If you can’t see capacity, delivery time and profit per job in one place, you’re managing by gut feel. That’s how teams get overloaded, jobs slip and margin quietly disappears. For a deeper systems view, cross-reference Business Operations: The Complete Systems Playbook for SMEs, then use this article to build a lean KPI set you’ll actually run weekly.
In this article, we’re going to discuss how to:
- Choose the few KPIs that tell you what’s really happening in delivery
- Calculate capacity, delivery times, profit per job and utilisation without a data project
- Set guardrails and a weekly rhythm so the numbers drive action not debate
Operations KPIs: A Practical Definition You Can Run Weekly
Operations KPIs are a small set of numbers that tell you whether your delivery engine can fulfil demand at the promised speed and quality, while protecting margin and team health. If a metric doesn’t change a decision this week, it’s not an operations KPI, it’s trivia.
Quick sense-checks for good operations kpis:
- Owner-actionable: You can change it with a policy, schedule or pricing decision.
- Lead and lag mix: You can spot a problem before it hits revenue.
- Measured from your own artefacts: Job tickets, timesheets, calendar bookings, invoicing, support queue.
- Comparable week to week: Same definition, same cut-off time, same owners.
Get Your Baseline Data In 2 Hours (Internal First, Public Second)
You don’t need a BI stack to start. You need consistent definitions and a fast baseline you trust enough to make a call. Block 2 hours, open your last 4 weeks of work and pull the numbers by hand if you must.
Internal Signals To Gather Before You Touch The Market
Pull these from the tools you already use. The point is speed, not perfection.
- Jobs delivered: Count of completed jobs/projects this week, plus any still open.
- Start and finish dates: From your project board, ticket system or calendar.
- Hours consumed: Timesheets if you have them, otherwise a quick estimate via calendar blocks.
- Revenue invoiced: Per job, not just total weekly revenue.
- Direct costs: Contractors, materials, transaction fees, mileage, hosting, anything that scales with delivery.
- Rework events: Number of times a job came back due to an internal miss.
If you’re missing items, don’t stop. Pick a ‘best available’ source and write it down so you can improve later.
Public Signals That Help You Sanity-Check
Only once you’ve pulled internal data, look outside. You’re not trying to copy competitors, you’re trying to sense-check expectations.
- Quoted lead times: What others claim for delivery and onboarding.
- Review patterns: Complaints about delays, communication, quality, handovers.
- Job ads: What roles competitors are hiring to fix bottlenecks.
Use public signals to inform your promises and positioning, not to set your internal targets blindly.
The Four Numbers That Stop Most Operational Fires
If your purpose is capacity, delivery times, profit per job and utilisation, you can run a tight operating system around four core metrics. Everything else can be supporting detail.
1) Capacity: How Much Work You Can Deliver Without Breaking People
Capacity is the amount of deliverable work your team can complete in a given period at your required quality level. Treat it as a constraint, not a wish.
A simple weekly capacity calculation for service businesses:
Weekly capacity (hours) = (Number of delivery people) × (Available hours per person) × (Focus factor)
Example:
- 6 delivery staff
- 32 available hours each (after meetings, admin and sales support)
- Focus factor 0.85 (to allow for context switching and small disruptions)
Weekly capacity = 6 × 32 × 0.85 = 163.2 hours
Completion check: if you regularly book 190 to 220 hours of work against a 163-hour capacity, the KPI isn’t the problem. Your planning is.
2) Delivery Time: How Long Work Really Takes End To End
Delivery time should be measured end to end, from when the job is accepted into delivery to when it is signed off and ready to invoice. Clients don’t care about your internal ‘work started’ date, they care about finished.
Two views you can track without drama:
- Cycle time: Median days from start to finish for completed jobs this week.
- On-time delivery: % of jobs delivered by the promised date.
Use the median, not the average, because one messy project will skew the mean and you’ll end up arguing about outliers instead of fixing flow.
3) Profit Per Job: The KPI That Pays Your Wages
Profit per job is where founders either get honest or go broke slowly. Revenue is vanity if delivery consumes more time and cost than you priced for.
Start with contribution margin per job:
Profit per job = Invoiced revenue − Direct costs − Delivery labour cost
If you don’t track labour cost per job, estimate it. Pick a blended hourly cost for your delivery team (salary + NI + pensions + software + overhead allocation) and get moving.
Example quick calc:
- Job fee: £3,000
- Direct contractor cost: £450
- Hours delivered: 22
- Blended hourly cost: £45
Profit per job = £3,000 − £450 − (22 × £45) = £1,560
Completion check: if you can’t explain why the bottom 20% of jobs are low margin, you’re flying blind. That bottom 20% is usually where the operational fixes live.
4) Utilisation: Are You Selling The Time You’re Paying For?
Utilisation is about how much of your paid capacity is spent on billable or value-creating delivery. It’s not about squeezing people, it’s about spotting structural waste: too many meetings, unclear briefs, rework, waiting on approvals.
A practical utilisation formula:
Utilisation % = (Delivery hours on client work ÷ Total available delivery hours) × 100
Good utilisation varies by model. A small agency with heavy client communication might be healthy at 70% to 80%. A productised service with tight SOPs might run 80% to 88%. If you’re at 55% and still overloaded, your work intake and planning are chaotic, or your time tracking is fiction.
How To Turn Operations KPIs Into A Weekly Operating Rhythm
Numbers don’t improve because they exist. They improve when you attach an owner, a review cadence and a specific action rule. This is how you keep operations kpis useful instead of performative.
Set A Single Weekly Cut-Off And Keep It Boring
Pick a time, for example Monday 09:00. Pull the same data each week. Don’t ‘adjust’ the method mid-quarter unless you document the change. Consistency beats elegance.
Use A 30-Minute Ops Review With A Fixed Agenda
Keep it short so it actually happens. Agenda:
- 5 mins: Capacity vs booked work, identify immediate overloads.
- 10 mins: Delivery time and on-time %, name the blockers and decide removals.
- 10 mins: Profit per job, scan low-margin jobs and decide changes.
- 5 mins: Commitments, who does what by when.
Completion check: if your ops meeting ends with ‘we’ll keep an eye on it’, you’ve wasted everyone’s time. End with decisions.
Make Action Rules That Trigger Without A Debate
Action rules protect you from mood-based management. Example rules that work in live ops:
- If utilisation exceeds 88% for 2 weeks: Pause new starts, raise price on the next 5 proposals, or add a contractor buffer.
- If on-time delivery drops below 90%: Reduce WIP by 20% and tighten acceptance criteria for new work.
- If median cycle time rises by 25%: Run a bottleneck review, identify the stage with longest wait and fix that stage first.
- If profit per job falls below £X: Trigger a scope reset or change the delivery method before repeating the offer.
A One-Sentence Offer Template That Fits The KPIs
If your offer is vague, your operations will be vague. Here’s a one-sentence template that forces clarity and links to capacity, delivery time and margin:
We help [specific customer] get [measurable outcome] in [timeframe] for £[price], using [method], with [clear boundary] included and [clear boundary] excluded.
Example for a UK-based B2B service:
We help SaaS founders reduce churn by fixing onboarding emails in 10 working days for £2,500, using a tested 7-email sequence, with copy and setup included and design work excluded.
This is operationally useful because you can now measure cycle time against 10 working days, capacity against how many sequences you can build weekly and profit per job against a stable scope.
Validate In 7 To 14 Days With Small Tests, Not Big Bets
Founders waste months ‘building the perfect service’ when they could validate the operations in days. You’re testing whether the work fits your delivery system, not whether you can do the work at all.
A simple validation path:
- Day 1: Write the offer sentence, define the acceptance criteria for a job and define ‘done’.
- Days 2 to 4: Sell 3 small jobs with identical scope, same delivery team, same tools.
- Days 5 to 10: Deliver, track hours and blockers, capture rework events.
- Days 11 to 14: Review profit per job and cycle time, then decide: raise price, tighten scope, improve SOPs or stop.
Completion check: you should be able to answer, ‘How many jobs like this can we deliver next week without lowering quality?’ If you can’t, your process is not defined yet.
Pricing And Unit Economics That Hold At Small Scale
At small scale, the enemy is underpricing mixed with inconsistent delivery. Your pricing needs to survive when you’ve got 10 to 50 customers, not when you’ve got 500 and a finance team.
Start With A Minimum Viable Margin Rule
Pick a rule that protects you from yourself. For example:
Every job must deliver at least £800 contribution or at least 40% contribution margin, whichever is higher.
This stops you filling your calendar with low-margin work that looks busy but starves the business.
Use A Two-Number Job Estimate Before You Quote
Don’t build a spreadsheet. Ask your team for two numbers:
- Expected hours: If everything goes normally.
- Risk buffer hours: If the brief is messy or approvals are slow.
Then price off the higher number until you’ve earned the right to be more aggressive. Early on, buffer is a strategy, not a mistake.
Watch The Blend: Price Per Job Versus Hours Per Job
Even if you price per project, your delivery engine runs on hours. Track:
Effective hourly rate = Job revenue ÷ Delivery hours
If your blended cost is £45 per hour and your effective rate is £70, your gross delivery margin is fine. If it’s £52, you’ll feel ‘busy’ and still not have cash.
Operational Guardrails That Protect Margin And Time
Guardrails are the rules that stop your team doing heroic work to cover poor process. They protect delivery time, utilisation and profit per job.
Guardrail 1: Control Work In Progress (WIP)
WIP is how many active jobs you’ve got open at once. Too much WIP increases cycle time because everyone context-switches and waits on each other.
Set a simple limit: no more than 2 active jobs per delivery person unless a director approves. You’ll be shocked how quickly delivery times improve once you stop starting everything.
Guardrail 2: Tight Intake Criteria
Define what must be true before work starts. Example intake requirements:
- Single owner on the client side: One person who can approve.
- Assets received: Logins, content, brand guidelines, access.
- Definition of done: What sign-off looks like.
No intake pack, no start. This one policy reduces rework and improves on-time delivery without adding headcount.
Guardrail 3: Make Rework Visible
Track rework as a count and as hours. Rework is usually not a ‘people problem’, it’s a briefing, scope or QA problem. Once rework is visible, profit per job stops leaking quietly.
Mini Examples: What Good Looks Like In Real Teams
These are short, real-world patterns you can copy this week.
Example 1: Logistics SME in Manchester
They were quoting 5-day delivery but averaging 9 days. A WIP limit and a stricter intake checklist cut median cycle time from 9 to 6 days in 3 weeks, with on-time delivery moving from 72% to 91%.
Example 2: Boutique finance consultancy in London
They thought utilisation was the issue, it wasn’t. Their utilisation was 76% but profit per job was inconsistent because senior staff kept doing ‘quick fixes’. They introduced a scope boundary and a paid ‘rapid response’ add-on at £350, increasing contribution per job by £420 on average.
Example 3: Productised web build service in Bristol
They raised price 12% and removed two custom steps that caused most rework. Utilisation stayed roughly flat, but median delivery time fell from 18 to 13 working days because handoffs became cleaner.
Common Risks And Simple Hedges
Most KPI programmes fail for predictable reasons. Here are the mistakes and the fixes.
- Risk: Measuring too much and acting on nothing.
Hedge: Start with the four core metrics, add only when a decision demands it. - Risk: Using KPIs to blame people.
Hedge: Focus on system fixes first: intake, WIP, SOPs, approvals, QA. - Risk: Incentivising the wrong behaviour, like hitting utilisation at the expense of quality.
Hedge: Pair utilisation with on-time delivery and rework rate so you don’t create a sweatshop. - Risk: Trusting bad data and arguing in meetings.
Hedge: Write definitions in plain English and use a single weekly cut-off.
Do And Don’t Checklist For Tracking Operations KPIs
- Do: Use median cycle time and on-time delivery % as your delivery signal.
- Do: Track profit per job on the bottom 20% first, that’s where fixes pay back.
- Do: Set action rules so your team knows what happens when a KPI moves.
- Don’t: Let ‘urgent’ work bypass intake, it will wreck delivery time and margin.
- Don’t: Chase 95% utilisation if it increases rework and churn, you’ll pay for it later.
- Don’t: Change KPI definitions mid-stream, you’ll lose trend visibility.
Download The Operations Dashboard Template And Run Tighter Weekly Reviews
If you want to put these numbers into a simple weekly rhythm, download the Operations Dashboard Template (KPIs, Tasks, Delivery Status) and use it for the next 4 Mondays. You’ll spot capacity pinch points, delivery slippage and margin leaks fast, then you can build the SOPs and systems to fix them.
Key Takeaways
- Pick a small set of operations kpis that drive decisions this week: capacity, delivery time, profit per job and utilisation.
- Validate your delivery model in 7 to 14 days by selling and delivering 3 identical jobs, then reviewing cycle time, rework and contribution margin.
- Protect margin and time with guardrails: WIP limits, strict intake criteria and visible rework tracking.
FAQ For Operations KPIs
What are the most important operations KPIs for a service business?
Capacity, delivery time (cycle time and on-time delivery), profit per job and utilisation will cover most of what matters. Add rework rate if quality issues are common.
How do I calculate utilisation if we don’t track time properly?
Start with a 2-week manual sample using calendars and task logs, then estimate delivery hours versus available hours. The goal is directional accuracy so you can spot waste and overload, not perfect forensic reporting.
What’s a healthy utilisation % for a small team?
Many service teams run well at 70% to 85% depending on client communication and complexity. If you’re consistently above 88%, expect delays and quality slips unless your processes are extremely tight.
How do I improve delivery time without hiring?
Reduce work in progress, tighten intake so you start jobs cleanly and remove blockers faster through a short weekly ops review. Most cycle time is waiting and switching, not doing.
How often should we review operations KPIs?
Weekly is the sweet spot for most SMEs because it’s frequent enough to correct course but not so frequent you chase noise. Keep the meeting to 30 minutes and focus on decisions.
What if profit per job varies wildly between projects?
That usually means scope is inconsistent or the delivery method changes job to job. Standardise the offer, define boundaries and introduce paid add-ons for ‘extras’ instead of letting them leak into delivery hours.
Should operations KPIs be shared with the whole team?
Share the ones people can influence and pair them with the ‘why’ and the action rules. If you share utilisation, also share quality signals like rework or on-time delivery so it doesn’t become a blunt pressure tool.
How do I stop KPIs becoming a vanity dashboard?
Attach an owner and a trigger action to each metric, then review the decisions you made last week. If a KPI never changes a decision, remove it.
