Sierra Window & Property Services

This is the steadiest month the account has had. Total spend moved by $72.87, Google Ads sat within $18 of last month and within $14 of the same window in 2025, and Meta rose 8.9%. After a summer of changing the way both channels work, the useful thing this month is that nothing had to be changed, and the numbers still improved underneath.

The comparison that carries the story is the year. The account is buying 72% more clicks than it did in August and September 2025 for 15.6% more money, which is a blended $1.23 a click against $1.83 then. One note on the month-on-month figures: last month’s Meta total included the technician job ad, which finished on August 7. Setting that aside, the service campaigns themselves went from $545.79 to $676.55, so they grew by about a quarter. The sections below break each channel down.

Impressions and reach are both down about 30%, and that is the campaigns working as intended rather than a drop in delivery. Optimising for landing page views tells Meta to stop buying the cheap, broad impressions that never turn into anything and spend the money on the smaller group who will actually tap through. It showed the ads to 30% fewer people and got 18% more of them onto the website. The click-through rate went from 2.5% to 3.6% on the back of it.

Two of last month’s figures are not like-for-like, and both are worth naming. The previous window still contained the technician job ad, which ended on August 7, so the spend comparison flatters last month slightly. And a fortnight of the previous window was still running the even-split structure. The year-on-year column is the clean one, and it is strong throughout: 86% more clicks, 89% more website visits and a cost per visit down from $0.81 to $0.65.

Comparing the last thirty days against the thirty before, a website visit now costs 10.6% less in Whistler and 20.3% less in Squamish. The mechanism is in the two lines underneath: the cost of a visit is the cost of being seen divided by how many of those views become a visit. Meta paid more per thousand views in both towns, up 30% and 37%, and turned far more of them into visits, up 45% and 72%. It is buying dearer attention from much better matched people and winning the trade comfortably.

Squamish is the one that changed most. It was the town where the even split had made Video 4 look weakest, and it has now improved twice as fast as Whistler. That is the consolidation doing the work as much as the choice of video: three ad sets each learn from a third of the data and bid against one another, one ad set learns from all of it. It is a good argument for leaving the structure exactly as it is.

Worth being straight about one thing. Some of this gain is the campaign simply having been running longer, because an ad set that has been learning for two months is better matched than one that has been learning for two weeks. That effect flattens off, so the honest expectation is that the cost of a visit settles somewhere near where it is now rather than continuing to fall at this rate.

The Local Services listings are the standout on value again, bringing 55 clicks at $2.09. That is 27% of every click in the account for under 10% of the budget, and it is where the Google Guaranteed badge shows. Whistler window cleaning is the most competitive term at $9.66, which is what the end of the summer glass season looks like in the auction.

Snow removal is the campaign to watch. It took only 75 impressions in the window because the searches have not started yet, while four of the ten website enquiries since we last met were for snow, and the team has separately quoted a machine clearing job at $1,650. The demand is arriving ahead of the search volume, so the decision for the call is how early and how hard to lean into it.

Whistler carries most of the visibility with 356 profile views, and it is the profile that converts to contact, taking all 6 calls and 21 of the 25 website clicks across the three. Squamish and Snow Removal are smaller but pull heavily on directions, 57 and 30, which is people navigating to a job rather than browsing.

The Snow Removal profile is the one to watch from here. It is already drawing 136 views and 30 direction requests before the season has started, and it is the profile ChatGPT names by title when asked about seasonal driveway work. It is also the template for how the lights profile should be built once it has a phone number of its own.

Snow is now the most requested service on the form, four of the ten, and it has overtaken window cleaning for the first time. Three of those four are driveway contracts for the coming winter and one is the commercial building asking for shovelling. Those are multi-winter relationships being decided now, months before the first snowfall, and they match what ChatGPT is telling people to do.

The enquiries stay traceable end to end, which is the strongest thing in this report. One submitted the form at 12:06 on August 21 and was on the team’s own estimate board thirty minutes later. Another came in on August 24 and was assigned the following day with the job booked for completion by the 28th. A third submitted on August 13 and left a five-star Google review on August 21, which is a complete run from search to enquiry to job done to public review inside eight days.

The tool is doing the job it was built for. It puts a number in front of people at the moment they are interested, and three of the five were returning customers, which suggests it is being used as a quick way to re-book as much as a way to shop around. The largest quote, $1,400 for a roof clean, is a firm figure rather than a range because roof cleaning is fully priced in the tool.

That is also the clearest argument for filling in the rest of the pricing. Window cleaning currently has one priced band, ten to twenty windows, so three of the five quotes landed on the same $200 to $355 range and two had extra services routed to a custom quote instead of a number. The pricing sheet is with you, and every gap filled is another quote that lands as a price rather than a follow-up.