What time of day should your ad actually run?

Somebody with a large budget runs ads continuously and lets the platform work out when they perform. Somebody with $40 has to pick a window, and ad scheduling stops being an optimisation and becomes the entire decision. The awkward part is that the advice available is almost all written for the first person. (Disclosure: published by BidSurvivor, which sells advertising in fixed blocks of hours, so we are not neutral on whether the block you choose matters.)

The advice you will find, and why it does not apply

Search for the best time to advertise and you will get charts: Tuesday mornings for B2B, evenings for consumer, avoid Fridays. These come from aggregate platform data across thousands of accounts, and they have a specific problem when applied to you.

They describe averages across advertisers, which means they describe where competition is. If Tuesday 10am is the reported best hour for B2B, every B2B advertiser reading the same article is bidding on Tuesday 10am. In an auction, the reported best hour is the most expensive hour, and the reported worst hour is where the discount lives.

That does not make the charts wrong. It makes them a description of the market, not a recommendation for you — and the two get conflated constantly.

What actually varies by hour

Three things move, and they move independently:

How many people are awake. The real one, and the biggest. Global internet traffic has a pronounced daily shape, visible in public aggregates like Cloudflare Radar. If your audience is concentrated in one country, half the day is close to empty.

What mood they are in. Someone browsing at 11pm is in a different state from someone at their desk at 10am. Neither is better; they are better for different asks. Impulse signups and long evaluations do not peak together.

What it costs. In any auction, price follows demand. The busiest hour is the dearest, so more attention is available and you are paying more per unit of it. Whether that is a good trade depends entirely on your conversion rate at that hour, which is the number you do not have.

The rule when you have no data

You do not have historical performance by hour. Nobody does before their first campaign. So the choice has to be made on structure rather than evidence, and the structure that holds up is:

Pick the hours your audience is at the thing you are asking them to do. Not the hours they are awake — the hours they are in the context. A developer tool is bought by a developer at a keyboard, and that is business hours in their timezone, which is a narrow window and worth paying for. A consumer app is opened on a sofa. A B2B service being evaluated by a committee is looked at during work and decided over a week, so the hour matters less than the day.

Pick one timezone and commit. The single most common mistake in small-budget scheduling is running "9am to 5pm" against an audience spread across three continents, which is nobody's 9am and everybody's random hour. Choose the timezone where most of your audience is and buy the block that lands there, even if it is the middle of the night where you are.

Prefer a contiguous block to a scattering. Twelve consecutive hours gives one clean result you can interpret. Four scattered two-hour slivers give you four numbers with samples too small to compare — which is the sample size problem in a different costume. When you can only afford one measurement, make it a legible one.

Do not buy the hour everyone recommends first. If you are testing, the cheap hour tells you more per dollar. If your product only works at the expensive hour, you will find out from the cheap one failing badly rather than from the expensive one succeeding mildly.

What one day can and cannot tell you

Be honest about the ceiling on a single block.

It can tell you whether the offer lands at all. A creative that gets no clicks in twelve hours at a reasonable slot is not suffering from a timing problem.

It can tell you roughly what a click costs you in that context, which is the number you need for any planning at all.

It cannot tell you that Tuesday beats Thursday. That comparison needs both, several times each, with enough volume in each to be distinguishable. One Tuesday against one Thursday is two coin flips.

It cannot separate the hour from everything else that happened that day. A news event, a competitor's launch, a holiday you forgot about — all of these live inside your one measurement and cannot be extracted from it.

This is the honest limit of a one-day ad campaign: it is an excellent instrument for "does this work at all" and a poor one for "which of these two is better". Use it for the first question.

A workable default

If you genuinely have no information and one block to buy:

1. Identify the single country most of your likely buyers are in.
2. Take the block that covers late morning to early evening there — the widest span of that country's waking, working, deciding hours.
3. Buy it, once, with one creative and one landing page.
4. Record clicks and what they did, and write down what else was happening that day.

Then do it again a week later with one thing changed. Two clean observations a week apart beat eight muddy ones in an afternoon, and the difference is entirely in whether you can say afterwards what caused what. That is also why buying attention at the last minute works better than it should: an unsold block at a known hour is a cheap, legible measurement, and legibility is the scarce thing when the budget is small.