Digital marketing for windows and doors companies UK

The surveyor’s van is already booked for a 1930s terrace in Solihull when the next enquiry comes in: “uPVC windows, three-bed, quote only.” You can tell in ten seconds whether that lead will pay for a full replacement or die as a tyre-kicker comparing £89 “from” prices on a comparison site. That split, not impressions, is the real job of digital marketing for windows and doors companies in the UK.

Owners and marketing leads in this trade do not need another lecture on “being online.” You need a mix that fills the measure diary with homeowners who have a real job (failed units, planning a kitchen, right-to-buy renovations, new-build snagging) and a cost per qualified survey that still works after VAT, FENSA paperwork, and the salesperson’s time on the drive.

This piece is a working map for that mix: search intent, paid media, local proof, landing pages, and tracking that survives privacy loss. It is written for the person who owns the marketing number, not for the homeowner browsing colours of casement.

Where UK window and door enquiries die before anyone books a measure

Most waste starts before the click even looks expensive. CPCs in home improvement Search campaigns have been running £2.50-£4.00 depending on geography and match type, with year-on-year inflation across the category. Twenty to thirty percent of SEM budgets still leak on irrelevant queries, neglected negatives, or a homepage that talks about “quality craftsmanship” while the ad promised “replace blown double glazing in Birmingham this week.” Mobile takes 63%+ of paid search clicks, yet conversion rates lag desktop by 30-40%. If your quote form is a five-field maze on a 4G connection in a conservatory, you are paying Google for traffic that never becomes a diary slot.

AI Overviews now sit on more than 20% of searches and can cut organic CTR by nearly 60%. Homeowners still type “composite doors near me” and “triple glazing grants,” but a chunk of those sessions never leave Google. Ranking without a click is not a vanity problem. It is a pipeline problem if your only proof lives on a slow brochure site instead of a page that answers size, lead time, guarantee, and next available survey.

Attribution is the other quiet killer. Teams lose 20%+ of conversion data to ad-blockers and consent banners, which starves smart bidding. Last-click then double-counts the same booked measure across Google Ads and Meta. In-platform CPA looks healthy while the CRM shows tyre-kickers. Google Ads average ROAS sits around 4.2x across industries; Meta sits nearer 2.8x. Those are not your numbers. A replacement-window job with a long consult cycle cannot be judged on last-click ROAS from a £40 click on “cheap windows.”

Creative fatigue shows first as rising CPMs, not as a sudden drop in form fills. UGC-style clips that worked on day one can die in five or six days. If you keep stacking 2024 lookalikes instead of treating creative as targeting, Meta will keep finding people who want a catalogue, not a surveyor.

If paid search is carrying the diary and the landing page still dumps everyone on the homepage, tighten the handoff before you raise budget. HeyLead’s Google Ads and paid search work is built around that click-to-measure path, not a media plan that stops at the session.

A UK channel mix that books surveys, not brochure sessions

Lead with high-intent search. Homeowners in Leeds, Bristol, and the Home Counties still use Google when a seal fails or a quote is due for a loft conversion. Structure campaigns around jobs, not product PDFs: replacement casement, tilt-and-turn, composite front doors, bi-folds for extensions, commercial fire doors if you actually sell them. Keep brand separate. Protect it. Then run non-brand on the phrases that mean “I need someone on site.”

Performance Max will try to take the bulk. That is fine if conversion signals are clean and you are not feeding it “brochure download” as a success. PMax in well-run accounts can land in a 4x-8x ROAS band. In a messy account it will harvest branded searches and cheap catalogue clicks. Pair it with Search that you still control, and with negatives that kill “DIY uPVC trim,” “spare keys,” and “job vacancies.” AI Max-style campaigns can cut CPC 12-18% versus standard Search, but only if the landing page matches the query and Enhanced Conversions is actually firing.

Organic still matters because zero-click search is real: a large share of Google searches end without a click. You need pages that can be cited and still convert when someone does click. AI-assisted copy is now common across top-ranking glazing pages, which means generic output no longer differentiates-Google’s March 2026 core update rewarded genuine information gain. That is not a licence to publish filler. A 900-word essay that could sit on any glazing site will not beat a page that names typical lead times in your depot, FENSA / Certass handling, and what happens on the measure visit.

Meta is the demand layer for people who are not searching yet: new-build snagging, “the conservatory is freezing,” a landlord replacing a block of flats. Run broad. Creative is your targeting. Short clips of a surveyor on a real elevation beat stock patio shots. Do not promise a ROAS or CPL before you have seen the account, the offer, and unit economics. Realistic lift takes 3-6 months, not one Reel.

Only 55.9% of origins pass all three Core Web Vitals in May 2026 CrUX data. INP on a quote form is a sales issue. If the page stutters when someone opens the colour picker, paid and organic both pay a tax. Use dedicated landing pages aligned to the ad. If an agency says they just send traffic to the homepage, that is the moment to walk.

The survey-booking playbook: search, pages, and signals

Here is the operational sequence a marketing lead at a UK glazing firm can run without waiting for a rebrand. It is checklist-grade on purpose. Overwhelm here is useful: this is the work that has to stay done every month, not a one-off “campaign.”

Paid social quality still depends on the page after the Reel. If Instant Forms fill WhatsApp with “how much for one window” at 22:40, send cold traffic to a page that asks for postcode and openings instead. One sentence of proof next to the button beats another 40 seconds of factory B-roll.

Two UK operators, two broken mechanisms, two fixes

A marketing manager at a Midlands fabricator was buying “windows” as a broad match theme and sending everything to the homepage. Cost per enquiry looked acceptable until the sales director counted diary slots. Eleven of every 14 forms were people wanting a single pane of glass or a job. The fix was not a new platform. They split Search into replacement vs repair, added negatives, and stood up a replacement-only page with a measure calendar embed. Booked surveys did not triple overnight. The junk rate fell hard enough that the same spend filled Fridays that used to sit empty.

A South Coast installer ran Meta with stacked lookalikes of “people who liked conservatories.” CPMs climbed for nine days while form volume held, which is the fatigue tell. They switched to broad, cut to two campaigns, and tested a 7-second clip of a surveyor explaining blown units on a 1970s estate. Creative became the targeting. Lead quality moved because the hook selected the problem, not because a spreadsheet of interests got more clever.

Neither team needed a guaranteed ROAS slide. They needed tighter positioning, a clearer offer, and tracking that counted a measure, not a thank-you page. Adding TikTok or another network would have multiplied the same leak.

If organic is the gap (thin service-area pages, no information gain versus the national catalogues), the same discipline applies: pages that a human would send a neighbour, published on a cadence you can sustain. For firms that need that cadence without a huge writing bench, HeyLead Auto Blogger is the managed publishing layer under human strategy, not a licence to ship generic glazing essays.

What marketing leaders are seeing

One fenestration manufacturer’s head of marketing found that switching the conversion event to booked measure wrecked in-platform CPA for a fortnight-then Fridays filled. Optimising for form volume had been filling the CRM with tyre-kickers comparing £89 from-prices; counting diary slots changed what the algorithms were trained to buy.

A regional installer saw the same pattern on Performance Max: the campaign looked like 4x until brand was excluded and non-brand was barely breaking even. The leak was the homepage, not the bid strategy.

FAQs

Should a UK windows company put most budget into Google Ads or Meta?

Put the first serious money on high-intent Search and a matching landing page. Use Meta to create demand and retarget people who bounced before booking a measure. Google’s industry-average ROAS is higher than Meta’s; your mix should follow booked surveys, not the prettier dashboard. A reasonable starting split for a regional installer is 70% Search / 30% Meta until you have 30+ booked-measure conversions in 90 days. Once signals are clean, test shifting 10-15% toward Meta retargeting of page visitors who didn’t book.

Is Performance Max enough on its own?

No. PMax can do the bulk once signals are clean, but you still need Search you control, negatives, and brand protection. Constant structure changes reset learning. Give it jobs as asset groups, not one campaign for every product in the factory.

Why do CPCs keep rising if conversion rate is “fine”?

Auction pressure is real, and mobile conversion often lags. If you only watch blended conversion rate, you hide a mobile form that fails. Fix INP, shorten the form, and stop paying for repair and DIY queries before you assume automated bidding is the villain.

Can we guarantee a cost per lead?

No honest partner can guarantee ROAS or CPL before seeing the account, pages, and unit economics. Plan on 3-6 months of signal cleanup and creative testing. Anyone promising a rocket from the first Reel is selling a timeline that will churn you.

Do we still need SEO if AI Overviews steal clicks?

Yes, because the remaining clicks are more valuable and Maps still drives measures. Write pages with genuine local detail and citation-worthy answers. Volume of sessions is a weak metric. Booked surveys from organic plus branded Search is the one that pays the factory.

Putting it to work

This week

• Pull 60 days of enquiries by source and mark booked measures versus quote-only junk.

• Write negatives for DIY, jobs, cheap, and single-pane repair if that is not your product.

• Check Core Web Vitals on the three money pages and fix the worst INP issue on mobile.

• Align one ad group’s headline to one landing H1. Kill homepage traffic from that group.

• Confirm Enhanced Conversions on the booked-measure event, not the raw form.

• Replace one fatigued Meta ad whose CPM rose while CTR went nowhere.

Next 30 days

• Stand up or rebuild dedicated pages for replacement windows, composite doors, and bi-folds.

• Give PMax job-based asset groups and a brand exclusion if Search already owns brand.

• Run a proof test next to the form (reviews, guarantee, photo of a real elevation).

• Set a weekly view of cost per booked measure, not last-click ROAS alone.

• Use the UTM builder and Insights-style behaviour review so you can see where surveys leak after the click.

Start by tagging last month’s forms as booked measure versus tyre-kicker, then freeze any campaign still optimising for the thank-you page. When you want the click-to-diary path (search terms, landing pages, conversion signals, and creative that selects real jobs) run as an ongoing programme rather than a side project, HeyLead can take that operational load so you stay on the marketing number. Chat with us on WhatsApp