SEO for construction companies in 2026: what works, what has stopped working and what AI search changed

Search engine optimisation for a construction company used to mean one thing: rank on page one of Google for “groundworks contractor Leeds”. That still matters, but it is now one of three jobs. The second is being named when a buyer asks ChatGPT, Gemini or Google AI for subcontractors in your trade and region. The third is surviving the check that follows, when the buyer opens your site and looks for the accreditations the AI said you had. This guide covers what construction SEO looks like in 2026, what has stopped working and the order to do things in.

What changed

Three things. Google’s results for trade and region queries now carry an AI Overview or AI Mode answer above the links, built from sites it can read as facts. Buyers use assistants directly: G2’s March 2026 research found half of business buyers start supplier research in a chatbot, and Gartner puts generative AI use in a recent purchase at 45%. And procurement is more standardised, with the Common Assessment Standard and PPN 03/24 setting what buyers check, so the evidence SEO has to surface is the same evidence a PQQ asks for.

What no longer works

• Bought links from directories and guest post farms. They are discounted at best and penalised at worst.
• Location pages that are the same text with the town name swapped. Google folds them together; AI tools ignore them.
• Keyword stuffing: “construction company Newcastle construction services Newcastle”. It reads badly to a QS and signals nothing to a model that reads meaning.
• Logos instead of text for accreditations. Invisible to search engines and assistants alike.
• Blog posts about nothing. “Five tips for a safe site” earns no links, no citations and no tenders.

What works, in order

1. A page per trade and a page per sector, each with scope, values, accreditations relevant to that work and two case studies. This is most of construction SEO on its own.
2. Accreditations written out and linked to the registers, with structured data stating them. The accreditations guide gives the format.
3. Business details identical across your site, Companies House, Constructionline, CHAS and Google Business Profile. Inconsistency is the commonest reason an AI tool describes a firm wrongly or not at all.
4. Case studies with value bands, programmes and sectors. They are the pages that get cited, because they contain facts.
5. A Google Business Profile with your trades as categories, your service area and reviews from contracts managers that name the project and the trade.
6. Regional pages with substance: the main contractors, frameworks and programmes active there and the packages you have delivered. One page per region you genuinely serve.
7. Structured data and an llms.txt file, so assistants read the site as a set of facts. Almost no subcontractor sites have these; it is the cheapest edge available.
8. Guides that answer buyers’ questions, like this one. They earn links from trade bodies and citations from AI tools, and they bring the right reader.
9. Speed on mobile. A site that takes five seconds on a site connection loses the estimator and the ranking.

How to measure it

Rankings for trade plus region in a private window, monthly. The ten buyer questions in the AI visibility audit, monthly, with a record of who is named. Enquiries, with the trade and value of each, so you know what a ranking is worth. Our website management plan does all three and sends a one-page report; the pricing page has the cost.

What to do this week

• Search your trade and region in a private window and note who holds the top three and whether an AI answer sits above them.
• Ask ChatGPT the same question and see whether you are named.
• Run the free Tender Check; the Found stage is the SEO score.