Keep the hail work in its own account, with its own phone number and its own budget
Storm-driven glass and screen inquiries behave nothing like retail whole-house replacement, and blending them corrupts the measurement of both.
After a hailstorm the search volume climbs, clicks get cheap, and cost per lead drops across the board. It looks like a good week. Most of what arrives is a broken pane, a torn screen or a question about a claim.
If those conversions sit in the same campaigns as your retail work, automated bidding learns toward the cheap ones. You end up buying more of the least valuable inquiry in the market and calling it efficiency.
Split it. Separate campaigns at minimum, a separate tracking number so whoever answers knows what kind of call is ringing, and a separate budget line that can be capped without touching retail spend.
Keep storm ad copy clear of anything that sounds like handling a claim. Point homeowners to their own insurer. In a field with plenty of operators who show up after a storm, restraint reads as credibility.
If you do not want storm work at all, exclude those queries deliberately rather than paying for them by accident.
Your negative list starts with evaporative cooler pads and ends with rebate hunters
A large share of window-related search in this metro is maintenance, parts and program shopping, and none of it turns into an order.
The Denver-specific entries first: swamp cooler pads and parts, whole-house fan questions, window well covers, storm windows. Then the usual drains, which are window cleaning, window tint, screen repair, glass repair, blinds, sash cords and anything about the operating system with a similar name.
Utility program searches deserve a decision rather than a default. People searching for an Xcel Energy rebate want the program, not a dealer. You may still want some of that traffic, but send it to a page that answers the program question honestly and offers a quote second, and never publish eligibility claims or amounts you have not confirmed with the utility.
Builder bids, new construction and rental turnovers should be spending somebody else's budget unless you actively serve them.
Read the search terms report weekly, and daily in the week after a storm. That is when the junk arrives faster than a monthly review can catch it.
Denver County is one jurisdiction and everything around it is five more, so split the campaign that way
The city and county of Denver, plus Jefferson, Arapahoe, Adams, Douglas and Boulder, are separate permitting worlds with different housing stock, and a single metro radius pretends they are one place.
Draw the geography around the counties and towns your crews actually serve rather than a ring around the yard. A radius from a Thornton address will happily sell you clicks from Castle Rock and from up in the foothills past Golden, which is a different day for an install crew.
Set targeting to people in your locations rather than people merely interested in them, so you are not paying for someone in another state researching a house they have not bought yet.
Bid by what the job usually is, not by what the neighborhood is worth. An older home with a dozen openings and a two-window townhome are different orders, and the housing pattern in an area is a fair proxy for which you are likely to get.
Exclude the places you will not drive to, in writing, and revisit the list when a crew moves or a second yard opens. Geography set once and never checked is where quiet waste lives.
February installs fine in Denver, so do not run a national seasonality curve
Cold, dry, sunny weeks here are workable, and a budget that goes quiet after the holidays hands the quarter to whoever stayed on.
National playbooks pause winter spend because they were written for markets where winter stops work. Denver winters are cold but often dry and bright, and replacement work continues through much of it. Confirm your own crew's limits, then set the budget to them rather than to a template.
The real constraints are a deep cold stretch, a snow day and the manufacturer lead time. Pace against the order book: if crews are booked out three weeks, you are buying week four, not stacking week one.
Storm weeks need a hard cap. Without one, a single hail night can spend a monthly budget in a few days on inquiries you did not want.
Paid social sits alongside this differently than search. It reaches the household that has been putting off the west-facing room for three summers and has not searched yet. Treat it as a separate budget with a slower read, not as a substitute for search.
Put the lead time in the ad and let the impatient clicks go elsewhere
Ordered units take time, and an ad that says so buys fewer clicks and better ones.
Name the sequence in the copy: measure, order, install. Someone who needs glass by Friday will keep scrolling, which is exactly what you want, because that call would have cost your office fifteen minutes and closed nothing.
Say what you install and where. Say who pulls the permit, since practice varies by jurisdiction and homeowners genuinely wonder. If you carry the licenses and insurance, state that plainly and accurately, and nothing more.
Skip the efficiency adjectives. Every competitor is running them, they say nothing, and they attract comparison shoppers rather than buyers.
The landing page has to repeat the promise the ad made, in the same words. A mismatch between ad and page is the most common reason a well-targeted click goes nowhere.
Feed the signed contract back into the account, or you will optimize toward screen repair
Form fills and calls are inputs, not outcomes, and the bidding will chase whatever you tell it counts.
Import outcomes from your CRM back into the ad account: booked measure, held measure, signed order. Value them differently, ideally weighted by opening count or contract tier, so the system stops treating a one-window inquiry and a whole-house order as identical events.
Report on cost per held measure and cost per signed job. Cost per lead is the number that looks best in a storm week and tells you least.
Keep storm weeks flagged in every report. An average that blends them describes a month that never happened.
Local Services Ads, where available for your categories, compete for the same phone as your search ads and show a rating rather than a message. Run them as their own line item with their own cost per booked job, and compare deliberately instead of assuming one cannibalizes the other.
Judge the account monthly. Daily readings in a market this seasonal produce panic decisions and not much else.
Questions we actually get
- What should we budget to start with?
- There is no honest published figure for this trade in this metro, and any agency quoting one is guessing. The workable approach is to start with enough daily budget to gather real search term data in your actual service counties, keep a hard cap on storm campaigns, and revisit after you have signed jobs to attribute against rather than form fills.
- Do we need separate campaigns for doors?
- Usually, yes. Entry and patio door inquiries move on a different clock than a whole-house window quote, they are a smaller order, and mixing them means one bid strategy is wrong for both. Separate campaigns also let you turn doors up in a season when window installs are already booked out.
- How do we handle the surge after a hailstorm?
- Cap the storm campaigns before the season rather than during it, keep them on their own phone number, and stay out of anything resembling claim advice or claim handling. Homeowners should be talking to their own insurer. Your job in that week is to route the small work efficiently and keep your retail campaigns from being reshaped by it.
- Is paid social worth running for window replacement?
- It reaches people who have not started searching, which is most of your future buyers, so it can be. Set expectations differently: the read is slower, the inquiries are earlier in the decision, and the follow-up has to be longer. Judge it on booked measures over a quarter, not on cost per click over a week.
- Should we bid on competitor names?
- Sometimes, and carefully. It is legal in general terms to bid on a competitor's brand while not using their name in the ad copy, though trademark policy and your own risk tolerance are worth confirming with your own counsel. Expect a lower conversion rate and treat it as a small, capped experiment rather than a core campaign.
