Two numbers that constantly get conflated
Every ads conversation contains two separate costs. Management is what you pay someone to build and run the campaigns. Media spend is what you pay Google for the clicks.
Media should go on a card attached to an ad account registered to your own business. That matters more than it sounds. You see every dollar, nobody can mark up your media without you noticing, and if you change agencies you keep the account and its performance history — which is the thing that makes the campaign work in the first place.
An ad account in an agency’s name is the single most common way local businesses lose years of accumulated data. Insist on ownership before you sign anything.
What clicks actually cost here
In competitive Metro Detroit home-service categories, expect roughly eight to eighteen dollars per click. Emergency and high-ticket categories run at the top of that; less contested trades run lower.
Do the arithmetic before you set a budget. At thirty-three dollars a day you are buying two to four clicks. If one in ten becomes a lead and one in three leads becomes a job, you are looking at a job every week or two — enough to learn from within a month, nowhere near enough to dominate a market.
This is why spending a few hundred dollars a month in a competitive category tends not to work. The campaign never accumulates enough conversions for the bidding to optimise, so it stays permanently in a learning phase and you conclude that Google Ads does not work. It works; it just was not given enough signal to work with.
The floor, and what to do if you are under it
For a competitive local service category, roughly a thousand dollars a month in media is the realistic floor. Below that, put the money somewhere it compounds instead.
Your Google Business Profile, filled in properly and posted to. A deliberate system for asking every completed customer for a review, with a sentence about what you actually did rather than just a star rating. Content answering the questions buyers ask before they call. None of those stop working the day you stop paying, which is the fundamental difference between them and paid search.
There is also a hard prerequisite that has nothing to do with budget: if you cannot answer a new lead within minutes, do not run ads at all. Paid leads decay fast. Somebody comparing three providers calls all three, and the one who answers wins regardless of who is better. Running ads with a slow response time is buying leads for your competitors.
The negative-keyword trap
This one costs real money in small accounts and almost nobody knows it, because negative keywords do not behave the way positive ones do.
A broad-match negative requires every word to be present. Adding “epoxy paint” as a negative does not block the search “garage floor paint,” because there is no “epoxy” in that query. People assume they have blocked a whole category and have blocked almost nothing.
Negatives also do not match plurals or close variants. Blocking “kit” does not block “kits.” And an exact-match negative blocks only that exact phrase, so “paint” as an exact negative does nothing for “cost to paint garage floor.”
The practical fix is to add negatives as phrase match, and to read the search terms report weekly rather than trusting your keyword list. The search terms report shows what people actually typed, which is always broader than what you thought you bought. In one local account we audited, roughly a third of attributable spend was going to do-it-yourself and research queries despite a negative list being attached — because every term on it was the wrong match type.
Where the budget quietly leaks
Beyond negatives, four things drain small local accounts consistently.
The landing page does not match the ad. Somebody searches for one specific service, clicks an ad promising it, and lands on a homepage covering eight services. Most do not hunt for what they came for. This costs twice, because Google scores landing-page relevance into your ad rank — so you also pay more per click.
Conversion tracking does not really work. If the conversion action never fires, or fires on the wrong thing, the bidding is optimising against noise and every report you read is fiction. Test it by submitting your own form and confirming the conversion appears. Do not trust the dashboard.
Assets sit disapproved. Call extensions and images can be rejected for months while the campaign appears to run normally. A disapproved call asset means your ads have never shown a phone number — which, for a trade whose customers phone, removes the primary conversion path entirely. Nothing alerts you loudly; somebody has to open the assets view and look.
And the landing page is invisible to the ad platform. If your site renders only in the browser, Google’s ad-quality crawler receives an empty document and scores you on it. In our audit of 1,348 funded companies, 28 of the sites serving an empty shell were running Google Ads.
A workable first ninety days
Weeks one and two: verify conversion tracking end to end before spending anything meaningful. Point every ad group at a page that matches its keywords. Set the geography tightly — presence in your actual service area, not interest-based targeting that reaches people merely reading about Detroit.
Weeks three to six: read the search terms report weekly and add phrase-match negatives for what you actually bought. Expect this period to buy data as much as leads. Do not judge keywords yet.
Weeks seven to twelve: now judge. Cost per lead and cost per booked job, not cost per click. Pause what has meaningful spend and no conversions, once the landing page has been right long enough to be a fair test. Consider switching from maximising clicks to maximising conversions once you have accumulated roughly fifteen to thirty conversions.
Committing to one month is not a test — a month is the learning period. Three months is a test. If you cannot commit to three, spend the money on reviews and your Business Profile instead and come back when you can.