Why Australian window companies leak quotes after the click

Nobody in Melbourne thinks about double glazing in February. They think about it on a July night when the heater is running flat out, the bedroom window is wet with condensation and the draught under the sliding door is back.

That is when they search, and they are not ready to buy yet. They want to know whether uPVC or aluminium is better, what a WERS rating means, whether the installer will still be around in ten years, and, in bushfire-prone areas, whether a door meets the right BAL rating. A form that only asks for a name and a phone number turns all of that into a pile of “how much for a window?” messages, and pouring more money into winter ads just makes the pile bigger.

The businesses that turn cold-weather searches into site measures answer those worries before the first call: a plain explainer on energy ratings, photos of real installs in nearby suburbs, an honest idea of how long the job takes from quote to install, and a form that asks what the buyer is actually trying to fix. The site measure then feels like the obvious next step, not a sales visit.

Window and door companies in Sydney see the quote leakage described in this article in its sharpest form, because high click costs make every abandoned enquiry expensive. Fast quoting, clear pricing cues and follow-up discipline close the gap. Target location: Sydney. The Sydney local firms page is the location overview for firms in Sydney.

Storm weeks fill inboxes. They do not fill measure vans

After hail in Sydney’s west or a week of southerlies in Adelaide, search volume jumps for “broken window repair,” “emergency glass,” and “replace sliding door track.” Plenty of teams treat that spike as a growth event and dump extra budget into broad match. What you actually get is a mix of insurance jobs, DIY hardware shoppers, and people who needed a glazier yesterday, not a custom window company with a six-week lead time.

The marketing failure is intent mixing. Emergency glass and full-home replacement sit on the same campaigns, the same landing page, and often the same “Get a quote” button. Sales then spends Tuesday ringing people who wanted a $90 pane, while the family in Toowoomba who needs twelve double-hung units with flyscreens goes cold. You did not have a lead-quality mystery. You had one offer serving two jobs.

Mobile makes it worse. More than 63% of paid search clicks now come from phones, and conversion rates on those devices still lag desktop by roughly 30-40% when the form asks for room counts, frame colours, and photos before anyone offers a measure slot. A homeowner standing in a wet laundry will not upload three elevations. They will bounce, then later search a competitor who promised “book a 45-minute measure this week.”

Creative fatigue shows up the same way on Meta. CPMs climb first. Lead volume holds for a few days, then the WhatsApp queue fills with “how much for one window” while your surveyors sit idle on Thursday. UGC that looked strong on day one can die in five to six days. If you are still running a 2020 lookalike stack and hoping the algorithm “finds homeowners,” you are paying for scroll-stopping video that never names the next step: a booked measure, not a catalogue PDF.

Attribution will not rescue you if the event is wrong. Last-click will credit the branded search that happened the morning the surveyor arrived, and it will ignore the Google ad that started the conversation 19 days earlier. In HeyLead accounts across home-improvement verticals, Search typically returns 4-6x when the conversion is a measure, not a form; Meta sits lower and falls faster when creative ages. Those averages mean little if your conversion is a form, not a diary slot with a postcode and a product family. Lose 20%+ of conversion data to ad blockers and privacy changes and smart bidding starts optimising for the wrong people.

If your paid traffic still lands on a homepage of 40 product photos, that is the leak. Dedicated pages aligned to the ad, with a single measure CTA, are the baseline. If you want a partner to rebuild that path rather than another round of keyword lists, HeyLead’s SEM programs start from the measure event, not the impression.

Intent clusters that actually book a site measure in AUD markets

Australian window and door demand clusters in a handful of jobs, not in “windows” as a category. Replacement of failed seals and fogged IGUs. New-build or renovation packages for builders. Energy-rating upgrades before winter in VIC and TAS. BAL-rated and security doors in QLD and WA. Commercial aluminium suites for small developers. Each cluster needs its own query set, ad copy, and proof.

Search still needs keywords even as Google pushes signals. Treat keywords as intent labels, then feed the account first-party events: measure booked, measure completed, quote sent, deposit paid. Enhanced Conversions and a clean GTM setup matter more than another ad group for “aluminium windows Sydney.” Performance Max will drive bulk spend in many accounts. Let it, but only after the conversion is a qualified measure, not a brochure download. Unoptimized ones buy cheap clicks from people comparing flyscreen mesh.

Organic still has a job while AI Overviews sit on a growing share of queries and cut CTR hard when they appear. SparkToro data suggests roughly half of Google searches in English now end without a click - AI Overviews are accelerating that in home-services queries. Australia is not immune to zero-click behaviour. Rankings without a measure CTA and a Google Business Profile that actually offers a booking path are vanity. Local pages for “double glazing Ballarat” or “security doors Townsville” need unique proof: WERS tables, BAL notes, installer photos from that suburb, review snippets that mention time-on-site. Thin AI without installer language will not earn citations, and information gain after the March 2026 core update is what moves the needle, not word count.

If your configurator or gallery stalls INP on a phone, you will lose the homeowner who was ready to book. Check that before you raise bids. Use the Core Web Vitals checker on every measure-booking URL, not just the homepage.

Meta still works when creative is the targeting. Run broader structures, fewer campaigns, and a testing cadence that replaces ads before CPM drift becomes a lead-quality problem. Short video should show a measure, a reveal, and a specific product (uPVC tilt-and-turn, BAL-40 door, commercial shopfront), not a lifestyle montage of a kitchen. Instant Forms without a follow-up path to a calendar will fill a spreadsheet and empty a van roster. Send high-intent traffic to a page, then watch where they stall with HeyLead Insights so you are not guessing whether the energy-rating table or the 14-field form killed the booking.

A measure-first playbook you can run in the next fortnight

Do not rebuild the whole site. Separate emergency glass from replacement programs, fix the conversion event, and give each cluster a page that can book a measure in AUD, with GST-inclusive language and a realistic lead time. Below is the audit we would run on an Australian windows and doors account that is “getting leads” but not filling vans.

A Brisbane aluminium shop ran Search and PMax into one “windows and doors” page for 11 months. Forms were cheap. Measures were not. They split BAL-rated doors onto their own URL, shortened the form to suburb plus product, and moved the conversion to “measure booked.” Cost per form rose. Cost per completed measure fell enough that they could raise bids in two postcode clusters without the sales manager screaming about tyre-kickers. The mechanism was the event and the page, not a new bidding algorithm.

Another operator in regional VIC kept Meta Instant Forms because the volume looked healthy. Sales was drowning in “send a brochure.” They killed the form, sent traffic to a measure page, and used session behaviour to drop the energy-rating accordion that nobody opened. Enquiry volume dropped 22%. Booked measures did not. That is the trade you want.

If organic content is the bottleneck (suburb pages, product explainers, winter energy pieces), cadence matters more than a one-off rewrite. HeyLead Auto Blogger is built for that kind of region-and-niche publishing under human strategy, not for dumping interchangeable AI onto every city page.

Patterns we see in accounts

When completed measures are tagged, a $41 CPL on Search often collapses into two postcodes carrying 80% of the diary, with the rest trained as catalogue traffic the algorithm keeps buying.

On Meta, CPMs on Reels can climb for four days while lead count holds; by day six the WhatsApp queue is price-only. The accounts that stay healthy kill a creative when CPM moves, not when the form count falls.

FAQs

Should Australian window companies still use keywords if Google wants signals?

Yes. Keywords label intent. Signals (measure booked, deposit paid, Enhanced Conversions) tell bidding which of those intents paid. Dropping keywords entirely usually dumps emergency glass and DIY queries into replacement campaigns.

Is Performance Max safe for windows and doors?

It is safe when the conversion is a qualified measure and brand exclusions and URL controls are in place. If PMax can hit a thin catalogue URL, it will. Feed it the three cluster pages, not the homepage.

Why does Meta look cheaper than Google but fill fewer vans?

Meta’s average ROAS sits lower in 2026 benchmarks, and creative dies fast. Broad targeting works only if the creative and the landing path qualify for a measure. Cheap Instant Form leads are not a channel win.

How long before this looks like ROI to a sceptical owner?

Honest programs take 3-6 months to show a stable cost per measure, especially with a four-to-eight-week sales cycle. Anyone guaranteeing a CPL before seeing your account, pages, and unit economics is selling a number they cannot own.

Do we need a new website first?

You need measure-specific landing URLs, fast mobile forms, and clean events. A full redesign can wait. If Core Web Vitals fail on the booking page, fix that before you add channels.

Putting it to work

Go back through last winter’s enquiries, June to August, and mark each one as measure booked, price check, or no fit. Note what the booked ones asked about first: warmth, noise, security, bushfire rating or price. Then rewrite the top of your main quote page so it answers that first question before it asks for anything. It is an afternoon’s work, and it will shape every ad you run before the next cold snap.

If turning cold-snap searches into booked measures is the work that keeps slipping, HeyLead can take it on. Chat with us on WhatsApp