Minimum viable ad spend for a local business

The number is different for a house cleaner and an injury lawyer, because the click prices are different by an order of magnitude. Anyone who quotes you a minimum without asking what you sell is guessing.

What does not change is the shape of the problem. Too little spend produces a handful of clicks spread thin across a month, no leads, and no way to know whether the campaign was wrong or the sample was small. You spend real money and learn nothing, which is worse than spending nothing.

How do I calculate my own minimum?

Three numbers, and you already have two of them.

Start with what a customer is worth to you in gross margin, not revenue. Then your close rate on quotes, which most owners know within a few points even if they have never written it down. Then the cost per click for your service in your area, which Google's own Keyword Planner will estimate in ten minutes.

For a placeholder while you pull your real numbers, LocaliQ, which publishes annual search advertising benchmarks under the WordStream name, puts the 2026 average cost per click at $8.33 for Home and Home Improvement, $8.00 for Dentists, $4.62 for Beauty and Personal Care and $2.05 for Restaurants and Food. Its average conversion rates for those categories are 8.05%, 10.67%, 10.35% and 8.05%. Those are medians rather than means, LocaliQ sells advertising management, and the 2026 edition does not disclose its sample size. The 2025 edition of the series disclosed 16,446 US campaigns with a minimum of 64 campaigns per category. Vendor research, and a starting estimate only.

Here is the chain, in six lines, with home services numbers.

One. Cost per click, $8.33.

Two. Conversion rate on the landing page, 8.05%. That means one lead every 12.4 clicks (derived).

Three. Cost per lead, $103 derived.

Four. Close rate on quotes. Say a third. That is one customer every three leads, so cost per customer is $310 derived.

Five. Gross margin on an average job. If the job bills $900 at a 40% margin, it contributes $360.

Six. The gap. $360 minus $310 is $50 per customer. That clears, barely, and it tells you the campaign is real but has no room for error.

Now the part that sets the floor. One customer is not a signal, and neither is three. You need enough customers per month that a bad month and a broken campaign look different from each other. Ten leads a month is the practical threshold where a monthly number starts carrying information for most local businesses. Ten leads at $103 is $1,030 a month, which is your floor in this example. Under it you are not running a test.

Rerun those six lines with your own click cost, your own conversion rate and your own close rate. If the number that comes out is larger than you can fund for ninety days, that is the answer, and it is a useful answer.

Why is a small budget worse than no budget?

Because it produces confident conclusions from bad data, and those conclusions stick.

A business runs a thin budget for a month, gets no calls, and decides paid search does not work for their industry. Nothing about that month tested the industry. It tested a sample too small to read.

This is where a piece of received wisdom deserves killing, because it is quoted constantly and it is not research. You will be told that split testing needs 10,000 monthly visitors. The figure has been traced. It appears as reference 29 in Kohavi, Deng, Longbotham and Xu's "Seven Rules of Thumb for Web Site Experimenters," a peer-reviewed KDD 2014 paper generalising from thousands of controlled experiments at Amazon, Booking.com, LinkedIn and Microsoft. In the paper it is attributed to a Neil Patel post on the QuickSprout blog dated 14 January 2013, and the four authors immediately qualify it: the guidance, they write, should be refined to the metrics of interest.

A blog post from 2013, promoted to an industry threshold by repetition. The paper's actual position is that minimum sample size is a function of the metric's variance and of the size of the effect you want to detect. Their worked table for Bing makes the point better than any rule of thumb: revenue per user, a wildly skewed metric, needs 114,000 users per variant to detect a 4.4% change, while sessions per user needs 4,700 for 5.4%. Same product, same traffic, twenty-four times the sample requirement depending on which number you are watching.

The transferable lesson for a local advertiser is the inverse of the usual one. You do not need a fixed traffic number. You need enough events of the specific thing you are measuring, and leads are rarer and more variable than clicks. A budget that buys plenty of clicks and four leads has produced a readable click number and an unreadable lead number, and the lead number is the one you are spending money to learn.

So the practical move is to commit to fewer things at a real budget rather than many things at a thin one. One service, one geography, enough money to matter. Then read the result.

The platform pushes the same way. Google's documentation on Smart Bidding notes that some strategies rely on a minimum volume of historical conversion data. An account with three conversions a month is not giving the bidding system anything to learn from either.

Where does a thin budget leak first?

Into searches you would never pay for if you saw them written down.

When budget is scarce, every dollar spent on the wrong search costs you a lead you could have had. The single highest return activity in a small account is opening the report that shows what people actually typed, because the search terms report is where the waste is visible and it is free to read.

The second leak is the ad itself. Copy that describes what you do without saying who it is for invites everybody, and a small budget cannot afford everybody. Ad copy that filters out the wrong customers is a budget decision before it is a creative one. Naming a price range, a minimum job size, or a specific neighborhood costs you clicks you did not want.

The third leak is the destination, and it is the biggest. Spend at a page that does not convert is not an investment with a slow payback. It is a loss with a receipt. Before you raise a budget by a dollar, understand why ads pointed at a weak landing page lose money faster than any other way of spending it. Go back to line two of the chain above and halve the conversion rate. Your cost per customer doubles and your $50 of margin becomes a $260 loss per job (derived).

What can I do with a small budget that actually works?

Narrow it until the money is dense enough to matter.

Cut geography to the area you actually want to serve, not the area you would technically drive to. Cut hours to the hours you answer the phone. Cut services to the one with the best margin, not the one with the most searches. A small budget concentrated on one profitable job in one city can produce a readable result. The same budget spread across five services and a county produces noise.

Then make every lead you do get count harder. Two cheap fixes do more for a thin budget than any bid change. Cut the fields on your quote form, because every extra field costs you submissions and at low volume you cannot afford the drop. And use the page people land on after they submit, since the thank you page is usually wasted space that could be scheduling the call or setting expectations.

Where does the floor argument break down?

In two places, and both are worth knowing before you treat the number as a law.

When the job value is large enough that one customer settles it. A remodeler at a $40,000 average job with a 30% margin does not need ten leads to know whether the campaign works. One closed job pays for a year of a small budget. The floor still applies to learning, since a single sale tells you nothing repeatable, but it does not apply to survival. Fund a long observation window instead of a large monthly number.

When the constraint is capacity rather than demand. If you are already turning work away, the correct minimum budget is zero. More leads at any price is a cost, because every unreturned call is a review risk and a reputation cost that does not show up in the ads account.

There is a third case that looks like a budget problem and is not. If your leads are cheap and your customers are expensive, the account is fine and the intake is broken. That gap is the subject of cost per customer rather than cost per lead, and it is the number that decides whether raising a budget is investment or waste.

What to do this week

Write down three numbers on one piece of paper. Gross margin on an average job. Close rate on quotes. Cost per click for your main service, from Google's Keyword Planner.

Chain them together using the six lines above and find how much you would spend to acquire one customer at your current conversion rate. If that number is larger than your margin, stop. The problem is not budget size. It is either the page or the offer, and adding money makes the loss bigger.

If the number works, multiply your cost per lead by ten. That is your monthly floor. Fund it for ninety days or do not start.

Then write both figures somewhere you will see them again, because the temptation in month two is to compare against a benchmark instead of against the number you set.

Questions about what to budget?

Email me at eric@seod.com with your average job value, your gross margin on it, your close rate on quotes, and the main service you would advertise. I will run the six lines and send back the monthly number I think is your realistic floor, plus whether I would start at all.

If the honest answer is that your numbers do not support ads yet, I will say that and tell you what I would fix first. That answer takes me five minutes and saves you a quarter.

Or keep reading more on paid search for local businesses.