Metric Definition
When demand concentrates
Track from
Peak hours analysis
Peak hours analysis is the practice of measuring how demand, such as orders, sessions, contacts, or transactions, distributes across the hours of the day so you can find the windows where volume concentrates. It turns a flat daily total into an hourly profile. That profile drives staffing, capacity, and pricing decisions that a single daily number cannot.
7 min read
What is peak hours analysis?
Peak hours analysis is the practice of breaking a daily volume total down by hour to find when demand concentrates and how sharp that concentration is. Instead of knowing that a shop served 1,200 customers in a day, you learn that 240 of them arrived between noon and one. The headline figure is the peak hour share, the proportion of a typical day that lands in its single busiest hour.
The value of the analysis comes from the gap between the average hour and the peak hour. If a venue is open for ten hours and serves 1,200 customers, the average hour carries 120. A peak hour carrying 240 is twice the average. Staffing, kitchen capacity, and queue length all have to cope with the 240, not the 120, so the average quietly understates what the operation actually needs.
Peak hours analysis applies anywhere demand arrives unevenly across the day. Restaurants and retail have lunch and evening rushes. Support desks spike after a product release goes out. Ecommerce checkout volume often climbs into the evening. Websites peak around commute times. In every case the same question applies: when does load concentrate, and how much.
A peak is only meaningful relative to a baseline. Always compare the busiest hour to the average hour for the same day, not to the quietest hour. The quietest hour flatters the comparison and makes every operation look like it has a severe spike. The average is the honest reference point.
How to calculate peak hours analysis
The core figure is the peak hour share: the volume in the busiest hour divided by the total daily volume. A bakery that takes 1,000 orders in a day with 180 of them between four and five in the afternoon has a peak hour share of 18 per cent. That single number tells you how lumpy the day is.
The share on its own is not enough to act on. You also need the peak to average ratio, which compares the busiest hour to the typical hour, and you need the analysis repeated by day of week, because a Saturday peak and a Tuesday peak rarely sit at the same time or the same height. Build the hourly profile first, then read these figures off it.
- 1
Hourly volume buckets
Group every demand event, an order, a session, a ticket, a transaction, into the hour it occurred. This is the raw hourly profile and everything else is derived from it.
- 2
Total daily volume
Sum the hourly buckets to get the full day. This is the denominator for the peak hour share and the basis for the average hour.
- 3
Busiest hour volume
Identify the single highest bucket. This is the load the operation has to be sized for, not the average.
- 4
Average hour volume
Divide total daily volume by the number of operating hours. Comparing the busiest hour to this average gives the peak to average ratio, the truest measure of how concentrated demand is.
Peak hours analysis in a metric tree
A peak hour share is a symptom, not a cause. The hour got busy because of upstream behaviour, scheduling, and capacity choices. A metric tree pulls those drivers apart so that when the peak shifts or sharpens, you can see which branch moved rather than guessing.
Metric tree insight
The tree separates two questions that operations teams constantly conflate: is the peak too high, or is the capacity too low. A 20 per cent peak hour share is not a problem if you can staff for it. It only becomes a problem on the capacity branch. KPI Tree lets you assign RACI ownership to each branch, so the demand smoothing levers sit with marketing while the staffing node sits with operations, and a moving peak pushes the alert to the owner who can actually act on it.
Peak hours analysis benchmarks
There is no universal good peak hour share, because it depends entirely on how concentrated demand naturally is in your context. A 24 hour ecommerce store spreads load across more hours than a lunch venue that does most of its trade in a two hour window. The ranges below give realistic peak hour shares for common operating models, measured against a typical operating day.
| Context | Flat profile | Typical peak share | Sharp peak |
|---|---|---|---|
| Quick-service food | 12 to 15 per cent | 18 to 25 per cent | 30+ per cent |
| Ecommerce checkout | 8 to 10 per cent | 11 to 15 per cent | 18+ per cent |
| Web and app traffic | 7 to 9 per cent | 10 to 14 per cent | 16+ per cent |
| Transactional services | 9 to 12 per cent | 13 to 18 per cent | 22+ per cent |
Benchmark the peak by day of week, never as a single blended figure. A Saturday lunch peak and a Wednesday lunch peak can differ by half. Blending them hides the day you most need to staff for and overstaffs the days you do not.
How to improve peak hours analysis
Improving peak hours analysis means two things: making the measurement sharper so the profile reflects reality, and using the profile to either match capacity to the peak or flatten the peak itself. The cards below cover both the measurement and the operational response.
Match staffing to the hourly profile
Build shift patterns from the hourly buckets rather than from a flat daily headcount. Bring people in for the window that carries 25 per cent of the day and release capacity in the quiet hours. The peak to average ratio tells you exactly how much extra cover the busy window needs.
Flatten the peak with incentives
If the peak strains capacity, shift some demand to adjacent hours. Off-peak pricing, happy hours, or scheduled pickup slots move willing customers out of the crush. A flatter profile lets the same capacity serve more demand without a queue.
Segment the profile by day and source
A single average curve hides the variation that matters. Split the profile by day of week, channel, and customer segment. The peak that comes from a scheduled email send behaves differently from the peak that comes from a commute, and each needs a different response.
Use consistent time zones and bucketing
A profile is only as honest as its timestamps. Normalise every event to a single operating time zone before bucketing, and decide whether a session or order counts at its start or its completion. Inconsistent bucketing smears the peak across two hours and hides where load truly sits.
Common mistakes when tracking peak hours analysis
- 1
Averaging away the peak
Reporting only the average hour hides the very thing the analysis exists to find. The average is a useful reference but it is never the figure you size capacity against. Always carry the busiest hour alongside it.
- 2
Blending days of the week
A combined weekly profile produces a peak that exists on no real day. Weekend and weekday demand peak at different times and heights, so a blended curve overstaffs some days and understaffs others. Keep the days separate.
- 3
Ignoring the capacity side
A peak hour share is only half the picture. A sharp peak is harmless if capacity can absorb it and a mild peak is painful if it cannot. Always read the peak against the capacity available in that window.
- 4
Treating a one-off spike as a pattern
A single campaign or outage can create an hour that never repeats. Build the profile from enough days to separate the recurring peak from the one-off event, and exclude known anomalies before you staff to them.
Related metrics
Ticket Volume
Customer Support MetricsMetric Definition
Ticket Volume = Total New Tickets Created in Period
Ticket volume is the total number of new support tickets created within a defined period. It is the fundamental demand metric for support operations, determining staffing requirements, budget allocation, and the urgency of self-service and product quality investments.
Average Order Value
Revenue per transaction
Operations Metrics
Metric Definition
AOV = Total Revenue / Number of Orders
Average order value measures the mean amount spent each time a customer places an order. It is a core e-commerce and retail metric that directly influences revenue, profitability, and customer acquisition efficiency.
Checkout Conversion Rate
E-commerce metric
Ecommerce & Marketplace MetricsMetric Definition
Checkout Conversion Rate = (Completed Purchases / Checkout Starts) x 100
Checkout conversion rate measures the percentage of users who begin the checkout process and successfully complete their purchase. It isolates the final stage of the buying funnel, from the moment a shopper initiates checkout to the order confirmation page. This metric is critical for e-commerce businesses because the checkout is where purchase intent is highest, and any friction at this stage directly destroys revenue that was nearly captured.
Cart Abandonment Rate
Checkout drop-off
Operations MetricsMetric Definition
Cart Abandonment Rate = (1 − Completed Purchases / Carts Created) × 100
Cart abandonment rate measures the percentage of online shopping carts that are created but not converted into completed purchases. It is one of the most impactful e-commerce metrics because it represents revenue that was within reach but lost at the final stage of the buying journey.
Why did my metric change?
Metric Definition
When demand concentration shifts, this diagnostic framework helps you trace why peak hours analysis moved and what drove the change.
Metric trees for operations teams
Metric Definition
Peak hours analysis sits at the heart of operations planning, and this guide shows how to place it within an operations metric tree alongside related capacity and demand metrics.
Turn the hourly profile into action with owners on every branch
Build a metric tree that decomposes the peak hour share into demand timing, demand drivers, and capacity. Give each branch a RACI owner so that when the peak shifts, the alert reaches the person who can adjust staffing or smooth the demand.
