Programmatic SEO

Programmatic SEO is building a page family off one template, populated by a dataset. Think city pages, model pages, or comparison pages at scale.

It earns when every page has data or an answer of its own that a reader would want. Example: /guides/espresso-machines/breville-bambino vs /guides/espresso-machines. The model page lists dimensions, pressure, parts, stock at your store, and support, not just a spun blurb.

• Pick one search intent per family, like “coffee subscription in London” or “compare X vs Y”.
• List the required fields that satisfy that intent. For a city page: delivery area, cut-off times, shipping fee, stock by postcode, local testimonials, support hours.
• Decide the unique key for each page: city name, product model, integration name, or route id.
• Sketch the template sections: summary, key facts table, pricing, FAQs tied to the entity, internal links to siblings and parent.
• Write the first page by hand, then fill the dataset. If the hand-written version is stronger, your fields are right. If not, add fields.

A good page looks like /locations/london with an H1 “Coffee subscription London”, a facts table with delivery windows, £ amounts you actually charge, and reviews from London buyers, plus links to /locations/manchester and /pricing.

• Avoid scaled content abuse: do not ship pages that only change a city name or a synonym. Add verifiable, page-level facts.
• Avoid site reputation abuse: do not host third-party templated posts on your domain to borrow your authority.
• Do not buy expired domains to rank templated pages. That is expired domain abuse.
• Use the same template, but keep the copy blocks fact-based, not spun. Google does not penalise AI-assisted writing as such, it acts on pages made at scale without value.
• Set canonical URLs for near-duplicates. Consolidate thin variants into one useful page instead of many weak ones.

The helpful content signals were folded into core ranking in March 2024. Quality is now baked into core updates, so weak page families can drop sitewide.

1. 1

   Baseline

   In Search Console, note impressions, clicks, and queries for the parent section, like /locations/.

2. 2

   Indexation

   Submit the XML sitemap. Check index coverage and that canonical URLs are respected. Fix any “Alternate page with proper canonical” mismatches.

3. 3

   Queries and intent

   Watch the Performance report by page and by query. Each page should earn queries with its entity in them, like “coffee subscription london”. If you only get generic terms, your fields are not specific enough.

4. 4

   Click-through rate

   Keep titles under about 60 characters. Write meta descriptions that reflect the entity fields. Compare CTR across siblings to spot weak pages.

5. 5

   Cannibalisation

   If multiple siblings rank for the same head term, tighten titles and internal links. Link up to the parent hub and across siblings with clear anchors.

Impressions may rise before clicks. That is normal while Google tests variants. If a page shows but is never clicked after 28 days, improve the fields or merge it.

• Thin fields: two numbers and a paragraph. Add stock, pricing, support, availability, delivery, compatibility, or local proof.
• Mass publication: thousands at once. Ship ten, measure, then batch in cohorts you can review.
• Duplicate titles: only the city or model changes. Add a unique fact to the title when possible.
• Slow templates: heavy maps and scripts on every page. Aim for LCP under 2.5 seconds, CLS under 0.1, and INP under 200 ms.
• Orphan pages: no links from the parent. Build a hub, facet links, and a sitemap.
• Stale data: outdated prices or unavailable stock. Hook fields to live sources or remove the page until accurate.