Your advertising runs 24 hours a day. Your customers do not buy that way. Every account has an hourly signature — conversion rates that swing 2–3× between the dead hours and the buying hours — and a flat, always-on bid strategy pays the same price for a 2am click that converts at 4% as for an 8pm click that converts at 12%.
Dayparting — scheduling bids and budgets against those hourly patterns — is one of the last "free" efficiency levers in mature accounts, and simultaneously one of the most misused tactics in immature ones, where thin hourly data gets tortured into schedules that optimise noise.
This guide covers the data you actually need, the patterns that repeat across categories, the budget-exhaustion dynamic that matters more than bids, implementation options, and the honest list of situations where dayparting is a distraction.
The Data First: Finding Your Real Hourly Signature
- Hourly campaign data is available in the Amazon Marketing Stream / hourly reports (and surfaced by most major PPC tools) — impressions, clicks, spend, sales by hour
- Volume threshold before believing anything: judge hourly conversion only where each hour-bucket has meaningful clicks across weeks — a good rule is analysing hour-of-week cells only after 100+ clicks per cell, else aggregate (weekday vs weekend, 4-hour blocks)
- Look at three curves separately: traffic (impressions), efficiency (CVR/ACOS), and competition (CPC) by hour — they do not move together, and the money is in their divergence
- Timezone discipline: reports and rules operate in specific timezones; a one-hour misalignment quietly shifts your whole schedule
The divergence insight: the best dayparting hours are rarely the highest-traffic hours — they are the hours where conversion holds up while CPCs sag (often late evening), and the worst are where traffic exists but intent is thin (typically deep night browsing).
Patterns That Repeat (and the Exceptions)
| Pattern | Typical shape | Caveat |
|---|---|---|
| Consumer B2C default | CVR builds through evening, peaks 7–10pm; trough 1–5am | Category-dependent; verify against your data |
| B2B/office products | Weekday 9–5 peaks, weekend soft | Inverse of consumer default |
| Payday/promotion rhythms | Month-start and event-driven surges | Layer onto daily shape |
| Weekend browsing | High traffic, mixed intent Saturday; Sunday evening strong | Split weekend from weekday in analysis |
| Impulse vs considered | Impulse converts flat-ish; considered purchases cluster in research hours | Price point drives this more than category |
Treat these as hypotheses to test against your own hourly export, never as schedules to copy. Two products in the same category with different price points routinely show different signatures.
The Budget-Exhaustion Trap (Bigger Than Bids)
Before optimising bids by hour, fix the silent version of dayparting you may already be doing by accident: budget exhaustion. A campaign that spends its daily budget by 2pm is dark during the evening conversion peak — you involuntarily day-parted yourself out of the best hours, and your competitors collect them.
- Diagnose: the "out of budget" indicators and hourly spend curves show when campaigns go dark; chronic early exhaustion on efficient campaigns is found money
- Fix order: raise budgets on efficient budget-capped campaigns first (cheapest win in most accounts), then shape bids by hour
- If budgets must stay capped, prefer lower bids all day over going dark at 2pm — presence through the peak usually beats intensity in the morning
Implementing Schedules: Rules, Tools, and Sanity
Native rule-based bidding
Amazon’s campaign rules support scheduled bid adjustments; the Marketing Stream powers near-real-time tooling. Native rules are coarse but free — sufficient for a first schedule (e.g. −20% 1–6am, +10% 7–10pm).
Third-party tools
Mature PPC platforms offer hour-of-week bid multipliers on top of base bids. Use one layer of automation — a tool’s dayparting stacked on Amazon’s dynamic bidding stacked on manual changes produces uninterpretable outcomes.
Rollout discipline
- Start with 2–4 bid blocks, not 168 cells: e.g. deep-night −25%, morning baseline, evening +10–15%
- Change one variable set, hold 2–3 weeks, compare like-for-like weeks (promo weeks excluded)
- Measure at the account/portfolio level — hourly reshuffling can look great per-campaign while total sales stay flat if you just moved when you win the same auctions
- Interactions: dynamic bidding up-and-down already modulates by predicted conversion; keep hourly multipliers modest so the layers do not compound into 2× swings
When Dayparting Is a Distraction
Skip or postpone dayparting when: the account is young or spend is low (data too thin — fix structure, keywords, and listings first); campaigns are budget-capped (fix exhaustion first — it dominates any bid schedule); TACoS problems are structural (wrong targets, weak listings — no schedule rescues a page that does not convert); or during launch phases where rank velocity matters more than hourly efficiency. Dayparting is a percentage-points optimisation on top of a healthy account — sequence it after structure, targeting, and budget health, and it reliably adds 5–15% efficiency; sequence it first and it mostly adds complexity.
Frequently Asked Questions
Does Amazon PPC support dayparting natively?
Yes, within limits: rule-based bid scheduling exists in campaign settings, and the Amazon Marketing Stream provides the hourly data that powers finer-grained third-party dayparting tools. Most serious implementations use a tool layered on Stream data.
What hours convert best on Amazon?
The most common consumer pattern peaks in the evening (roughly 7–10pm local) with a deep trough 1–5am — but the honest answer is "your hourly report knows and generic advice does not". B2B products, considered purchases, and international audiences all invert pieces of the default curve.
How much can dayparting realistically improve my ACOS?
On healthy accounts with real hourly contrast, 5–15% efficiency improvement is a realistic band. Accounts rescued from budget exhaustion during peak hours sometimes see more — though that is really a budget fix wearing a dayparting costume.
Should I pause campaigns overnight entirely?
Rarely. Pausing sacrifices auction presence, slows reattribution, and some overnight orders are same-session completions of evening research. Reduced bids (−20–40%) overnight capture most of the saving while keeping presence; full pausing is for extreme cases with genuinely worthless night traffic.
Do bid changes take effect immediately for dayparting?
There is operational lag in auction systems — schedule transitions are not knife-edge precise, which is another reason to prefer a few broad blocks over hour-by-hour micro-schedules. Build schedules that are robust to ±1 hour of slop.