Generic advice says pause ads overnight. Sometimes that is right. Sometimes overnight is when your best customers shop. The only schedule worth running is one built from your own account's data, and building it takes an afternoon, not a data science team.

This playbook walks through the full process in six steps: pull the data, score each hour, decide on actions, group campaigns, roll it out and measure the result.

Step 1: pull the right data

You need spend, clicks, orders and sales by hour of day, ideally split by day of week, for at least four weeks. Amazon's hourly reporting and Amazon Marketing Stream are the usual sources. Reading hourly performance data covers where to find it and how to export it.

Exclude event days such as Prime Day or Black Friday week. Their hourly patterns are unusual and will distort a normal schedule. Make sure the data uses the marketplace's time zone, since that is the clock your schedule will run on.

Step 2: score each hour

For each hour, compute conversion rate (orders divided by clicks) and ACoS (spend divided by sales). Then compare each hour to the account average.

A simple scoring scale works well:

Strong hours. Conversion rate clearly above average and ACoS below target.

Average hours. Close to the account average either way.

Weak hours. Conversion rate well below average, for example under half of it, with ACoS well above target.

Dead hours. Meaningful spend and almost no orders over the whole period.

Check the number of clicks behind each score. An hour with very few clicks can look terrible or wonderful by chance. Only score hours with enough clicks to trust, and leave the rest as average until more data arrives. The hourly heatmap is a helpful way to see the whole week at once.

Step 3: decide the action for each hour

Strong hours: full bids, and make sure the budget lasts into them. If campaigns run out before strong hours, the schedule needs to free budget earlier in the day.

Average hours: leave alone. Cutting hours that perform near average saves little and costs sales.

Weak hours: lower bids rather than pause. A reduced bid keeps some presence and lets you see if the hour improves.

Dead hours: pause. These are the hours where scheduling saves the most.

Pause vs lower bids covers the tradeoff in more depth.

Step 4: group campaigns by pattern

Not every campaign shares the same curve. A gift product and a work product in the same account can peak at opposite times. Score hours for each product group separately if you suspect a difference, and build one schedule per group with a clearly different pattern.

Keep the number of schedules small. Two or three is manageable. One per campaign is not. Best times to run ads by category gives typical patterns to compare your groups against.

Decide whether weekdays and weekends need separate schedules. If weekend shopping starts later or runs more evenly, a single schedule will be wrong for part of the week. Day of week performance covers how to check.

Step 5: roll it out safely

Start with half. Apply the schedule to about half of the campaigns in a group and leave the rest unchanged. This gives you a comparison over the same weeks, which removes seasonality from the test.

Start gentle. Use bid reductions before pauses on all but the clearest dead hours.

Automate it. A schedule that depends on someone logging in at midnight and 6 a.m. will be missed. Use rules so it runs every day.

Write it down. Record the schedule, the date it started and which campaigns it covers, so the measurement later is clean.

Step 6: measure and adjust

After four weeks, compare scheduled and unscheduled campaigns on ACoS, ad sales and total sales. If the scheduled half shows better ACoS with similar sales, extend the schedule to the rest. If sales fell noticeably, look at which paused hours had more orders than expected and restore them.

Then set a review date. Once a quarter, pull fresh data and repeat steps 2 and 3. Patterns shift with the seasons, and a schedule left untouched for a year will drift out of step with your shoppers. Dayparting ROI covers how to measure the savings in detail.

Common mistakes to avoid

Scheduling on too little data. Two weeks of hourly data can make a random slow night look like a pattern. Use at least four weeks and check the click counts behind every hour you cut.

Pausing average hours. The savings come from the clearly weak and dead hours. Trimming hours that perform near average saves little and quietly costs sales.

Ignoring time zones. If your data is in your local time and your schedule runs in the marketplace's, every action lands in the wrong hour. Check both before you start.

Setting it and forgetting it. A schedule is a hypothesis about your shoppers. Review it each quarter, and before Q4, when evening and weekend shopping often grows. Each review takes far less time than the first build, because the method and the comparison are already in place.

Frequently asked questions

How much data do I need to build an Amazon ad schedule?

At least four weeks of hourly data, so each weekday appears several times. Eight weeks is better for smaller accounts, since hours with few clicks need more time to show a reliable pattern. Avoid periods that include major sales events, which distort normal hourly behavior.

Should I build one schedule for the whole account?

Start with one schedule for campaigns that share a similar pattern, usually grouped by product type. Products bought by different customers at different times need separate schedules. Branded and non-branded campaigns can also behave differently enough to justify their own.

How often should I update my Amazon ad schedule?

Review it at least once a quarter, and before and after major seasonal shifts such as Q4. Hourly patterns change with the season, the weather and the school calendar, so a schedule built in one season can be wrong in the next.

Off Hours turns the schedule you build here into dayparting rules that run on a 15-minute cadence, with every change logged for your four-week check. Start a free 14-day trial.