The stream ran four hours and brought in eleven thousand. Two of those hours produced nine thousand of it. The other two produced two thousand between them. Looking at one number for the night, the seller booked the same host for the same four hours the following week. Four hours paid for, two hours earned. No TikTok live analytics view had ever split it by the hour.
This is what a nightly total does. It converts a scheduling question into a single figure that looks like a verdict. TikTok live analytics at hourly resolution turns it back into a question with an answer. Where did the money actually land? Was the room full or empty? How long does the next slot need to be?
This is written for the seller running live two to five nights a week across one or two markets. The host takes a session fee rather than a share of sales. At that size, the gap between a two-hour slot and a four-hour slot is not a rounding error. It is the whole margin on the night.
A nightly total is a verdict; an hourly view is a question
Every live produces a curve. Viewers arrive in waves that follow the host’s notifications and the platform’s own distribution. The buying concentrates where a wave meets a product the audience already wanted.
The consequence is that teams repeat the same slot whether it worked or not. Nothing in the nightly total said which part of it worked. Four hours is the default, so four hours is what gets booked. The last hour, usually the thinnest, is where the host’s fee is worst spent.
The hourly view does not make the decision for you. It tells you where the money was. That is the only input the decision needs.
What the hourly view actually shows
Four figures per hour are enough. Anything more becomes a dashboard nobody opens twice.
| Per hour | What it tells you | The decision it changes |
|---|---|---|
| GMV | Where the buying actually happened | Which hours to keep, which to cut |
| Viewers at the top of the hour | Whether the room was full or empty | Whether the hour failed or was never given a chance |
| Units per minute | The pace inside the hour, not just the total | Whether to repeat the product sequence or change it |
| Product switched in that hour | Which item the peak belongs to | What to lead with next time |
DAMI shows those four by hour rather than as a nightly average. The average is exactly what hides the decision.
The second row is the one that stops a wrong conclusion. An hour with low GMV and a full room is a content problem. The product or the pitch did not land. An hour with low GMV and an empty room is a distribution problem. Cutting the host for it punishes the wrong person.
The hour you should not book again
Nearly every live has a dead hour. It is usually a predictable one. The hour after the host’s energy drops is one. Another is the gap between two product blocks, when the room empties. A third is the hour nobody knew about, because the announcement went out after it had started.
Three options exist, and only one of them is usually taken. The obvious one is to shorten the slot. A better one is often to move the block. Hour three may be dead because the audience is commuting. That block may work at a different time rather than not at all. The third option pays best over a quarter. Stop treating the dead hour as a hosting cost. Give it to a different host instead.
None of these need a longer stream. Duration is not a lever on revenue once the audience curve has flattened. It just adds hours at the thin end.
Duration is a scheduling decision, not an endurance test
Duration stops being a matter of taste once the hourly split exists. TikTok live analytics at hourly resolution turns it from a preference into a decision. Hosts are usually paid by the session. Some are paid by the hour, occasionally by a share. Whichever it is, the last hour costs the same as the first. The revenue is not the same. A four-hour slot that produces two good hours is a two-hour slot with a surcharge attached.
The workable pattern is to decide duration from the curve rather than the calendar. Run the full block once. Read the hourly split. Then cut to the hours that produced at least 20 percent of the night’s GMV. DAMI keeps that curve per session, so the cut is arithmetic rather than a guess. Most sellers land on two and a half to three hours. That is also close to where host attention holds.
The same logic applies to the host. A host with a peak in hour one and a flat hour three is not a bad host. She has a short, high-intensity format. Book her for four hours and you will judge her on the wrong number.
Where this belongs in the wider read
The hourly live view is one of four reports, and it is the one teams skip most often. Live looks like an event rather than a dataset. The four reports and the question each one answers puts it in order. The seven-stage map is the place to start. It shows which part of the workflow is actually leaking.
One honest limit. Hourly attribution is approximate when the same product sells through video at the same time. A single night is also a sample of one. Read the curve over three or four sessions before you restructure a schedule around it.
DAMI keeps the live breakdown at hourly resolution next to the video and product views. The hour that earned and the video that fed it read in one pass. No more piecing them together from three screens. DAMI also keeps the host, the slot and the product sequence attached to each session. That is what makes the comparison possible at all. The DAMI analytics workspace is where those three sit together.
FAQ
How many hours of live does a small shop actually need per week?
Two sessions of two and a half hours beat one session of five for almost every shop under fifty SKUs. It doubles the number of audience waves without doubling the thin hours. What matters is that the slot is the same each week. The audience then knows when to arrive.
What if every hour looks equally bad?
Then the problem is upstream of scheduling. Most often the product is not one people buy in a live format. Or the room is never full enough to test anything. Look at viewers at the top of each hour before changing duration. A flat curve in an empty room tells you nothing about the host.
Should you pay the host per hour or per session?
Per session prices the risk into the host. Per hour prices it into you. Most shops at this size are better off with a session fee plus a small share above a threshold. It keeps the cost tied to the same number the report is measuring. Creating an account gives you a shop small enough to read the curve within a month. Read TikTok live analytics hourly from the first session and the second one is cheaper to schedule.
