Consulting & AdvisoryFree
Process Observation Study and Time Log
Duration comes from the two days you spent on the floor, frequency from the client's own logs, and every pause is marked paid or unpaid.
River takes the durations out of your observation notes and the frequencies out of whatever the client already counts, then multiplies the two. Back comes the observed workflow written up step by step with the friction in place. Beside it sits an observation log where every row carries how long the step took, how often it happens in a year, which system said so, and whether the operator was being paid to stand there. The annual figure falls out of the sheet rather than out of an assumption.
Search the query and page one is time study forms. Columns for the element, the stopwatch reading, a performance rating and an allowance factor, all of it aimed at turning observed minutes into a standard time. That work is real, and it is half the problem. Nothing on those pages tells you what to do with the fact that your fourteen observed hours have to stand in for nearly nineteen thousand. That is the step where a number in a deck goes wrong by a factor of three.
Built for the consultant who spends two days on a client's floor and then defends a figure in front of a finance director who will ask where it came from. It reads next to the process register assembled from interviews and feeds a diagnostic that scores what it finds. Reach for it in a machine shop, in a claims department, or anywhere somebody is about to call a process inefficient and be asked what that is worth.
Why an observed minute is not yet a dollar
Two days on the floor tells you how long a step takes. It cannot tell you how often the step happens, and extrapolating frequency from the window you happened to watch is where observation studies come apart. The federal cost principles draw the same line. 2 CFR 200.430 accepts a rolling time study in place of time records, but only where the entire time period involved is covered by the sample and the result is statistically valid for the period being sampled.
The other half is whether a pause costs anything at all. Under 29 CFR 785.15 a factory worker waiting for machinery to be repaired is working, because the wait is unpredictable, short, and controlled by the employer. That operator is engaged to wait. Section 785.16 is the other side and it is strict. Time is not hours worked unless the employee was told in advance they could leave and need not restart until a stated hour.
Ridgeway Machine Works, a contract machine shop, eleven machinists at 1,720 direct hours each. Fourteen observed hours caught four tool crib trips averaging 11.5 minutes, which extrapolates to 5,406 a year. The MRP unplanned issue log recorded 2,455, so the observation ran 2.2 times high. At $58 fully burdened that is $27,291 rather than $60,097. Across three friction items the observation alone priced at $210,322. The logs and the pay classification bring it to $64,381.
How it works
Send the notes
Whatever you wrote on the floor, in whatever order it came out, plus the photographs.
Name the systems
Which of the client's systems already timestamps these events, even if nobody has exported it.
Read the log
One row per friction point, with its duration, its annual frequency, its source and its pay status.
Push back in chat
Ask why a row counts as paid time, or drop in the export the client finally sent.
What you get
- Every row carries the duration you observed and the frequency the client's system recorded
- The name of the log behind each frequency, so the number has an address
- Each pause marked as paid working time or as time the operator was released
- The basis for that classification written out, because a finance director will test it
- The naive extrapolation kept beside the corrected one, so the size of the correction shows
- Photographs and stopwatch readings tied to the step and the operator they came from
Common questions
How many hours of observation is enough?
Enough to get a stable duration for each step, which is usually a handful of clean cycles rather than a whole shift. Frequency is the part that needs a full year, and that never comes from watching. Keeping the two in separate columns means a thin observation count weakens one of them instead of quietly poisoning the annual figure.
What if the client has no system that counts these events?
Then the row says so, its frequency is a named person's estimate, and it is priced as a range and left out of the total. At Ridgeway the print revision hunt ran between $1,933 and $7,733 depending on whose guess you took. Half a day in the transaction data usually turns up a proxy: a stores issue, a scan, a status change, a reprint.
Why does it matter whether the operator was paid to wait?
Because it decides whether removing the wait saves money or only saves time. Paid idle time converts into capacity the client is already buying. Time where the operator was released until a stated hour is not on the payroll, so eliminating it shortens lead time and recovers no wage cost. Mixing the two is how a savings case gets taken apart.
The operations director says my frequency is wrong. What happens?
They open the log and you both read the same number, because the frequency came from their system and the row names it. That argument takes a minute and ends. Extrapolating from two days is the version that ends the meeting instead, which is the same reason the cited evidence log exists for paperwork.
Does the correction always shrink the number?
No, and that is the point. At Ridgeway the re-measure step showed up twice in fourteen hours, which extrapolates to 2,703 a year. The coordinate measuring machine had logged 3,910. That row moved up 45 percent, from $16,984 to $24,568. A method that only ever cuts your number is a haircut rather than a correction.
Where does this sit in the engagement?
In discovery, once you know which process the answer turns on. Interviews give you what people say the process is, and the theme register holds that. This gives you what the floor actually does and what the gap between them costs. Where a peer figure has to carry the case, the graded comparator sheet takes it on.
Process Observation Study and Time Log
Fill in the form and your workspace opens with the work already underway.