Your conversion rate at 9 p.m. is not your conversion rate at 3 a.m. Your cost per click at 7 a.m. on a Monday is not your cost per click on a Saturday afternoon. The console knows none of this: it sets one bid per keyword and runs it around the clock. Every account is paying the same price for hours that behave completely differently.

This playbook is about closing that gap. It covers finding your account's hourly shape, turning it into a bid schedule, and running the schedule without doing it by hand.

Why hour matters

Amazon's auction prices a click the same whether or not the shopper is likely to buy at that hour. A bid that is right for your best hour is overpaying during your worst, and a bid that is right for the average is overpaying for half the day and underpaying for the other half.

The size of the gap varies by category. Impulse and consumer categories tend to peak in the evening and go quiet overnight. B2B and work-related categories peak in weekday business hours. Some categories are flat. The point is not that every account has a dramatic hourly curve; it is that most sellers have never checked whether theirs does, and the ones that do usually find a wider spread than they expected. The overnight ad cost post shows what the quiet hours tend to look like.

Step one: find your shape

Pull hourly performance data for the account. Amazon's reporting has offered hourly breakdowns for some report types, and third-party tools expose it more directly. You want, per hour of day and ideally per day of week: spend, clicks, orders, and sales.

Compute conversion rate and ACoS per hour. Then rank the hours. What you are looking for is a tiered shape: a set of hours where conversion is clearly above the daily average, a set where it is near average, and a set where it is clearly below or where orders are so sparse the hour is effectively dead.

Use at least four weeks of data, more for lower-volume accounts. A single hour with two orders in a month is not a strong hour; it is noise. Set a minimum order count per hour before you trust the number.

Step two: build three tiers

More than three tiers is precision the data does not support. Fewer than three misses the shape.

Peak hours. Conversion rate meaningfully above the daily average. These hours deserve a bid at or above the keyword's current level, because the clicks are worth more here. If the keyword is already at a bid that works on average, peak hours can take a modest increase.

Base hours. Near the daily average. The keyword's normal bid. Nothing to change.

Weak hours. Conversion rate well below average, or orders too sparse to measure. Two choices: lower the bid to match the hour's value, or pause the campaign for the hour entirely. Lowering keeps the ad visible at a price that fits; pausing removes the spend. For hours with some conversion, lower. For hours that are effectively dead, pause.

Write the tiers down as a table: hour ranges, day-of-week variations if the data shows them, and the bid treatment for each. That table is the schedule.

Step three: set the amounts

How much to raise in peak and lower in weak hours is a judgment call with a few guardrails.

Start modest. A 10 to 20 percent change per tier is enough to see whether the schedule moves ACoS without risking large swings. Larger changes can come once the first pass has run for a few weeks.

Keep floors and ceilings. A weak-hour reduction should never take a bid below the level where the ad stops showing at all, unless pausing is the intent. A peak-hour increase should never push a bid above what the keyword's conversion rate at that hour can justify against your ACoS target.

Treat campaigns separately where they differ. Brand campaigns, category campaigns and exact match campaigns often have different hourly shapes. One schedule for the account is a start; one per campaign type is better once the data supports it.

Step four: run it

This is where most sellers stop, because the console cannot do it. Changing a bid at 7 p.m. and changing it back at midnight, every day, on dozens of keywords, is not a task a person sustains.

The practical options. Pause and resume by hour, which the console also cannot do but which scheduling tools handle: the weak-hour tier becomes "paused" and everything else runs at the normal bid. This captures most of the value with the simplest rule. The dayparting guide covers this approach in full.

Bid changes by hour, where a tool supports it: the full three-tier schedule, applied to bids on a cadence. More control, more to maintain.

Bulk uploads on a schedule: possible by hand, two uploads a day, every day. Nobody keeps this up for more than a few weeks, which is why it belongs to software.

Step five: read it and revise

After the schedule has run for four weeks, pull the hourly data again and compare. The questions: did ACoS in the weak hours fall without total sales falling? Did peak-hour sales rise enough to cover the higher bids? Did the hourly shape itself change, which it will as seasons shift?

Revise quarterly at minimum. Hourly shapes move with daylight, with school calendars, with events. A schedule built in April is stale by October, and the one built in October is wrong by January.

What this does not replace

Hourly bidding is a layer on top of a keyword and bid structure that already works. It does not fix an irrelevant keyword or an uncompetitive listing; it makes a good keyword cheaper during the hours it is worth less. Get the structure right, then schedule it. Budget rules handle the daily cap side of the same problem, and the two work together.

Frequently asked questions

Can you change Amazon PPC bids by time of day in the console?

No. The console sets one bid per keyword or target with no hour-of-day dimension. Placement modifiers and dynamic bidding adjust bids by position and by Amazon's conversion estimate, not by clock time. Hourly bid control comes from outside tools or from manual changes made by hand at the hour.

Is it better to pause campaigns or lower bids during weak hours?

Lowering bids keeps the campaign visible at a price that matches the hour's conversion rate, and keeps impression history steady. Pausing removes spend entirely. For hours that convert at a fraction of the daily average, a lower bid is usually right. For hours that produce clicks and essentially no orders, pausing is cleaner.

How much hourly data do I need before building a bid schedule?

Enough that each hour has a readable number of orders, not just clicks. For most accounts that means four to eight weeks of hourly performance data. Below that, a single large order at 3 a.m. can make a dead hour look alive.

Off Hours runs the schedule: dayparting rules that pause and resume campaigns by hour and day, every day, from a table you build once. Start a free 14-day trial.