Capability template for operations and process improvement
Process map and bottleneck review
Map one process step by step: who does each step, how long work waits and how long it takes, where it is handed over and where it gets sent back. Then find the one step that caps how much gets finished, and what to try there first. The owner decides what changes.
Listed under
Add it to a workflow in the Fully Powered Planner, beside the teams and roles that will run it. Book a fit call
The outcome
Everyone has a theory about why the work is slow. This review replaces the theories with one page of figures. Every step of one process is laid out in order, with the role that does it, the hours an item sits waiting before it and the hours of real work it takes. Hand-overs between roles are counted, and every loop where work is sent back to an earlier step is on the map, not hidden.
From those figures the review names the one step that limits how much the process can finish each week: the role with more work arriving than its hours can absorb. It shows the arithmetic, checks it against what the team actually finishes, and says plainly whether the step everyone complains about is that limit or not.
Then it lists what to try first at that step, using the time and people you already have, before anyone talks about hiring or new software. It names the steps to leave alone for now, because making them faster would leave the bottleneck exactly where it is, and it names where the bottleneck is likely to move next. Fix the right step and the whole process moves. That is how a team does far more with the people it already has.
The review proposes. The process owner decides what changes, and nothing is changed until they do.
Who starts the work, and when
An operations lead or the owner of the process starts it, when people disagree about why a process takes so long, when work is piling up in one place, before adding people or buying software to speed something up, or before deciding which piece of work to improve first.
One run covers one process that repeats, such as turning a request into a quote, onboarding a new customer or approving a supplier bill. Run it again after a change has been tried, with fresh figures, because the bottleneck moves.
What they supply
- The process, in a few lines: its name, its owner, what starts one item and what finishes it, the one number that should rise (quotes sent a week, orders shipped a week) with its current value, how many items arrive a week, and any time target.
- A step table: each step in order, the role that does it, the typical hours an item waits before the step and the hours of work it takes, any step it sends work back to and how often, and whether each time was measured or estimated. A spreadsheet export is fine; the template names the column headings it expects.
- The hours each role gives the process in a week, everyone in that role together.
- Short notes from the people who do the work: when approvals happen, why items come back, what else pulls on their time.
Rough estimates will do to start. The review marks every estimate and says which ones matter most to firm up.
What comes out
One review, in a fixed order:
- The map: every step with its role, wait, work, hand-overs and send-backs, and which steps each role does.
- Where the time goes: the hours one item spends in the process, how much of that is work and how much is waiting, what send-backs add, and the longest waits.
- Load by role: the hours of work arriving at each role every week against the hours it gives, as a percentage, and how many items it can finish.
- The constraint, with its evidence, the check against what is finished, and a straight answer on the step people complain about.
- What to try first at the constraint, up to three options, each tied to a fact on the map.
- What comes after, if those are not enough, and the steps to leave alone for now.
- Where the bottleneck goes next, figures to firm up, decisions for the owner and open questions.
A good review lets someone who was not in the room see why that step is the bottleneck, from figures they can recompute.
An invented example
Illustrative example with invented data. The company is an invented 25-person maker of custom signs for shops and offices. Its people, process, figures and notes are all invented, and everyone is named by role. First come the inputs the operations manager supplied, in full. Then comes the review the capability should return from them, worked to the template's rules.
The inputs
1. The process request
Process: Quote for a custom sign, from request to quote sent
Owner: Operations manager
Starts when: a customer request arrives by email or the website form
Ends when: the quote is sent to the customer
The number that should rise: quotes sent per week. Now 17 a week.
Arriving: 20 requests a week
Figures cover: the four weeks to 30 October 2026 (weekly averages)
Hours in a working day: 8
Time target: quote sent within 3 working days (24 working hours)
The step people complain about: the production check (step 4)
Prepared: 10 November 2026, by the operations manager
2. The step table (exported from the job system, then completed by hand)
step,name,role,wait_before_hours,work_hours,sent_back_to,sent_back_per_10,basis,notes
1,Log the request in the job system,Sales coordinator,1,0.3,,,measured,
2,Check drawings and sizes are complete; ask the customer for anything missing,Sales coordinator,2,0.5,,,measured,
3,Price materials and labour,Estimator,12,1.5,2,2,measured,Missing sizes found during pricing go back to the coordinator
4,"Check it can be made: fixings, sizes and site access",Production manager,8,0.5,3,3,estimated,Changes to fixings or sizes go back for repricing
5,Approve the price,General manager,18,0.2,,,estimated,
6,Write up and send the quote,Sales coordinator,3,0.5,,,measured,
3. Role hours
role,hours_per_week
Sales coordinator,36
Estimator,38
Production manager,15
General manager,6
4. Notes from the people who do the work
Sales coordinator, 3 November: About 2 in 10 requests come back from
pricing because the estimator finds a size or a drawing missing. I chase
the customer, then the request goes back into the pricing queue.
Estimator, 3 November: I work 42 hours a week. About 4 of them go on
questions from the workshop about jobs we have already won, so about 38
are left for quotes. When the production manager changes the fixings or
sizes, I price the whole job again.
Production manager, 4 November: I do quote checks when I am off the
floor, about 15 hours a week. Roughly 3 in 10 need a change to fixings
or sizes. That is a guess; I do not track it.
General manager, 4 November: I approve quotes on Tuesday and Friday
afternoons.
The output
PROCESS MAP AND BOTTLENECK REVIEW: Quote for a custom sign, from request to quote sent
For: Operations manager. Prepared 10 November 2026 from the step table (6 steps), the role hours, notes from four roles. Figures cover the four weeks to 30 October 2026.
One item: starts when a customer request arrives by email or the website form; ends when the quote is sent to the customer.
The number that should rise: quotes sent per week, now 17 a week. Arriving: 20 a week. Time target: quote sent within 3 working days (24 working hours).
1. The map
Step 1. Log the request in the job system. Sales coordinator. Waits 1.0 h, then 0.3 h of work. Measured.
Step 2. Check drawings and sizes are complete; ask the customer for anything missing. Sales coordinator. Waits 2.0 h, then 0.5 h of work. Measured.
Step 3. Price materials and labour. Estimator. Waits 12.0 h, then 1.5 h of work. Hand-over from Sales coordinator. Sent back to step 2 in 2 of 10. Measured.
Step 4. Check it can be made: fixings, sizes and site access. Production manager. Waits 8.0 h, then 0.5 h of work. Hand-over from Estimator. Sent back to step 3 in 3 of 10. Estimated.
Step 5. Approve the price. General manager. Waits 18.0 h, then 0.2 h of work. Hand-over from Production manager. Estimated.
Step 6. Write up and send the quote. Sales coordinator. Waits 3.0 h, then 0.5 h of work. Hand-over from General manager. Measured.
Hand-overs: 4 (Sales coordinator to Estimator, Estimator to Production manager, Production manager to General manager, General manager to Sales coordinator). Send-backs: step 3 to step 2, 2 in 10, Estimator to Sales coordinator; step 4 to step 3, 3 in 10, Production manager to Estimator.
By role: Sales coordinator: steps 1, 2, 6. Estimator: step 3. Production manager: step 4. General manager: step 5.
2. Where the time goes
First pass: 47.5 h (5.9 working days): 3.5 h of work (7%), 44.0 h waiting (93%).
Send-backs add 9.8 h on average (step 3 to step 2: 0.2 x 16.0 h = 3.2 h; step 4 to step 3: 0.3 x 22.0 h = 6.6 h), 57.3 h (7.2 working days) in all.
Against the time target of 24.0 h: over by 23.5 h first pass and 33.3 h with send-backs.
Longest waits: before step 5 (18.0 h), before step 3 (12.0 h), before step 4 (8.0 h).
3. Load by role
Sales coordinator (steps 1, 2, 6): 28.0 h of work a week against 36.0 h given: 78%. Can finish about 25.7 a week.
Estimator (step 3): 45.0 h of work a week against 38.0 h given: 118%. Can finish about 16.9 a week.
Production manager (step 4): 13.0 h of work a week against 15.0 h given: 87%. Can finish about 23.1 a week.
General manager (step 5): 4.0 h of work a week against 6.0 h given: 67%. Can finish about 30.0 a week.
Send-backs counted: each one repeats the steps from where the item returns to where it was found, once. Expected work per item: step 2, 0.5 h x 1.20 = 0.60 h; step 3, 1.5 h x 1.50 = 2.25 h; step 4, 0.5 h x 1.30 = 0.65 h; all other steps as listed.
4. The constraint
Estimator, at step 3: 118%. More arrives than it can finish: 20 a week against about 16.9; the queue in front of step 3 grows by about 3.1 a week.
Check against what is finished: about 16.9 a week expected (the constraint's capacity, the smaller of the two) against 17 now: the figures agree.
The step people complain about: step 4 is not the constraint: the Production manager is at 87%, below the Estimator's 118%, so a faster check on its own would not raise quotes sent. Its send-backs do cost the Estimator time; see option 2.
The longest wait, before step 5, is not at the constraint: the General manager approves quotes on Tuesday and Friday afternoons, and is at 67%.
5. Try first at the constraint
1. Stop requests with missing sizes or drawings reaching pricing. Based on: step 3 sends 2 in 10 back to step 2, because the Estimator finds a size or a drawing missing. Returns up to about 6.0 h a week of Estimator time if the cause went away (20 x 0.2 x 1.5 h). Arithmetic on the supplied figures, not a forecast.
2. Stop the repricing caused by changes to fixings or sizes found at the make-it check (step 4). How is the owner's choice; the check itself stays. Based on: step 4 sends 3 in 10 back to step 3, and the Estimator prices the whole job again each time. Returns up to about 9.0 h a week of Estimator time if the cause went away (20 x 0.3 x 1.5 h). Arithmetic on the supplied figures, not a forecast.
3. Take the workshop's questions about jobs already won off the Estimator, to a role the owner chooses. Based on: the Estimator's note that about 4 of 42 hours go on those questions. Returns up to about 4.0 h a week of Estimator time if the cause went away (42.0 h - 38.0 h). Arithmetic on the supplied figures, not a forecast.
6. Then, if that is not enough
If the Estimator is still the most loaded role after these are tried and measured, adding capacity at step 3 is the owner's decision; adding it at any other step would not raise quotes sent per week.
7. Leave alone for now
Step 1 (Sales coordinator, 78%): faster logging would not raise quotes sent while the Estimator can finish only about 16.9 a week.
Step 2 (Sales coordinator, 78%): checking faster would not raise quotes sent. What it checks matters to option 1.
Step 4 (Production manager, 87%): a faster check would not raise quotes sent. What it finds matters to option 2.
Step 5 (General manager, 67%): the longest wait, but not a limit on quotes sent per week. It bears on the time target; see decision 2.
Step 6 (Sales coordinator, 78%): writing up faster would not raise quotes sent: at most about 16.9 a week reach it, and the Sales coordinator can finish about 25.7.
8. Where the bottleneck goes next
Production manager, at 87% today. Option 2 would also take the repeat checks off this role.
9. Figures to firm up
Step 4, wait 8.0 h, work 0.5 h and 3 in 10 sent back: estimated. The Production manager says the share is a guess; it sets the size of option 2 and the Production manager's load.
Step 5, wait 18.0 h: estimated. It is the longest wait and the basis of decision 2.
Estimator, 38.0 h given: estimated (role hours count as estimated unless said to be measured; the Estimator's note gives about 38). The 118% rests on it.
10. Decisions for the owner
1. Which of the three options to try first, who carries it, and when to run this review again with new figures.
2. Whether approval stays at two afternoons a week. The wait before step 5 is 18.0 h of the 44.0 h spent waiting on a first pass, and the time target is 24.0 h. It does not limit quotes sent per week.
11. Open questions
1. How long a customer usually takes to reply when a request goes back to step 2 for missing sizes, for the Sales coordinator. The notes give no reply time, so the review counts only the step's typical 2.0 h wait on the return trip.
Draft for the Operations manager. Nothing has been changed, assigned or sent. The owner decides what changes.
Where it fits
- Operations and process improvement, where slow processes are noticed, mapped and fixed.
- Quality management, because work sent back to an earlier step is usually the most expensive time on the map.
- Professional services, where proposals, onboarding and approvals queue behind one senior person.
- Manufacturing, for quoting, order entry and engineering checks around the line.
- Logistics and trade, where bookings and documents pass through several hands before anything moves.
- Construction, where estimates, submittals and change requests wait on a few busy people.
- Growth & scaling, when demand outruns the team and you need to know where the next hour of capacity counts.
- Established businesses, where processes grew by habit and nobody has timed them in years.
What has to change for your company
Choose the process and name its owner. Agree the one number the process exists to raise; most teams pick items finished a week. Decide how figures are gathered, whether from a ticket system, a job system or a week of people jotting times down, and use your own column headings if they differ. Set the load at which a role counts as nearly full; the template starts at 80 percent. Decide whether the review may propose moving work between roles or only changes within the constraint's own step. Agree where reviews are kept, so the next run starts from the last one. It runs in ChatGPT, Claude, Copilot or Gemini from the tables and notes you supply, with no connection to your systems.
What has been checked
Fully Powered AI wrote this capability template. It was editorially reviewed on 10 November 2026. It was tried once on its invented example, in Claude on 10 November 2026, and the record of that run is published with it. The figures and the finding matched the example, and the record lists the places where the instructions left the assistant a choice. Small clarifications made afterward, to settle those places, have not been run again; the record lists them. One run on an invented example is not a test. It has not yet been tested in a client environment. It has not yet been tested as a packaged version in ChatGPT, Claude, Copilot or Gemini.
Testing in your environment means running it on a process whose bottleneck you already know, and on one you changed in the past where you know what happened next, then checking every figure by hand. Hard cases belong in the test set: a step everyone blames that is not the limit, a long wait caused by an approval day rather than by lack of time, and estimates that disagree with what is actually finished. Pilot acceptance-test preparation builds that test set. When the review points at a step worth handing to AI, Workflow fit check says whether it is ready, and Write an SOP from how the work is done writes down the changed step once the owner approves it.
The full capability template
Everything above is the outline. The complete capability template is the working version: the roles, the inputs, the procedure, the exceptions and stop conditions, the output format, the reviewer checklist, the acceptance tests, the instructions for an AI assistant and what to customize for your company. It comes with a download and the link to its folder in Fully Powered’s public repository.
Free with an account. Sign in by email: no password and no card.
Next step
Put this capability to work for your team: on your plan, or built with you in a sprint.
Build this in a sprint
In a sprint, Fully Powered adapts this capability template to your rules, proves it on your own work and puts it in the hands of the roles that own it.
Two ways to begin: