How Much to Spend on Ads to Test a Business Idea Before Launch

Budget for about 300 search ad clicks, which at a realistic cost per click comes to roughly $500 if your category is among the cheapest, about $1,600 at the cross-industry average. Treat anything under 100 clicks as noise: even at 100 clicks and a 5% conversion rate, the plausible true rate spans a factor of five. Before you spend anything, read the cost per click itself, because it tells you what the demand is worth.

How many clicks before a conversion rate means anything?

When you run ads to test an idea, you are buying one number: the share of visitors who take the action you care about. That share is only useful if it's precise enough to support a decision. Precision depends on how many visitors you send, and it improves much more slowly than most people expect.

The table below shows the 95% confidence interval (Wilson method) around an observed 5% conversion rate at different traffic levels. The interval is the range where the true rate plausibly sits.

Clicks	Conversions at 5%	Plausible true rate	High end ÷ low end
60	3	1.7% to 13.7%	8x
100	5	2.2% to 11.2%	5x
200	10	2.7% to 9.0%	3.3x
300	15	3.1% to 8.1%	2.6x
400	20	3.3% to 7.6%	2.3x
800	40	3.7% to 6.7%	1.8x

At 60 clicks, a "5% conversion rate" could be a dead idea at under 2% or a great one near 14%. The data can't tell those apart, so you haven't learned anything. At 300 clicks the range is narrow enough to compare against a break-even threshold. Going from 300 to 800 clicks multiplies the cost by 2.7 and narrows the high-to-low ratio from 2.6x to 1.8x, which rarely changes a go or no-go decision.

For a low conversion rate, precision is driven almost entirely by the number of conversions, not the number of clicks. Fifteen conversions give you a usable estimate and five do not. So think of the target as "enough clicks to collect about 15 conversions at the rate I need to beat."

Where is the line between signal and noise?

The line is 100 clicks per thing you're testing. Below that, don't read the conversion rate at all, not even as a "directional" signal. "Directional" is usually what people call noise they've already paid for.

This is why the common "$100 on ads to see what happens" test is weak. At the 2026 average search CPC of $5.42, $100 buys about 18 clicks. Even at the median landing page conversion rate of 6.6%, 18 clicks produce about one conversion. One conversion or zero is a coin flip, not a verdict on an idea.

There is one exception where 100 clicks can decide on its own: zero conversions. If 100 people clicked and none converted, the upper end of the plausible true rate is about 3.7%. (The quick version is the rule of three: with zero events in n trials, the true rate is probably below 3/n.) If your economics need 5% or more, zero in 100 is enough to kill the idea. A positive result needs more traffic than a negative one.

What will those clicks cost?

Cost per click varies more than any other input, so use a range rather than one number. Across industries, the average search CPC is $5.42 in LocaliQ's 2026 data. WordStream's 2025 figures put the cheapest categories at Arts & Entertainment ($1.60), Restaurants & Food ($2.05), and Travel ($2.12). The table prices the low end at those WordStream categories and the high end at the LocaliQ 2026 average.

Clicks	At $1.60 CPC	At $2.12 CPC	At $5.42 CPC
100 (noise floor)	$160	$212	$542
300 (decision-grade)	$480	$636	$1,626
800 (fine-grained)	$1,280	$1,696	$4,336

These averages are only a starting point. Your own category's CPC is what matters, and Keyword Planner shows it.

Display and social clicks usually cost less than search clicks. Don't use that to save money on a validation test. A search click comes from someone who typed the problem into a box. A display click comes from someone who was curious about a banner. Their conversion rates measure different things, and only the first one tells you about demand.

Why is the cost per click a demand signal on its own?

A cost per click is a price set at auction. Every advertiser bidding on a keyword has decided what a click is worth to them, and advertisers who overpay for long run out of money. A high, stable CPC on a keyword means several businesses are turning those searches into revenue. A high CPC can also be distorted by a few well-funded bidders, so read it alongside the competition level.

Google says Keyword Planner shows what other advertisers are paying for keywords as historical bid ranges, and ranks each keyword's competition as low, medium, or high. Those top-of-page bid ranges come from other advertisers' bids. They are not a quote for your own account.

You can also estimate what the market pays per customer. Divide the CPC by a plausible conversion rate. A $6 click that converts to a lead 10% of the time means advertisers are paying $60 per lead, close to LocaliQ's 2026 average cost per lead of $66.69. Businesses only keep paying $60 per lead if a lead is worth more than that.

Here is how to read the combination of CPC, competition and volume:

What Keyword Planner shows	What it means	What to do
High CPC, high competition, solid volume	Proven money. Incumbents are profitable.	Test only if you have a sharper angle or a cheaper way to reach the same buyers.
Low CPC, low competition, solid volume	Either a gap, or nobody has worked out how to make money from these searchers.	Test. This is where a cheap test can find something real.
Low CPC, near-zero volume	People aren't searching for this.	Search ads can't test it. Use a different channel or rethink the problem.
High CPC, near-zero volume	A niche worth a lot per customer, usually B2B.	Clicks will be scarce. Direct outreach may be faster than ads.

Your own test also gives you a CPC signal. Inside your own campaign, the top-of-page bid estimate in the Keywords report reflects your Quality Score, so a very high estimate can mean your Quality Score is poor rather than that demand is strong. Plan for a new account with a new landing page to pay above the estimate. If you're bidding well above the range and still not getting impressions, the problem is usually your ad's relevance, not demand.

What does a full test budget look like?

Say you're testing a $49/month scheduling tool for independent dog groomers. Every figure in this example is hypothetical. Keyword Planner shows top-of-page bids of about $1.80 to $4.20 on your core terms, and you plan around $3.50.

Step 1: Put a value on a conversion. Your conversion is someone clicking "Start trial" on a page that shows the price, then entering an email. Assume a paying customer stays 8 months: $49 × 8 = $392. Assume 1 in 4 signups would actually pay. Each signup is then worth $392 ÷ 4 = $98.

Step 2: Find the break-even conversion rate. $3.50 ÷ $98 = 3.6%. Below that, every signup costs more than it's worth. You want room to spare, so set a healthy target at double: 7%.

Step 3: Size the test. If the true rate is 7%, about 120 clicks is enough for the interval's low end to clear 3.6%. If the true rate is 3.6%, it takes about 220 clicks for the high end to fall below 7%. Plan for 300 clicks so either outcome can resolve.

Step 4: Price it. 300 × $3.50 = $1,050. Cap it at 300 × $4.20 = $1,260 in case clicks come in at the top of the range.

Step 5: Decide in advance what each result means.

• 0 signups at 100 clicks: kill.
• Interval low end above 3.6%: pass, keep going.
• Interval high end below 7%: kill.

Now suppose you finish at 300 clicks with 15 signups, a 5% rate. The interval is 3.1% to 8.1%, which crosses both lines. Don't spend another $1,000 to settle it. A result that lands between the lines means the idea is marginal at this price. Raise the price, which lowers the break-even rate, or move on.

How do you run the test without wasting the budget?

1. Make the conversion cost the visitor something. A price on the page plus an email, or a card, is the minimum. A free waitlist signup inflates the rate. Unbounce's median covers 464,000,000 visits to 41,000 landing pages, and what counts as a conversion varies a lot between them. Don't compare a harsh test against that benchmark.
2. Use exact and phrase match on problem keywords. Broad match sends low-intent clicks, and your conversion rate stops measuring demand.
3. Set the daily budget to total cap ÷ test days, and watch it. Google says a campaign might spend up to twice your average daily budget on any given day. It caps spend at 30.4 times the daily budget over a month, not over your two-week test. Pause the campaign yourself when you hit the total.
4. Test one pitch, not three. Splitting 300 clicks across three ad variants gives you three samples of 100, which is three results at the noise floor. Comparing variants takes far more traffic. If you need to compare, use a proper tool such as Evan Miller's sequential sampling calculator, which tells you how many conversions a two-variant test needs.
5. Don't stop early because the numbers look good. Early leads fade. Check the kill rule at 100 clicks and make every other decision at 300.

When should you not use ads to test at all?

Don't use search ads when Keyword Planner shows almost no volume for the problem. You can't buy clicks that don't exist, and a two-month trickle of traffic gets contaminated by seasonality and ad-auction changes.

The same goes when a break-even conversion rate comes out above 20%. That usually means the CPC is too high for your price point. That's a finding in itself: the channel your competitors use to acquire customers won't work at your price. You learn this from the arithmetic before spending anything, which makes it the cheapest result a paid test can give you.