Most Amazon PPC decisions get made on daily numbers. A keyword spent $20 yesterday and made $60, so ACoS is 33%, so the bid is fine. But that $20 wasn’t spent evenly. Some of it went out at lunchtime, when shoppers were buying, and some went out at 2 AM, to people who were just scrolling.

Dayparting means changing your bids by time of day: bidding more in the hours your ads sell, and less in the hours they don’t. Sellers argue about whether it’s worth it. So we looked at 60 days of hour-by-hour data from our own account.

What the hourly data showed

Across the whole account, ACoS for those 60 days was 35%. But almost nobody’s ads actually ran at 35%. Midday ran at 23%. Overnight ran at 86%, meaning almost every dollar spent overnight went to ads that barely paid for themselves.

The clearer way to see it is where the money went compared with where the sales came from:

Two things jump out:

• Overnight took 11% of the budget for 4% of the sales. It’s not a huge slice of spend, but it’s the least efficient money in the account.

• Midday was underfunded. It got 21% of the budget and produced a third of the sales. Clicks were cheaper too: $1.52 on average after 11 AM, against $2.08 in the morning.

Why hours behave so differently

A few things change over the course of a day:

• Who’s shopping. Late-night shoppers browse more and buy less. Midday shoppers are more likely to be ready to buy.

• Who’s bidding against you. Many advertisers set daily budgets that run out by the afternoon. As they drop out, there’s less competition, and clicks get cheaper. That’s one likely reason our afternoon clicks cost about 25% less than morning ones.

• Time zones. Amazon reports in your account’s time zone, but your shoppers are spread across the country. 6 AM Pacific is already 9 AM on the East Coast.

Your pattern won’t match ours exactly. Different products sell at different times. That’s the point: you can only see your pattern with hourly data, and daily reports can’t show it.

How to daypart without breaking your campaigns

Dayparting goes wrong when it’s done with too little data or too heavy a hand. Five rules we follow:

1. Use weeks of data, not days. One hour on one day is a handful of clicks. Look at the same hour across at least a few weeks before you act on it.

2. Adjust bids, don’t switch ads off. Turning a campaign off overnight throws away the sales that do still come in. Lower the bid instead.

3. Start with the extremes. Cut the hours that are clearly worst and raise the clearly best. Leave the middle alone at first.

4. Move in small steps. A 15–30% change is plenty to start. You can always go further once the numbers back it up.

5. Check that budget lasts into your best hours. If your campaigns run out of budget by noon, your best hours never get a chance. Fixing the budget may matter more than any bid schedule.

Here’s what a first schedule might look like for an account like ours. It’s an example only, not the math Pulse Pilot itself uses:

Three dayparting mistakes we see often

• Deciding from a week of data. A bad Tuesday at 3 PM is noise. A bad 3 PM across six weeks is a pattern.

• Setting it once and forgetting it. Shopping patterns shift with seasons and holidays. A schedule that worked in September may be wrong by December.

• Dayparting a campaign that’s out of budget. If it stops spending at 1 PM every day, its “afternoon” numbers are empty, and the schedule will be built on a gap.

How to set this up in Pulse Pilot

Pulse Pilot runs your Amazon ads two ways: Autopilot, which runs them hands-off toward your goal, or Pilot, your own sandbox for building rules.

• Set your own hour-by-hour schedule. In Pilot, open Automations → Dayparting to set an hour-by-hour bid schedule for the keywords, targets or campaigns you choose. You pick the hours; the schedule runs on the clock.