How much Google Ads budget do you actually need? Calculate the floor

Almost every answer to "how much should I spend on Google Ads" is a percentage. Five to ten per cent of revenue. Ten to twelve for aggressive growth. These rules tell you what you can afford to lose. They tell you nothing about whether Google Ads can work at that number, which is the only question that matters before you start.

Why percentage-of-revenue budgets tell you nothing

A percentage rule answers an affordability question: what can this business absorb. That is a real question, and finance should ask it. But it produces the same recommendation for a personal injury firm paying $9.87 a click and a restaurant group paying $2.05, and those two businesses need budgets that differ by a factor of five to reach the same amount of learning data.

The mechanism nobody prices in is that Google Ads is not a linear medium any more. It is an auction fed by a bidding model that needs a minimum volume of conversion events to work at all. Spend below that volume and you do not get proportionally less performance. You get a campaign stuck in a permanent learning state, with unstable cost per acquisition, because the model never accumulates enough signal to make the estimates it exists to make.

That gives you a floor that is calculable, and it is the number worth arguing about internally, not the percentage.

The number that sets your floor: conversions per month

Every Smart Bidding strategy has a data appetite. Google publishes recommendations, and practitioners running these strategies at scale report a higher number where they actually become stable. Both are worth knowing, because the gap between them is where most underperforming accounts sit.

Read the third column as the honest one. The 30-conversion recommendation for target CPA is a threshold at which the strategy is permitted to work, not the point at which it works well. Lead-generation accounts in particular sit at 30 to 40 conversions a month and wonder why CPA swings 40% between weeks: the model is making estimates from a sample that thin.

The learning period compounds this. In 2026 it typically runs two to six weeks, and campaigns clearing roughly 30 conversions a week exit it fastest, while low-volume campaigns can sit in it indefinitely. A bid strategy change, or a budget change over about 20%, restarts it for one to two weeks. A campaign that is both underfunded and frequently adjusted can spend most of the year learning and never trading.

The floor formula

Work backwards from the conversion requirement to a budget in three steps.

1. Pick your data target. 30 conversions a month if you intend to use target CPA, 50 if target ROAS, 15 to 20 if you will stay on Maximize Conversions. Use the practitioner column instead if stability matters more to you than starting quickly.
2. Convert that into clicks. Clicks needed = conversions needed ÷ your landing page conversion rate. If you do not know your conversion rate, use 2% as a conservative planning figure for a lead-gen site and check it against your own analytics within the first month.
3. Convert clicks into money. Monthly floor = clicks needed × your cost per click. Use Keyword Planner's top-of-page bid estimates for your actual keywords, not an industry average, if you can get them.

Compressed: Monthly floor = (conversions needed ÷ conversion rate) × CPC. A home services company aiming for target CPA needs 30 conversions. At a 4% conversion rate that is 750 clicks. At $6 a click, the floor is $4,500 a month, or $150 a day. If the available budget is $1,500, the honest answer is not "run a smaller campaign", it is "this campaign cannot reach a stable state as scoped, so change the scope".

What the floor looks like at 2026 prices

WordStream's 2026 benchmark study, drawn from more than 13,000 search advertising campaigns across 23 industries between April 2025 and March 2026, puts the all-industry averages at a $5.42 cost per click, a 6.64% click-through rate, an 8.18% conversion rate and a $66.69 cost per lead. Cost per click ranges from $1.63 in arts and entertainment to $9.87 for attorneys and legal services, which also carries the highest cost per lead at $131.63.

Applying the formula to 30 conversions a month, at three conversion rates and three price points:

Two things in that table are worth sitting with. First, the spread runs from about $750 to about $14,800 a month for the same amount of learning data, which is why any single-number answer to this question is wrong. Second, the right-hand column is the optimistic one and probably not yours. Benchmark conversion rates from campaign-level datasets are click-to-lead rates on campaigns that agencies have already optimised; a new advertiser sending traffic to an untuned page should plan against the 2% column and be pleasantly surprised. If you use the benchmark column to set your budget and land at 2%, you will underfund by a factor of four.

The four moves when you cannot afford the floor

Being below the floor is common and it is not a reason to abandon paid search. It is a reason to change one of the four variables in the formula. In rough order of how often they are the right answer:

1. Shrink the campaign until the budget clears the floor. The most under-used option. Cut to the two or three highest-intent keyword themes, one geography and your best-converting hours. The same $1,500 that is spread thin across 40 keywords in a whole state can be a genuinely competitive budget on six keywords in one metro. You are not reducing ambition, you are concentrating the data so the model has something to learn from.
2. Raise the conversion rate instead of the budget. The formula divides by conversion rate, so doubling it halves the floor. Going from 2% to 4% on the landing page cuts a $8,130 floor to $4,065. For most businesses under the floor, landing page work is a cheaper path to viability than more spend, and the gain persists across every channel. This is the same reason a paid traffic landing page pays for itself faster than most ad optimisations.
3. Lower the conversion requirement by choosing a simpler strategy. Maximize Conversions without a target needs materially less data than target CPA and far less than target ROAS. Start there, let volume accumulate, and add the target once you are genuinely clearing 30 to 50 events a month, accepting a fresh learning period when you do.
4. Use a cheaper conversion event to exit learning, then ladder up. If sales are too rare to feed the model, optimise initially toward a more frequent upstream event. Do this with a clear exit plan: optimising permanently toward form fills is exactly how cost per qualified lead drifts upward while cost per lead looks fine. The cheap event is scaffolding for the learning period, not the goal.

What is almost never the right move is running the full-scope campaign at a quarter of its floor and waiting. That spends the money without buying the data.

How to tell you are below the floor

Four signals, all visible inside the account, usually appear together:

• Search impression share lost to budget above roughly 30%. The campaign is being switched off during the day, so the model sees a truncated and non-representative slice of the auction.
• Cost per acquisition swinging more than 30–40% week to week with no change to the account. That is sampling noise from a thin conversion count, not performance change, and reacting to it makes things worse by restarting learning.
• The learning label reappearing repeatedly, or a campaign that never leaves it. Two to six weeks is normal once. Permanently is a data problem.
• Fewer than about 15 conversions in the last 30 days at the campaign level. Below this, the reported CPA is a small-sample artifact and should not be used for decisions.

If you see the first and the fourth together, more budget is the fix and it is worth calculating exactly how much. If you see the second and third without the first, the problem is more likely conversion tracking than budget, which is worth ruling out before you spend more.

Two things that changed the floor in 2026

The floor moved this year for two reasons worth planning around.

The first is the AI Max transition. Google began auto-upgrading campaigns using automatically created assets and campaign-level broad match to AI Max in September 2026, with Dynamic Search Ads following in February 2027. Broader matching finds queries you were not bidding on, which raises spend at a constant budget cap and typically widens the query mix before it narrows again. Budget calculated for a tightly matched campaign will not behave the same way after the upgrade, so recalculate after it lands rather than assuming the old floor holds. We covered the audit steps in the AI Max pre-migration checklist.

The second is Q4 auction pricing. Cost per click rises through November and December in most consumer categories, and since CPC is a multiplier in the formula, the floor rises with it. A budget that clears the floor in September may sit below it on Black Friday week without a single setting changing. If you are planning peak spend now, the Q4 planning guardrails are the companion piece to this one.

When the answer is not to run Search at all

If your floor comes out at $9,000 a month, you have $2,500, and there is no realistic way to narrow scope or lift conversion rate enough to close the gap, paid search is not the right first channel this quarter. That is a legitimate finding, not a failure, and it is better discovered on a spreadsheet than after two quarters of underfunded campaigns.

The alternatives that work at genuinely small budgets are the ones where the auction is cheaper or absent: local SEO and Google Business Profile for service businesses, organic and owned channels where you already have content, and lower-CPC platforms where the click price does not carry legal-vertical economics. None of them are as fast as search advertising. All of them are better than a campaign that never exits learning.

Frequently asked questions

What is the minimum Google Ads budget in 2026?

There is no universal minimum. Your floor is set by how much conversion data your bid strategy needs, calculated as (conversions needed per month divided by your conversion rate) multiplied by your cost per click. For a campaign targeting 30 conversions a month at the 2026 all-industry average cost per click of $5.42, that is roughly $2,000 a month at an 8% conversion rate and about $8,100 at 2%.

How much should a small business spend on Google Ads per month?

Enough to generate at least 15 to 30 conversions a month in the campaign, which depends entirely on your click price and conversion rate. A restaurant paying around $2 a click can reach that on roughly $750 to $3,000 a month. A law firm paying close to $10 a click needs roughly $3,600 to $14,800 for the same amount of data. Calculate your own number rather than using a percentage of revenue.

Is $500 a month enough for Google Ads?

Only if your cost per click is low and your conversion rate is high enough that $500 buys 15 or more conversions. At a $2 click price and a 5% conversion rate, $500 buys 250 clicks and about 12 conversions, which is marginal. At a $6 click price it buys 83 clicks, and no bid strategy can learn from that. At $500 the usual answer is to narrow the campaign to one geography and a handful of high-intent keywords, or to use the budget elsewhere.

How many conversions does Google Ads Smart Bidding need?

Google recommends at least 30 conversions in the previous 30 days before using target CPA and at least 50 for target ROAS. Maximize Conversions has no stated minimum and is workable from roughly 15 to 20 a month. In practice, advertisers report stable performance closer to 50 to 80 conversions a month for target CPA and 100 or more for target ROAS.

How long is the Google Ads learning period in 2026?

Typically two to six weeks, depending on conversion volume, budget and conversion delay. Campaigns generating around 30 or more conversions a week exit fastest; low-volume campaigns can stay in it far longer. Changing the bid strategy or moving the budget by more than about 20% restarts learning for roughly one to two weeks, which is why frequent budget adjustments hurt underfunded campaigns most.

What happens if my Google Ads budget is too low?

The campaign does not simply deliver proportionally less. It shows high impression share lost to budget, cost per acquisition that swings 30 to 40% week to week from sampling noise, and a learning status that keeps reappearing. Reported CPA at fewer than about 15 conversions a month is a small-sample artifact and should not drive decisions.

Should I lower my target CPA to make a small budget work?

No. Lowering the target restricts the auctions the campaign enters, which reduces conversion volume further and makes the data problem worse. If the budget is below the floor, narrow the campaign's keywords and geography so the same money concentrates, improve the landing page conversion rate, or move to a bid strategy with a lower data requirement.

The takeaway

Stop starting with a percentage. Take the conversion volume your intended bid strategy needs, divide it by your real landing page conversion rate, multiply by your real cost per click, and you have a number you can defend in a budget meeting. If the figure is affordable, fund it properly and leave it alone long enough to exit learning. If it is not, narrow the campaign until it is, or fix the conversion rate so the floor comes down. The one option that reliably wastes money is running the campaign you planned at a fraction of the budget it needs.