Google Ads for small business: when it works and what it costs

Across 13,474 US search campaigns measured between April 2025 and March 2026, the average advertiser paid $66.69 for a lead. On its own that number decides nothing. An owner paying $267 for a customer worth $400 in gross margin and an owner paying $267 for a customer worth $4,000 are running the same campaign at the same cost, and only one of them is buying something worth having.

So the useful version of the question is a sum you can run in two minutes from numbers you already have, plus a short list of things that rule you out whatever the sum says. Both are below. So is the budget floor Google's own help pages imply, the point where Local Services Ads beat Search, and the awkward math on paying someone else to run it. Setup belongs in a course, not an article, and four free lessons wait at the end of this one.

Start with the click, then leave it behind. WordStream by LocaliQ publishes the largest open benchmark set in the category, drawn from 13,474 US search campaigns run between April 2025 and March 2026. The all-industry cost per click averaged $5.42, click-through rate 6.64%, conversion rate 8.18%, and cost per lead $66.69. Much of page one still quotes a click at one to four dollars. Budget on a two-dollar click when the auction charges six and the campaign runs dry before anyone can judge it.

One honesty note that most pages leave out: the dataset blends Google Ads with Microsoft Ads, and Microsoft carries cheaper clicks, so a Google-only click sits above $5.42. The average is also the least useful line in the report. Cost per lead runs from $26.84 to $131.63 across industries, and that spread settles the decision for most owners.

Read the real estate row against the dentistry row. Real estate buys clicks at $3.22, less than half what a dental practice pays, and still lands a $102.51 lead, because only 3.70% of those clicks convert. Dentistry pays $8.00 a click, converts at 10.67%, and gets a lead for $72.97. An account tuned toward cheap clicks looks healthy on the dashboard while it buys the wrong audience.

The direction of travel changed this year. The 2025 edition of the same report found cost per click up 12.88% and rising in 87% of industries. The 2026 edition found cost per lead down for the first time in five years, with conversion rates up in 87% of industries. Costs rose, and then conversion caught up.

Treat the table as a starting estimate you replace. Your own account outranks it the moment it has thirty conversions in it.

Four lines of arithmetic settle this, and you can do them on the back of an invoice.

Cost per click

What you pay when someone clicks your ad, whatever happens after they land.

Conversion rate

The share of clicks that turn into a call, a form or a booking.

Cost per lead

Ad spend divided by the enquiries it produced, or cost per click divided by conversion rate.

Close rate

The share of leads that become paying customers.

Gross margin

Revenue from a job minus what it costs you to deliver, before overheads.

1. 1

   Work out your cost per lead

   Divide your industry cost per click by the conversion rate of the page the click lands on. At $8.33 a click and an 8.05% conversion rate, that is about $103 a lead, against a published home improvement average of $90.92. Start with the published figure and swap in your own.
2. 2

   Work out your cost per customer

   Divide the cost per lead by the share of leads you close. Ninety dollars a lead at one in three is $270 a customer, and at one in six it is $540.
3. 3

   Work out your margin per customer

   Take the revenue from an average job and subtract what it costs you to deliver. Gross margin, not revenue. A $1,200 job that costs $700 in labour and materials leaves you $500.
4. 4

   Compare the two, and leave yourself room

   A cost per customer under half your margin leaves room for the slow month and the lead who stops answering. A cost per customer close to your margin breaks even at best, and any tightening of the auction pushes it under.

If your customers come back, run the comparison against two or three years of margin rather than the first job. A pest control route or a dental patient can cost more to win than the first visit returns and still pay for itself twice over.

Google sets no minimum, and nobody should set one for you without knowing your trade. Its own help pages imply a floor. The guidance on Target CPA bidding says to judge a campaign over the last 30 days, and to have at least 30 conversions inside that window.

Multiply that threshold by your cost per lead and the floor appears. At the all-industry average of $66.69 it comes to about $2,000 a month. In restaurants, at $30.57 a lead, closer to $920. For a law firm at $131.63, nearly $3,950. That arithmetic is ours, built from two published figures rather than lifted from a report. Below that spend you can still advertise, and you cannot yet tell whether it worked.

The same threshold explains why a small budget has to be narrow. Split $1,500 a month across four services and three cities and you own twelve segments, none of which reaches thirty conversions in a quarter. Nothing can be read, so nothing improves. One service and one town until the numbers are legible.

Trim the figure before you commit it. Lunio's study of 2.7 billion clicks found 8.51% of paid traffic invalid, about one click in twelve, which costs a $2,000 budget around $170 a month. Lunio sells protection against the problem it measured, so take that as a haircut rather than a law.

Most pages on this subject are written by someone who sells advertising, which is why the disqualifiers go missing. Six conditions make paid search a bad buy, and five of them cost nothing to check.

• Too few people search for what you sell. Google Keyword Planner is free and answers this in ten minutes. Google marks low-traffic keywords as low search volume and leaves them inactive until that traffic rises.
• You cannot answer the phone reliably. On Search, an unanswered call is a click you paid for and set on fire. On Local Services Ads it is worse, because Google counts responsiveness in ad ranking and says missed calls may hurt it.
• Your margin per customer sits below your likely cost per customer. That is the sum above, and it rules out more businesses than the other five conditions combined.
• You have no conversion tracking. Without it the bidding optimizes toward whatever it can see, usually form fills, and you pay to teach the system to find people who never buy.
• The demand you would buy is demand you already own. A large field experiment run with eBay found brand-keyword ads produced no measurable short-term benefit, because that traffic arrived through the organic result anyway.
• You cannot fund the evaluation period. Thirty conversions at your cost per lead is the floor, and a budget that empties in week three buys you a bill and no answer.

The eBay finding needs its caveats said out loud. Blake, Nosko and Tadelis found returns on non-brand keywords positive for new and infrequent buyers, and negative on average, because regular buyers soaked up most of the spend. eBay also owns its organic results. A new contractor with none is the opposite case, which makes this a finding about brand terms and mature demand rather than a verdict on paid search.

Google frames its own two products as a choice between paying for clicks and paying for leads, and for many service businesses the second is the easier start. Local Services Ads charge when a customer contacts you, not when they click, and the budget runs as a weekly average under a monthly cap.

The pitch you will hear is that Local Services Ads charge only for good leads. That was closer to true two years ago. Google now assesses leads automatically at first contact and does not charge for the ones it judges invalid. It may credit a charged lead later, usually inside 30 days, with the original charge still sitting on the invoice. Manual disputes are gone, there are no credits for a job type or an area you do not cover, and none at all in health care, tax or Europe.

The phone rule is stricter here than on Search. Google lists responsiveness among its ranking factors and says missed calls may affect it, so one missed call costs you the lead and suppresses the next one. Google also charges for a missed call you return later.

Check coverage before you plan around it. The eligible list runs to roughly 150 categories, from acupuncturist through yoga studio, and several are limited to a couple of states. Architects and barber shops can run them in California and Florida only. Our piece on local paid ads that convert covers the lead mechanics in more detail.

They are different instruments, and the comparison people reach for is usually the wrong one. Ads buy volume today at a permanent price per lead, so stop paying and the leads stop that afternoon. Organic search is capital spending with a lag, expensive at the start and cheaper per lead every quarter it keeps working. Run ads if you need customers this quarter and the break-even sum clears. Invest in search if you can wait two or three quarters for an asset you keep.

One caveat about a statistic you may have seen. Seer Interactive tracked eighteen months of falling paid click-through on queries carrying an AI Overview, and the drop travelled a long way. Their April 2026 update, drawn from 53 brands and 5.47 million tracked queries, found the trend reversed. By February 2026 those queries returned 16.21%, against 21.85% on queries with no AI Overview. The tax on paid search is real and much smaller than the panic around it.

To see how your own site reads to both engines before you choose, our free SEO report scores organic and AI visibility in one pass and costs nothing.

The arithmetic on management is uncomfortable to publish from an agency, so we will publish it. Agencies commonly quote 10 to 20 percent of ad spend, or a flat monthly retainer for a small account. At 15 percent, a $2,000 budget justifies $300 a month of management, and almost nobody takes that work at that price. Below roughly $2,500 a month in spend, a flat fee can rival the budget it manages.

That is why the setup sits in a free four-part course rather than behind a sales call. Lesson one sets up the account, including the campaign Google builds for you during signup and how to skip it. Lesson two installs conversion tracking before you spend a dollar. Lesson three builds the first Search campaign. Lesson four covers the first 90 days, where small accounts are won or lost.

Two things earn their place before any management fee. Call tracking, from CallRail, WhatConverts or CallTrackingMetrics, ties a phone call back to the keyword that produced it, and all three bill call minutes on top of the base price. The second costs nothing: import your closed deals back into Google Ads so the bidding optimizes toward jobs won rather than forms filled.

Hiring earns its keep higher up. Above roughly $3,000 a month in spend, or across several services and markets, the percentage buys real work: search term review, landing page testing, and bid changes with enough conversions behind them to mean something. That is the shape of our own paid ads work, on two conditions we hold to. You own the ad account, and the spending cap stays yours.

Whatever you decide, decide it with the sum in front of you. The owner paying $267 for a $400 customer and the owner paying $267 for a $4,000 customer see the same cost per lead in the same dashboard, and only one of them should leave the campaign running.