking (Indexing Insight (2026)). In a programmatic system, orphan pages are the default failure mode: a template can generate a URL without anything linking to it, and a URL with no internal links is a page Google may never discover or choose to index. The finding turned internal linking from an SEO preference into an engineering requirement, the only way linking survives at scale.

Index coverage as the leading indicator

The team's north-star metric is worth copying directly. On index coverage, Indeed found near-100% coverage was easy at low page volumes, roughly the first 1,000 pages, but once pages grew past about 10,000 a consistent 5 to 10% stayed unindexed (Indexing Insight (2026)). The team uses its index-coverage rate as an early leading indicator to forecast success, before clicks, impressions and conversions follow (Indexing Insight (2026)). The lesson is that a program that is not getting indexed cannot generate traffic, and coverage moves first, so it is the earliest signal a new section will succeed or stall. Gus Pelogia, Indeed's Senior SEO & AI Product Manager, described constantly building and launching new pages and checking index coverage in daily email alerts (Indexing Insight (2026)).

Why it worked

Three conditions explain why the pattern matched the demand so well, and they are the parts to check before copying it.

First, the queries were explicit and the pages were exact. A person searching nurse jobs in Austin is stating a title and a place, and the Indeed page for that title and place is the closest possible match. There is no gap between what the searcher typed and what the page promises, which is the same alignment that makes a well-built directory page outperform a generic article. When the template's variables map directly onto the query's variables, the page is structurally the best answer before any authority is considered.

Second, the demand was already there and it belonged to the whole labor market, not to Indeed. People search for jobs by title and city whether or not they know Indeed exists, because that is how people describe the job they want. Indeed did not have to create the query or educate the market; it had to be the page that answered a query the market was already typing millions of times. That is a far cheaper position than building category awareness, and it is why the play scaled with the number of listings rather than with advertising spend.

Third, the pages sat at the moment of intent. A title-plus-city job search is a mid-funnel query: the searcher has already decided roughly what job they want and where, and they are looking for the listings that match. That is the step before an application, which made the pages valuable to employers, and it is how Indeed generates revenue by selling premium job posting and resume features to employers and companies hiring (Wikipedia (2026)). The organic pages and the paid product fed the same funnel, which gave the SEO program a commercial reason to exist beyond raw traffic.

There is a fourth, quieter condition: freshness. A job listing decays quickly, and a page showing current, open roles is more useful than one showing last month's postings. Aggregation gave the pages a steady supply of new listings, which kept the templates current without a writer refreshing each one, and the billion-document engineering work was the infrastructure that made that possible at scale (Indeed engineering blog (2013)).

What could break it

The pattern has structural vulnerabilities, and the public record now shows several of them acting at once.

The first is thin content. A page that differs from its neighbor only by a swapped city name and a re-sorted list of listings is exactly the kind of scaled content Google's quality systems are designed to devalue. Indeed's pages avoided the worst of this because the underlying listings are real, current and specific to the title and place, but the risk never disappears. Any copy of the pattern that generates pages without a real, fresh, specific dataset behind each one is generating thin pages, and those pages will not earn the authority that data-backed pages did.

The second is dependence on Google's referral traffic. Third-party estimates reported by AIM Group showed Indeed's total visits falling 10% globally and 9.4% in the United States between January and May 2024, figures that are estimates rather than company numbers (AIM Group (2024)). A directory built on aggregated, templated pages earns a large share of its visits from search, and when that referral narrows, the whole corpus earns less per page. The decline also matters because it coincides with the strategic shift described next.

The third is competition at the funnel. ERE reported that Indeed faces competition from programmatic job-advertising companies Appcast and PandoLogic, as well as from LinkedIn (ERE (2024)). Programmatic job advertising attacks the same employer budget from the paid side, while LinkedIn attacks the same job-search intent with a network that holds its own profile data. Neither of those competitors needs to win the title-plus-city organic page to take share, because they monetize the same demand through different channels.

The shift away from organic traffic

The most consequential recent development is strategic rather than algorithmic. In April 2024, industry reporting through Steve Rothberg's LinkedIn posts and the Chad & Cheese podcast said Indeed.com was potentially phasing out organic, non-paid job traffic in favor of sponsored postings (ERE (2024)). ERE framed this as industry reporting rather than an official Indeed announcement, so it should be read as a reported possibility, not a confirmed company policy. The significance for this case study is direct: the company that built one of the web's largest organic footprints in jobs was, by these accounts, considering walking away from part of it.

The stated reason in that reporting was Google's dominance in job indexing. Google was reported to have surpassed LinkedIn and Indeed as the largest non-paid job-traffic source, which inverts the position Indeed spent years building (ERE (2024)). When the search engine that sends the traffic also answers the query itself, the page that used to capture the click loses it. For a title-plus-city page, the threat is Google's own jobs features answering nurse jobs in Austin directly rather than sending the user to an aggregator's page. The company's 2024 traffic decline, as estimated by third parties, is consistent with that pressure even if the precise cause is not proven (AIM Group (2024)).

The strategic read is that programmatic SEO built the distribution, and then the economics changed. A page that monetizes through free listings is worth less once the same demand can be sold as a sponsored posting and Google captures the free-traffic role anyway. This is a reminder that the title-plus-city pattern is a distribution layer, not a business model, and the distribution layer is only worth maintaining while it earns more than it costs.

How to apply it

The copyable core is not the job title template itself; it is the sequence of decisions that made the template work. Any marketplace, directory or data product that has a title and a location, or any two combinable attributes, can run the same play with a smaller dataset.

First, find the query that states two variables and map it to a page. The Indeed equivalent is title plus city, but the same structure works for service plus neighborhood, practitioner plus city, or product plus region. The test is whether a searcher naturally types both variables in one query. Use the keyword research tool to list the variable combinations people actually search, and accept that many of the long-tail combinations will show zero or near-zero volume individually; the value is in the aggregate, not in any single row.

Second, get the data layer before you build the template. A title-plus-location page is only as good as the listings behind it, so the source of those listings, whether it is your own inventory, partner data or user submissions, has to exist and refresh before you generate pages.

Third, make the section a new URL path and plan for its ramp. Google needed time to establish trust in Indeed's new hyper-local section, and the same is true for any new path (Indexing Insight (2026)). Expect near-total indexing early at low volume and a stubborn unindexed tail as you scale, and treat internal links as a hard requirement rather than an afterthought. After you generate a section, run the sitemap checker to confirm the generated URLs are declared and reachable, which is the first step toward coverage.

Fourth, watch index coverage before you watch traffic. Indeed's team uses coverage as the leading indicator that forecasts clicks and conversions, and it is the earliest signal a new section will work (Indexing Insight (2026)). Check the actual pages rather than the template by running one through the on-page SEO checker to confirm the title, heading and canonical structure resolve correctly, then monitor coverage as the section grows.

The pattern also sits inside a larger set of decisions covered elsewhere in this library. The keyword research guide covers how to build the title-plus-location query set without chasing head terms, and the technical SEO guide covers the crawl, canonical and sitemap work a programmatic section needs to index at scale. For the adjacent case where a company turned structured salary and job data into pages, the Glassdoor salary pages case study follows the same data-to-page logic inside the same Recruit family of companies. The local-directory version of the same two-variable pattern appears in the Zocdoc doctor directory case study.