Programmatic SEO: one page template, one page per long-tail query

Programmatic SEO means you build one page template and fill it from a dataset, so each long-tail search query gets its own page. The one cited case is Avvo. It published rated profiles for 97% of practicing US lawyers, and by 2018 it drew more than 100 million annual visits. Its sources never say how much of that traffic came from the profile pages.

How it works

1. Find a query pattern with many variants, such as 'divorce lawyer in [city]' or '[tool] alternatives'. Each variant gets little search volume, but the full set is large.
2. Build or license the data that fills each page. Every row needs something unique to say, not just a swapped keyword.
3. Design one template. It needs a title and heading built from the query, the data block, and supporting text that answers the searcher's actual question.
4. Generate the pages. Connect them through hub pages, for example by city or category, and submit a sitemap.
5. Track indexing and clicks. Prune or merge pages that never get indexed or clicked.

Avvo, a legal marketplace founded in Seattle in 2006, matches this shape. It kept a rated profile for almost every practicing lawyer in the country. It also ran a free forum with more than 11 million questions and answers. In our reading, the profiles gave a page for each lawyer name, and the forum added original text that a bare template cannot supply. That pairing would work like a content growth loop only if visitors' questions and lawyers' answers became indexed pages, which the sources do not say. Our guide to growth loops covers how to draw one.

Internet Brands agreed to acquire Avvo on January 11, 2018. The Internet Brands acquisition release put Avvo at more than 100 million annual visits across its website and apps. It also counted more than 300,000 attorneys as active users. Stanford Law School's note on the deal repeats that attorney figure. The full write-up is on the Avvo case page.

What it takes

Our estimate: about 60 hours of work, and about 45 days before you can read a signal. Most of the hours go into four jobs: choosing the query pattern, cleaning the dataset, building the template with internal links, and checking a first batch of a few hundred pages by hand. The signal is slow because search engines find, crawl and index new pages over weeks. Rankings also keep moving for a while after a page is indexed, so early readings are noisy.

What to measure

Indexed share comes first. A low share tells you the engine sees the pages as thin or duplicate, and no other metric matters until you fix that. Clicks per indexed page, compared against a holdout batch, tells you whether the template earns traffic or only adds pages. Tie this one to your north star metric, not to raw sessions. Impressions at day 30 versus day 60 shows the direction of travel. Rising impressions with flat clicks points to weak titles or the wrong intent. Falling impressions points to a quality problem.

When it backfires

• Thin pages. The keyword changes but the text does not. Engines skip indexing these, and a large batch of weak pages can hurt how the rest of the site is treated. Our list of tactics that stopped working covers similar platform crackdowns.
• Errors at scale. One bad field repeats across thousands of pages. A directory of real people, like Avvo's lawyer profiles, carries extra risk, because a wrong profile damages trust with the person it describes.
• Traffic that never converts. A large visit count is not revenue. Avvo's figure of more than 100 million visits is a traffic number, and its sources tie no signups or sales to the pages.
• Wrong intent. The query pattern draws researchers or job seekers, not buyers.
• No unique data. If competitors can generate the same pages from the same public data, nobody wins the ranking.

Run it as an experiment

Hypothesis: pages built from our template for query pattern X will reach a useful indexed share and bring signups at a lower cost than our paid channels.

Comparison: we recommend you pick the eligible query variants, randomly assign half to a first batch, and hold the rest back. If you want to test two templates, assign pages at random to each version and compare them over a fixed 60 day window. Do not compare against a moving site average.

Log for every page: publish date, first indexed date, impressions, clicks, average position and signups. Our guide to running a growth experiment has a write-up format, and the experiment backlog template helps you rank this test against other ideas.

Decision rule, our recommendation: at day 60, expand to the held-back variants if at least half the batch is indexed and cost per signup beats your paid channel. If the indexed share is low, rework the template before you add pages. If traffic is fine but signups are not, change the query pattern.

How strong the evidence is

The grade is case with numbers: at least one company documented the tactic and published results as figures. Avvo carries the grade alone, through the visit and user counts in its acquisition announcement. The evidence has clear limits. The figures come from the companies in the deal, there is no baseline or growth timeline, and the visits are not split by page type. The acquisition itself is not a result of the tactic. Treat this page as a well-reasoned bet, and test it.