UrbanTechFlow · Gatineau–Ottawa

Google Ads for Ottawa Small Business: Where the Money Actually Leaks

Picture a six-person heating and cooling shop working out of a unit off Merivale Road in Nepean. Call it Rideau Ridge Heating & Cooling, an illustrative composite, not a real company. The owner turned on Google Ads in April because a competitor's ad kept showing up above his own listing. He picked a daily budget, typed in about fifteen keywords, accepted the campaign settings Google suggested, and got back to work. Four months later he has spent a real amount of money, the phone has rung a bit more, and he cannot tell you whether a single furnace install came from it.

Then somebody opens the search terms report for the first time. His ads have been showing for hvac technician salary ontario, furnace filter sizes, free ac repair, hvac courses algonquin, and, the expensive one, hvac companies hiring ottawa. Job seekers. Students. DIYers. People in Kingston. A meaningful slice of his budget went to clicks that could never, under any circumstance, become a booked job.

This is the ordinary case, not the horror story. Nothing was broken. The account did exactly what it was configured to do.

The four places the money actually goes

Leak one: match type. Broad match is the default path in most simplified campaign builds, and Google's auto-applied recommendations quietly widen keywords over time. Broad match means your keyword is a suggestion about topic, not a contract about wording. In a large market that's a feature. In ours it's a tax.

Leak two: no negative keyword list. Every account needs a growing list of words that stop an ad cold, salary, jobs, free, DIY, course, rental, wholesale, plus every city you don't serve. Most small accounts have none.

Leak three: a budget spread so thin it cannot learn. Trades clicks in the National Capital Region are not cheap. If your daily budget buys three or four clicks, you collect roughly a hundred clicks a month across every campaign, keyword and device. That is not enough data for you to see a pattern, and it is not enough for Google's automated bidding either, it will spend your money exploring while you wait for a signal that never arrives.

Leak four: no tracking worth the name. If the account counts "clicked the phone number" as a conversion, you are optimizing for taps, not for installs. Taps are free to give away. Installs are not.

You pay for the click whether or not the phone rings. Everything that happens after the click is where the money is either made or lost, and almost none of it happens inside Google Ads.

What the system actually looks like, in order

1. Decide what a booked job is worth before the first click

Not revenue, margin, minus the share of leads that never close. That single number sets your ceiling on cost per lead, which sets your bids, which decides which keywords you are allowed to compete for at all. Skip this step and every later decision is a guess. Result: a hard number that ends most arguments about whether a keyword is "too expensive."

2. Map the words people here actually type

Not the words you'd use. The words a homeowner in Barrhaven uses at 6 a.m. in February. The map has to separate three intents that look identical in a keyword list: someone ready to buy, someone researching, and someone looking for a job or a part. It also has to separate emergency language from planned-project language, because they convert at completely different rates and deserve different budgets and different hours.

This is also where the size of the market becomes real. Our own team measured total monthly search volume for every honest term describing our own category in this region: about 4,900 searches a month across six terms, total. Small markets are the normal condition here, not a failure. Result: a keyword set built around a handful of terms that can actually produce revenue, instead of forty that dilute it.

3. Treat match type as a spending decision

The sequence matters more than the choice: earn the data first with tight matching, then widen deliberately once the account knows what a good lead looks like. Result: your budget concentrates on demand you can serve.

4. Negatives are a weekly habit, not a launch task

The search terms report is the only place where the account tells you the truth. Someone has to read it on a schedule, decide which terms are genuinely off-target, and push them into shared negative lists that protect every campaign at once. It is unglamorous and it never ends, because the market's vocabulary keeps shifting. Result: month over month, a rising share of spend lands on searches that could become work.

5. Draw the geography like a dispatcher, not a marketer

Where will your crew actually drive in January? Location settings also have a nasty default, targeting people merely interested in your area, which is how a Nepean furnace ad ends up in front of somebody in Sudbury. And in this region there is a river. If your ads reach Gatineau, Hull or Aylmer, your landing page falls under Quebec's Charter of the French Language: section 52 requires that commercial documents available to the public, which the OQLF's own guidance extends to websites, be available in French, with no size threshold and no "small business" exemption, and the French version may not be offered on less favourable terms than the English one. Each day of non-compliance counts as a separate offence. Result: your money stays inside your service area, and crossing the river is a decision instead of an accident.

6. The ad's job ends at the click

The page that receives the click has to match the promise, load fast, and make contacting you nearly effortless. The strongest public evidence here is Google's own randomized Vodafone test (web.dev, 2021): a 31% improvement in Largest Contentful Paint produced an 8% increase in total sales and a 15% uplift in the lead-to-visit rate, in a genuine A/B split. Be clear about what that is, it's a conversion argument, not a ranking argument. Nothing in the 2026 expert survey of 187 local ranking factors puts page speed near the top of the local pack. Speed makes the traffic you already bought worth more.

Then there is the part no software fixes: someone has to answer. The study June 2026 survey of 1,227 US consumers found 75% decide which business to use in under 30 minutes and 28% decide in under five, with 72% considering three or fewer businesses. It's a US panel and self-reported, so treat it as direction rather than an Outaouais measurement, but the direction is unambiguous. Result: the clicks you paid for convert instead of bouncing to whoever picked up first.

7. Count booked jobs, not form fills

Proper measurement means call tracking that distinguishes a two-minute booking from a nine-second wrong number, form submissions tied to a specific ad and search term, and, the step nearly everyone skips, feeding the outcome back in once a lead becomes a signed job. That last step is what lets automated bidding chase revenue instead of chasing whichever keyword generates the most cheap noise. Result: you can name the keywords that pay for themselves, and kill the ones that don't, with evidence.

8. Concentrate, then be patient on purpose

In a market this size, five thin campaigns learn nothing while one well-fed campaign learns quickly. And with volumes this low, a bad week is usually noise. Deciding in advance how much data you need before you act is what separates management from fiddling. Result: fewer changes, better ones.

Before and after, illustrative

Numbers below are illustrative, built to show the shape of the change, not measured from a real account.

Fewer clicks. Better clicks. And for the first time, the owner can point at a line and say which words paid the crew.

What it looks like when it's working

• You can name your top three revenue-producing search terms without opening a dashboard.
• The search terms report is boring, the junk was negatived out weeks ago.
• Calls arrive during hours you actually staff, from places you actually drive.
• Spend shifts by season on purpose: emergency terms in the cold snap, planned-project terms in spring.
• When you raise prices, and the 2026 US home-service survey of 1,050 owners found 65% did, lead volume dips and lead quality rises, and you can see both.
• Ads are treated as their own channel, not as a substitute for your Google Business Profile. They don't improve each other.

Why this is a full-time job

None of the above is beyond you. You run a business in a region with more than 42,000 employer businesses, over half of which have four employees or fewer (Statistics Canada, Canadian Business Counts, released February 2026). You already manage crews, quotes, permits, warranty calls and a payroll, and you're doing it in a market where the National Capital Region shed roughly 9,400 federal jobs in two years (Treasury Board of Canada open data, March 2026), which tightens every household budget your ads are aimed at and makes every wasted click cost more than it did in 2023.

The reason this leaks money for good operators is not intelligence. It's that a paid search account rewards steady weekly attention, reading reports, harvesting negatives, checking that tracking still fires after a website update, resisting the urge to change everything after one slow Tuesday. That's someone's whole trade, the same way HVAC is yours. Our team does this full time, across Ottawa, Nepean, Barrhaven, Kanata, Orléans and the Outaouais, in both languages, because the Gatineau side genuinely requires it.

If you'd rather understand the shape of it and run it yourself, take this article and go, it's honest work either way. If you'd rather it just be handled, UrbanTechFlow can be reached at 819-550-9153.