Google Ads Pricing: What Actually Decides the Number

Anyone searching for google ads pricing is usually asking one of two questions, and the answer they find is almost always to the other one. The first question is what Google charges to run ads. The second is what it will cost me to get customers this way. Google publishes an answer to neither, because there is no price list — there is an auction, and the auction produces a different number for every advertiser, every day.

This article explains what actually determines the number, which parts of it you control, what the surrounding costs are that nobody quotes, and how to build an estimate for your own business that is defensible rather than borrowed from a benchmark article about a different market.

So what is google ads pricing, in one paragraph?

There is no price list. You set a daily budget and pay when someone clicks. What a click costs is decided by an auction between everyone targeting that search, where position depends on your bid and on a quality assessment of your ad and page. A stronger ad pays less for the same position.

Everything else in this article is detail on that paragraph, and the detail is where the money is.

The four numbers that decide what you pay

1. Cost per click, which you influence but do not set

Your bid is a ceiling, not a price. You are entered into an auction, and what you actually pay is closer to the minimum needed to hold your position against the advertiser below you. This is why raising a bid often does not raise costs proportionally, and why lowering a bid sometimes costs nothing in position and sometimes costs everything.

The gap between advertisers on the same keyword is wide. Two businesses bidding on the same term can pay meaningfully different amounts per click because one has a more relevant ad and a landing page that matches what the ad promised. That difference is the closest thing to a lever you have on unit cost.

2. Volume, which is set by demand, not by you

You cannot buy more searches than exist. This is the ceiling nobody mentions in cost discussions: if two hundred people a month search for what you sell, no budget produces a thousand visitors. Spending more in that situation means bidding more for the same clicks, or expanding into terms with weaker intent, both of which raise cost per customer.

Check search volume before building any budget model. A plan that assumes you can spend a given amount every month is a plan that assumes demand you may not have.

3. Conversion rate, which is mostly not about ads

Cost per click times clicks gives you spend. Spend divided by conversions gives you the number that matters. And conversion rate is decided almost entirely by things outside the ad account: the offer, the price, the landing page, how quickly you follow up, whether the form asks for eleven fields.

This is why two businesses with identical cost per click can have costs per customer that differ by a factor of three. The ad account is the smaller half of the equation, which is inconvenient because it is the half with the dashboard.

4. The value of a customer, which decides whether any of it works

A cost per customer is meaningless until it is set against what a customer is worth. The same figure can be excellent for a business with repeat purchases and ruinous for one selling a single low-margin item.

Work out the gross margin on a first purchase and, if you can, the probability and value of a second. Businesses that only compare cost per sale to first-purchase margin systematically underinvest; businesses that assume a lifetime value they have never measured systematically overspend.

The costs nobody quotes

The click cost is the visible part. Four others sit around it, and together they routinely exceed it for small advertisers.

Cost	Typical shape	Who forgets it
Management time	Several hours a week while learning, an hour or two once stable	Everyone running it themselves
Creative and landing pages	Front-loaded, then recurring as pages and offers are tested	Anyone treating the ad as the whole job
Measurement setup	One-off but real: conversion tracking, server-side events, CRM connection	Almost everyone, until the numbers stop making sense
The learning period	The first weeks buy information rather than customers	Anyone judging the channel on month one

The learning period is the one that causes the most arguments internally. Early spend is genuinely worth less per unit than later spend, because you are paying to discover which terms, audiences and messages work. Treating month one's cost per sale as the expected steady state produces one of two errors: cancelling something that was about to work, or celebrating a number that will not hold.

Building an estimate for your own business

Four steps, half an hour, and the result is more useful than any published benchmark because it is built from your numbers.

Step one: find the demand. List the searches someone would type when they are ready to buy what you sell — not research terms, buying terms. Get monthly volumes for your country. This gives you the size of the pool.

Step two: estimate a click cost range. Google's own planning tools give a range per keyword. Take the higher end for planning; the low end assumes a quality position you have not earned yet.

Step three: assume a conversion rate you can defend. If you have web analytics, use your existing rate for comparable traffic and reduce it, because cold ad traffic converts worse than people who arrived by search or referral. If you have no data at all, model three scenarios rather than one number.

Step four: compare to margin, not to revenue. Cost per sale against gross margin on the first purchase. If it clears with room, the channel works at that volume. If it clears only on assumed repeat business, write down the assumption and check it before scaling.

What comes out is a range, not a number, and that is the correct output. Anyone offering you a single figure for what Google Ads will cost your business is offering you a guess with the uncertainty removed.

What actually moves your cost per click down

Three things move it, in descending order of leverage, and none of them are bidding tricks.

Relevance between search, ad, and page

The auction rewards ads that answer the search and pages that deliver what the ad promised. The most common structural mistake is one landing page serving twenty different searches. Each of those searches is a slightly different question, and a page that answers none of them precisely pays a premium on every one.

Splitting a campaign so that closely related searches share a page written for them is the single highest-return piece of account work available to most advertisers, and it costs nothing but effort.

Excluding searches you do not want

Broad targeting brings in searches adjacent to yours: people looking for free versions, jobs at your company, definitions, competitors' product names, or the same words meaning something else entirely. Every one of those clicks costs the same as a good one.

Reviewing the actual search terms — the real queries, not the keywords you chose — and excluding the ones that can never convert is unglamorous weekly work with a direct effect on cost per sale. Most accounts that have never done it will find between a fifth and a third of spend going somewhere it should not.

Not competing with yourself

Accounts grow by accretion. Someone adds a campaign for a new product, and it overlaps with two existing ones. Now three of your campaigns are eligible for the same search and the account is bidding against itself, raising the price of the auction it is winning either way.

Where the money leaks, in the order you will find it

If you audit an account that has been running without close supervision, the leaks appear in a predictable sequence.

Search terms nobody excluded. Always first, always largest. Months of spend on queries that were never going to buy.

Ads still running for things you no longer sell. A discontinued product, an old price, a promotion that ended in March. These convert badly and, worse, produce customer service work.

Budget stranded in a campaign that cannot spend it. A campaign with narrow targeting sits at half its budget while a campaign that could use the money is capped. Nothing is being wasted exactly, but the money is not working.

Geographic settings nobody checked. The default targeting includes people merely interested in a location, not only people in it. For local businesses this is a steady quiet drain.

Devices and hours performing very differently, treated identically. Not always worth acting on — but worth knowing before you conclude the whole channel is expensive.

None of these are exotic. They are what an account accumulates when nobody has time to look, which is precisely the situation most small advertisers are in.

Doing it yourself, hiring an agency, or automating it

The three routes cost different things in different currencies.

Yourself. Cash cost is only the ad spend. The real cost is attention, and the failure mode is not incompetence but discontinuity — the account gets three good weeks, then a busy month with no attention at all, and the leaks accumulate during the gaps.

An agency. You buy continuity and expertise. The economics work when spend is large enough that a percentage or retainer is small relative to the improvement. Below that threshold you are paying a meaningful fraction of your budget for someone who can only afford to spend an hour a month on your account. Ask how many hours yours will receive; the answer is more informative than the fee.

Automation. You buy continuity without buying hours. It handles the recurring, rule-shaped work — the search term exclusions, the pacing, the pausing of things that have clearly failed — and it does it on Saturdays. It does not handle strategy, creative, or the decision that the offer is wrong.

Orova Ads sits in the third category. It connects Google, Meta and TikTok, and its automation is built from 214 coded optimisation actions — 101 for Google, 58 for Meta, 55 for TikTok — each with explicit bounds, assembled into policies you write in plain language from 19 starter templates. It will not write your ad copy, design your landing page, or decide your positioning, and it cannot make a channel work where the margin does not support the cost per sale. What it removes is the gap between something going wrong and someone noticing.

How much should you start with?

The wrong way to set a first budget is to pick a round number you are comfortable losing. The right way is to work backwards from how much data you need before the numbers mean anything.

Here is the logic. To judge whether a campaign works, you need enough conversions to distinguish a result from luck. A handful is not enough — at five conversions, the difference between a good week and a bad one is one customer changing their mind. Somewhere around twenty to thirty conversions you can start forming an opinion, and the opinion firms up from there.

So: estimate your cost per conversion from the exercise above, multiply by the number of conversions you need to learn something, and that is the size of your first commitment. If that figure is uncomfortable, the honest options are to narrow the targeting so the money buys deeper data on a smaller question, or to accept that the test will take longer, or to conclude the channel is not affordable at your margin. What does not work is spreading a small budget across a wide target and stopping after three weeks, which produces a number with no information in it and a firm conclusion drawn from it anyway.

Spread that commitment across at least a month rather than compressing it. Daily budgets that are too low starve campaigns of the consistency the bidding systems need; budgets compressed into a fortnight buy a period of unusually expensive clicks as the account learns.

Why published benchmarks mislead

Search for what a click costs in your industry and you will find tables of averages. They are real numbers and they are close to useless for planning, for four reasons.

They average across countries. The same keyword can differ by an order of magnitude between markets, driven by how many advertisers compete and what a customer is worth locally.

They average across intent. "Legal services" includes both people looking for a definition and people who need a lawyer this week. Those are not the same auction.

They are survivor-weighted. Benchmark data comes from accounts that are still running. Advertisers who found the channel unaffordable stopped, and their numbers left the sample.

They tell you nothing about your conversion rate, which is half the equation and the half you control most.

Use benchmarks for one thing only: a sanity check on order of magnitude. If your estimate is thirty times the benchmark, you have made an arithmetic error. Beyond that, your own account's first month is worth more than any published table.

What changes between markets

Cost is not a property of the platform; it is a property of the auction, and auctions are local.

In markets with many well-funded advertisers, clicks are expensive and the quality bar is high, so relevance work pays for itself quickly. In thinner markets, clicks are cheap and the constraint moves to volume — you can afford the traffic but there is not enough of it, so growth comes from broadening what you target rather than optimising what you have.

Two practical consequences. First, cost advice written for one market can be actively wrong in another: "improve quality score to reduce costs" is excellent advice in a crowded auction and close to irrelevant where you are one of three bidders. Second, if you sell across several countries, resist a single blended cost target. A blended figure hides the country that is subsidising the others, and the usual result is scaling the wrong one.

The same reasoning applies to channels. Search costs more per click than most feed platforms and converts better, because the person typing has already decided they want something. Comparing cost per click across channels is comparing prices for different goods; compare cost per customer instead.

Questions people actually ask

Is there a minimum spend?

No minimum is imposed. There is a practical floor set by learning: below a certain daily budget, campaigns get too little data to stabilise, and you pay a premium for inconsistency. The floor is set by your cost per click, not by a published rule.

Does spending more make clicks cheaper?

Not directly. It can indirectly, because more data lets bidding tune better and because more volume makes structural improvements worth doing. But there is no volume discount. Someone spending a hundred times more than you is in the same auction on the same terms.

Do I pay for people seeing the ad?

On search, generally not — you pay on click. On display and video placements, other models apply, which is one reason mixing them into one cost figure produces confusion.

What is a good cost per click?

The wrong question, asked constantly. A high cost per click with a strong conversion rate and a valuable customer is a good business. A low cost per click on traffic that never buys is expensive at any price. Ask instead what a good cost per customer is for your margin, and work back.

Can I set a hard ceiling on total spend?

Daily budgets are the mechanism, and daily spend can exceed the daily budget on individual days while averaging out over the month — a fact that surprises people who set a budget expecting a hard cap. If a hard ceiling matters, monitor at the account level and use spend pacing rules rather than assuming the budget setting will enforce it.

How long before I know if it works?

Long enough to collect the conversion volume described above, plus a settling period. For most small advertisers that is six to eight weeks, not two. Decisions made at three weeks are usually reversed at seven.

A short checklist before you spend anything

1. Conversion tracking works and counts the right event. Not page views. Not form loads. The thing that means money.
2. You know your gross margin on a first purchase. An actual number, not a feeling.
3. You have checked there is search volume for buying-intent terms, in your country.
4. Your landing page answers the search the ad was shown for, above the fold.
5. Someone is responsible for reviewing search terms weekly for the first two months.
6. You have written down what result would make you stop, before you start. This is the one people skip, and it is why unprofitable accounts run for years.

How the auction resolves, in enough detail to be useful

You do not need the exact formula, and it is not published. You do need the shape, because the shape explains most of the surprises in a cost report.

When someone searches, every eligible advertiser is ranked by a combination of what they bid and how good their ad is expected to be for that specific search — expected click-through, how well the ad matches the query, and the experience on the page it leads to. Position follows that ranking. What you pay follows the ranking of the advertiser below you, not your own bid.

Three consequences follow, and each one explains a common complaint.

"I raised my bid and nothing happened." If you were already winning the position, raising the ceiling changes nothing, because the price is set by the competitor beneath you. Bids are ceilings, not offers.

"My costs jumped and I changed nothing." Someone else changed something. A new advertiser entered, or a competitor raised budgets. Your account is not a closed system; roughly half of what happens to your numbers is other people's decisions.

"My competitor is above me and I know they bid less." Entirely possible. A stronger ad and a better-matched page can outrank a higher bid, and pay less per click while doing it. This is the mechanism that makes relevance work financially worthwhile rather than merely virtuous.

The practical takeaway is that cost is partly a competitive variable you do not control, which is why cost targets should be reviewed rather than set once. An account that was comfortably profitable in March can be marginal in September without anyone doing anything wrong.

Seasonality, and the budget you set in January

Auction prices move with demand. In seasons when your competitors want customers most, clicks cost more — sometimes substantially. A budget set as a flat monthly figure will therefore buy very different amounts of traffic across the year, and a cost-per-sale target that is achievable in a quiet month can be impossible in a peak one.

Two ways to handle it, both better than a flat budget. Set the budget as a share of expected revenue rather than a fixed amount, so it flexes with the season it is buying into. Or set explicit seasonal targets: a tighter efficiency target in quiet months where you are buying at a discount, a looser one in peaks where the customers are actually there.

What causes real damage is holding a single annual efficiency target and cutting spend whenever it is missed. That systematically withdraws from the market exactly when demand is highest, which feels disciplined and is expensive.

When search advertising is the wrong channel

Three situations where the arithmetic will not work no matter how well the account is run.

Nobody is searching for what you sell. New categories have this problem. If the product solves a problem people have not named, there is no query to bid on, and the money belongs in channels that create demand rather than capture it.

The margin cannot absorb a competitive click. If a click costs a meaningful fraction of your gross margin per sale, you need a conversion rate that few businesses achieve. Check this before building the account, not after two months.

The purchase involves a long, human sales process you cannot staff. Search can generate enquiries faster than a small team can follow up. Leads that go cold cost exactly as much as leads that convert. The constraint is the follow-up capacity, and buying more leads makes it worse.

Noticing any of these early is worth more than any optimisation. The most expensive Google Ads accounts are not the ones with high costs per click; they are the ones that should never have been switched on and ran for a year because nobody had written down what failure would look like.

Reading the first month's numbers without fooling yourself

Three habits make the difference between a first month that teaches you something and one that generates a confident wrong conclusion.

Look at search terms before you look at totals. The totals tell you the outcome; the search terms tell you why. An expensive month where nine tenths of the queries were irrelevant is not evidence about the channel — it is evidence about targeting, which is fixable in an afternoon.

Separate the learning period explicitly. Mark the first two or three weeks in whatever you report. Blending them into the month drags every average down and makes the trend invisible.

Count conversions where the money is, not where it is convenient. If the conversion you track is a form submission but half of those are unqualified, your real cost per customer is double what the account reports. Tracking the further-down event — the qualified enquiry, the closed sale — is more work to set up and it is the difference between managing a business and managing a dashboard. Connecting the outcome back into the ad platform is what makes automated bidding optimise toward revenue instead of toward form fills.

Two numbers to put on the wall

Accounts that stay profitable over years tend to have two figures that everyone involved knows without looking them up.

The maximum you can pay for a customer and still make money. Derived from gross margin, not revenue, and stated as a single number. Everyone who touches the account should be able to say it. Without it, every conversation about whether costs are too high becomes a matter of opinion, and opinions default to whoever is most senior rather than whoever is right.

The point at which you stop. Written down before launch. Something like: if after eight weeks and a defined spend the cost per customer is above the ceiling and not trending toward it, the channel closes. This sounds pessimistic and is the opposite — it is what makes it safe to start, because the downside is bounded in advance rather than discovered through attrition.

Both numbers should be revisited when the business changes: a price rise, a new product, a shift in what a customer is worth. What should not happen is quietly raising the ceiling because the account keeps missing it. That is the mechanism by which unprofitable advertising survives — not a bad decision, but a series of small revisions each of which felt reasonable.

What automation changes about the cost question

Automation does not lower the price of a click. The auction does not care whether a human or a system placed the bid. What it changes is the second-order costs, which for most small advertisers are larger.

It shortens the gap between something going wrong and something being done about it. A search term that starts consuming budget on a Tuesday gets excluded on Tuesday rather than at the next review, which for a lot of accounts means a fortnight later. Over a year, the sum of those avoided fortnights is usually a larger number than any bidding improvement.

It also removes the discontinuity problem. Accounts run by busy people do not get consistent attention; they get bursts. The leaks accumulate in the quiet weeks, and the burst of attention is spent finding them rather than improving anything. A system that handles the recurring work continuously changes what the human hours are spent on — from finding leaks to deciding strategy.

What it does not change: the margin, the offer, the volume of demand, or whether the landing page answers the question. Those set the ceiling, and no amount of optimisation lifts a ceiling. If the arithmetic does not work at a competitive cost per click, automation makes the losses more efficient rather than turning them into profit — and it is worth being clear-eyed about that before buying anything, including ours.

A worked example, with the arithmetic shown

Numbers invented for illustration — the point is the shape of the calculation, not the values.

A business sells a service with a first-purchase gross margin of a hundred units. Buying-intent searches in its country total around two thousand a month. Planning tools suggest a click cost in the region of two units at the top of the range. The site converts warm traffic at four per cent, so cold ad traffic is modelled at two and a half.

If it captured a fifth of available searches — an ambitious share for a new advertiser — that is four hundred clicks, costing eight hundred units, producing ten customers at eighty units each. Against a hundred units of margin, that clears, but only just: a twenty per cent buffer leaves no room for a bad month or a competitor entering.

Three readings follow, and all three are legitimate. The optimist notes that conversion rate is the softest assumption and that landing page work could plausibly move it, turning a thin margin into a comfortable one. The pessimist notes that the click cost estimate is the high end of a range that could go higher in a peak season, and that ten customers a month is too few to distinguish a trend from luck. The realist notes the ceiling: even at perfect execution, two thousand searches caps this channel at a size, and if the business needs to grow faster than that cap allows, search is one channel among several rather than the plan.

Run the same three readings on your own figures. The exercise takes twenty minutes and it converts an argument about whether Google Ads is expensive into a specific disagreement about which assumption is wrong — which is a disagreement you can settle with a month of data.

The mistake that costs the most

Of everything in this article, one error dominates the others in cost, and it is not a bidding mistake or a targeting mistake. It is running an account for months without a defined answer to the question "is this working".

It happens because the account always produces some result. There are always clicks, always a few conversions, always a plausible story about why last month was unrepresentative. Without a ceiling written down and a stopping rule agreed in advance, that story can be told indefinitely, and each month's version is individually reasonable. The accounts that quietly consume a year of budget are almost never run by people making obviously bad decisions; they are run by people who never agreed what success would look like and therefore could not recognise its absence.

The fix costs an afternoon. Write the maximum cost per customer your margin supports. Write the spend and the timeframe you are willing to commit. Write what result ends the experiment. Put all three somewhere a colleague can see them. Then start.

The short version

Google ads pricing is not a price list. You pay per click, what you pay is decided by an auction weighted by relevance, and the number that matters is not cost per click but cost per customer measured against your margin. Volume caps what you can spend, conversion rate decides whether spending is worth it, and both are mostly determined outside the ad account.

Budget from the data you need rather than from the money you can spare, treat the first six weeks as buying information, ignore published benchmarks beyond a sanity check, and audit search terms before concluding the channel is expensive. Most accounts that look unaffordable are not paying too much per click; they are paying full price for clicks that were never going to buy.

One closing caution about the search that brought you here. Queries about what a channel costs tend to be answered by people selling the channel, or by people selling an alternative to it, and both have an interest in the number. The only figure that will settle the question for your business is the one your own account produces in its first eight weeks, measured against a margin you calculated before you started and a stopping rule you wrote down while you were still calm.

None of which is a reason to avoid the channel. Search advertising remains the most direct way to reach someone who has already decided they want what you sell. It is simply a channel whose price is set by an auction rather than a list, and the advertisers who do well in it are the ones who treat the number as something to be measured against their own margin rather than compared against somebody else's average.

Related reading: the real cost of managing ads manually, agency versus in-house versus an AI agent, and blended CAC explained. The product is at orova.vn.