Fifteen geothermal markets, measured.
The problem is not your budget.

If you install geothermal, someone has pitched you Google Ads. Probably several someones. It is very unlikely that any of them checked how many people in your city are actually searching.

We checked fifteen markets. In all fifteen, a year-round campaign cannot buy enough enquiries to tell you whether the advertising worked — and the reason is the opposite of what almost everyone expects.

The fifteen markets

These are metro market areas rather than city limits — the area a contractor actually serves. The enquiry column assumes you win about 5% of the available clicks and that your page converts 5–10% of visitors.

On an annual average, every one of them fails. The best market reaches three-quarters of the bar at its most optimistic, and about a third of it at its most likely.

There is no cost-per-click column, deliberately. We pulled the price data and then decided we could not stand behind it. In six of the fifteen markets the reported average cost per click is higher than that market’s own top-of-page bid, which cannot be true, and in the thinnest markets only one of the seven phrases returns a price at all. The argument does not need click prices, so we have left them out rather than publish figures we would not defend.

Does a bigger city change the answer?

That is the obvious objection, and it is a fair one: sure, but I am not in Des Moines.

So we added the three largest geothermal metros we could. Philadelphia, Boston and Chicago all fail too — 1,200, 1,080 and 930 searches a month.

Look at what that means. Philadelphia has roughly ten times the population of Madison, Wisconsin. It has about six times the geothermal search volume. Demand in this trade does not scale with city size — it tracks housing stock, lot sizes, heating costs and local incentives. A bigger city gets you a bigger number, not a different answer.

This is not a price problem

The instinct when a campaign underperforms is to look at what clicks cost. Ignore it.

A ninety-day test is only worth running if it can produce enough enquiries to tell you something. Below roughly thirty, you cannot separate a real result from chance — you will have spent a quarter and a few thousand dollars to learn nothing, and whoever ran it will tell you it needs more time.

Indianapolis has about 630 people a month searching. There is no budget at which that becomes thirty enquiries in a quarter. You cannot buy demand that is not there.

Here is the cleanest test of that. Suppose your agency were twice as good at winning clicks as we assumed — capturing 10% of the available searches instead of 5%.

Doubling the click share on its own — the part a better-run campaign actually improves — changes nothing at all. To rescue even two of the fifteen — Philadelphia and Boston, and only with everything going right — you have to double the click share and double the conversion rate at the same time.

Price is not the variable. That is why a bigger budget is always the answer you get: a price problem sounds solvable, and a demand problem does not.

Not even the best month of the year

The obvious next question is whether timing rescues it. Demand in this trade is seasonal and the swing is large — Philadelphia moves between 970 and 1,920 searches a month across the year, Des Moines between 240 and 470, Madison between 130 and 240. Perhaps a campaign concentrated into the busy season could work where a year-round one cannot.

It cannot. We checked every market's best month, and not one of them reaches thirty enquiries.

Philadelphia at its January peak, with everything going right, reaches 28.8. That is the single best month in the single largest market we could find, and it still falls short.

One thing worth knowing if you do advertise seasonally: the peak is not where most people assume. Nine of the fifteen markets peak in January, which fits the heating-bill theory. But four peak in July — Chicago, Denver, Des Moines and Buffalo — which is a summer installation season, not a heating panic. Toronto peaks in May. If you time a campaign around a story about January heating bills, in those markets you will have picked the wrong month.

So the question is not “what should my budget be”, and it is not “when should I run it”. In these fifteen markets, there is no budget and no month that produces a decision.

Nobody selling you a twelve-month contract is going to volunteer that.

If you are in Canada

All of Canada produces about 1,910 buy-intent geothermal searches a month. That is one-sixteenth of the United States, for a country with roughly one-eighth of the population.

Toronto, the largest city, shows 300 a month — and that figure is city-level rather than metro, so the true catchment is somewhat larger. It does not change the conclusion.

What we would do instead

If you are a geothermal contractor in a market like these, your money does better in local search: your Google Business Profile, your review flow, and the service pages that appear when someone finally does search. That demand is small, but it is not auctioned, and you do not pay for each visit.

We would rather tell a contractor that than take their money for a ninety-day test we already know cannot conclude.

Check your own market in a minute

Fifteen markets is fifteen markets. Yours may differ — a provincial incentive, an unusually old housing stock, a local utility programme that has moved demand.

The check costs about nine cents and takes a minute, and any competent agency can run it. Ask for two numbers before you agree to any advertising:

1. Monthly search volume for install-intent phrases in your metro — not your state, not nationally.
2. How many enquiries that produces in ninety days, at their assumed click share and conversion rate.

If the answer to the second is under thirty, ask what the test is meant to prove. If nobody will give you those two numbers, that is your answer.

What we are not claiming

We would rather draw the lines ourselves than have you find them.

The click-share and conversion figures are assumptions, not measurements. That you would win about 5% of available clicks, and that your page converts 5–10% of visitors, are ordinary industry rules of thumb rather than facts about your business. The table above shows how much they would have to move to change the answer.

Where it does break down: double the click share and the conversion rate together and four markets become testable. If your landing page genuinely converts one visitor in five, and you genuinely capture a tenth of your metro’s geothermal searches, the arithmetic here does not describe you. Very few contractors are in that position, and you should ask for evidence before believing you are.

We are not claiming geothermal advertising never works anywhere, that it cannot work in a market we did not measure, or that a seasonal campaign in a large metro is a bad idea. Our claim is narrower: across these fifteen markets, a year-round ninety-day test cannot produce enough enquiries to justify its own conclusion — and in none of them can the best month of the year either.

Seven phrases is not every phrase. A contractor bidding on a wider or more local set would see somewhat more volume. It would have to be several times more to change the outcome.

How we measured this

Seven buy-intent phrases: geothermal heat pump installer, geothermal installation, geothermal heating and cooling, geothermal heat pump installation cost, ground source heat pump installation, geothermal contractor, and geothermal heat pump repair.

Pulled from the Google Ads search volume data on 5 August 2026, at fourteen US metro market areas, Toronto, and both national baselines — all in the same run, so every comparison is like-for-like. Seasonal figures come from the twelve months of history returned with each phrase. An eighth phrase, “geothermal heat pump installation”, was dropped on 16 August: Google returns it as the same close-variant group as “installer”, with identical figures in every location, so counting both double-counted one bucket. Total cost of the data: US$1.53.

This is market measurement, not client results. We have no client results in this trade to report, and we say so on the evidence page.

Published by Groundline Marketing, 5 August 2026. Corrected 16 August 2026.