Almost every contractor asks me the same question, and it is the wrong one.
Which channel should we be on. Local Services Ads or Google Ads. Should we try Angi. Somebody mentioned we should be advertising inside ChatGPT now.
It is not a menu. It is a sequence, and the order is not a matter of preference. Each step caps the one above it, which means skipping a step does not save you money, it just moves the cost somewhere less visible.
Here is the arithmetic that makes the point. Traditional Google search ads in this category are reported to average somewhere around $91 per lead. If half your leads never turn into a booked job, you did not pay $91. You paid $182 per job, and you paid it for a step you could have fixed for nothing.
So the rule that generates the whole order: fix the cheapest thing that limits everything above it, first.
Step one, and it is not marketing
Answer the phone. Book the call.
I went through the numbers on this separately so I will not repeat them here. The short version is that the gap between a mediocre booking rate and a good one is worth more than any campaign you are considering, and it costs a training habit rather than a budget.
Move on when: you know your booking rate, by person, from your own records, and it is above 70%. Not estimated. Counted.
If you skip this step, every number below it gets multiplied by your leak.
Step two: the free thing that outperforms your website
Google Business Profile, and reviews.
For most home services businesses this single free asset produces more leads than the website, social media and paid advertising combined. It is also where nearly half of all Google searches now land, given that something like 46% carry local intent, and a homeowner typing “AC repair near me” at nine at night is about as far down the funnel as a person gets.
The work is unglamorous: correct categories and service list, photographs of real jobs added regularly, hours that are actually right including holidays, questions answered, and every review replied to including the bad ones.
Reviews are the compounding half. Earning them properly is a process at the point of work, not a campaign, and it is the thing that decides whether you appear in the map pack at all.
Move on when: the profile is complete, you are adding photos most weeks, and you have a repeatable way of asking that does not depend on somebody remembering.
Step three: Local Services Ads
Now you can buy leads, and this is the first place to buy them.
They sit above every other paid result on the search page, they are priced per lead rather than per click, and the buyer arriving through them has high intent. I covered what changed and what it costs in detail, including the parts that are no longer true in the way most guides still describe.
The reason this comes before search ads is simply position and pricing model. You are paying for a lead in a placement above the ads you would otherwise be bidding for.
Move on when: LSA is producing leads at a cost per booked job you can defend, and you have hit the volume ceiling for your service area. Not before. A lot of contractors add channels while their best one still has room.
Step four: search ads
Now the auction, and only for the queries where the intent is unambiguous.
Emergency terms, your specific services plus your area, and a tight negative list from day one. Not broad campaigns, not Performance Max as a first move, and not display.
This step exists to capture demand LSA missed and to own your own brand term, which somebody else is probably bidding on.
Move on when: you are not sure you should. For a large number of trade businesses, the sequence correctly stops here.
Step five: marketplaces, as a bridge only
Shared-lead platforms are a way to fill a calendar while the steps above are still maturing. They are not a foundation, because the model prices your lead by selling it to several people at once, and I went through when they genuinely make sense and when they become a trap.
Use them deliberately, with an exit date, and know your cost per booked job from that source specifically rather than blended.
Step six: the new one, and where it actually belongs
Something real did change this year, so it is worth saying plainly rather than either ignoring it or overselling it.
On 5 May 2026 OpenAI opened a self-serve ads manager to US businesses of all sizes, with CPC bidding, a conversions API and no minimum spend. The previous arrangement required six figures, so this is genuinely open to a plumbing company for the first time.
And the underlying behavior is real. Homeowners do now ask an assistant what is wrong with their water heater or what a repair should cost before they call anybody, and the sources those answers draw on are heavily Google Business Profile, Local Services Ads and ordinary search results.
Read that last sentence again, because it is the strategic point. The work that makes you visible in AI answers for this category is mostly steps two and three, done properly. The ad product is separate and unproven here.
So: worth a small test if the earlier steps are working and you have budget you can afford to learn with. A terrible first channel. The one thing I would not do is treat a new ad platform as a shortcut past the free asset that the assistants are reading anyway.
The order, in one place
Answer the phone. Fix the profile and the reviews. Buy leads where they are cheapest and highest intent. Then bid on the queries you are certain about. Marketplaces only as a bridge. New platforms last, and only with money you can afford to spend learning.
The through line is speed to lead. Every paid channel on this list gets more expensive the slower you respond, and none of them compensate for a phone nobody picks up. That is why the free step is first, and it is the step almost everybody skips because it does not feel like marketing.
Before lunch
Three things, about an hour.
Work out which step you are actually on, honestly. If you cannot state your booking rate from records, you are on step one regardless of what you are currently spending.
Then open your Google Business Profile and fix whatever is stale. Hours, services, photos older than a month, unanswered questions.
Then look at your current paid channels and calculate cost per booked job for each, separately. Not cost per lead, and not blended across sources. That number will usually tell you which channel to stop and which has room left.
My notes on lead generation cover the detail behind each step. If you’d like to think through where your own business is in this order, drop me a line on email, WhatsApp or LinkedIn and we can have a quick chat. I’m contracted full time so this isn’t a pitch.