Every quote is a guess
You price from memory because nobody records what the last job of that type cost to deliver. Without delivered cost by job type, each quote starts from the founder's instinct and the client's budget. The fix is a feedback loop: record the gap between quoted and delivered margin, review it by job type each quarter, and change the rate.
Written for firms between £500k and £15M selling projects, contracts or matters. A firm can quote accurately in aggregate and still lose money on two job types out of five.
Why the guess never improves
Estimating improves only when the estimator finds out what happened. In most founder-led firms the person who quotes never sees the delivered cost, the delivered cost is never calculated by job, and the job that went wrong is remembered as a difficult client rather than as a pricing error.
Nobody writes down the planned cost
You price on a day rate, a percentage or a feel for the job. With no planned cost recorded at the point of sale, there is nothing to compare the outcome against, so the comparison never happens.
Costs land in the month, not on the job
Your ledger groups spend by type: salaries, subcontractors, materials. Useful for accounts, useless for pricing. Nobody can answer what job 402 cost without reconstructing it by hand.
Overruns get explained, not recorded
The job ran over because the client changed their mind, the survey was wrong, the weather turned. Each explanation is true. Together they hide a pattern that would have changed the rate.
The estimator never sees the result
Whoever quotes moves on to the next opportunity. Without a quarterly review putting delivered margin in front of them, they repeat the estimate that felt right last time.
The loop that closes the gap
Four steps, run quarterly. Record planned cost when you quote. Capture delivered cost against the job. Compare the two by job type. Change the rate, the scope wording or the qualifying question that let the job in. Firms that run this twice usually find one job type carrying the rest.
Record the planned cost
Hours by grade and external spend, written on the quote itself. Your estimator already did this arithmetic to arrive at a price. Save it rather than discarding it.
Tag every cost to a job
Hours booked weekly, supplier costs coded to the job when committed. This is the step firms skip, and skipping it makes the other three impossible.
Compare by job type
Group jobs into five or six recognisable types. Average the gap between quoted and delivered margin within each. One type usually stands out by a distance.
Change one thing
Raise the rate for that type, tighten the scope wording, or add the qualifying question that keeps the worst version of it out. One change per quarter, then measure it.
Getting delivered cost per job is the same mechanism described in the weekly project margin view. Build it once and it answers both questions: how this job is going, and what the next one should cost.
Three pricing errors this exposes
| Error | How it shows up | What to change |
|---|---|---|
| The loss leader you forgot to stop | One job type shows a negative gap every quarter, defended as a route to better work that never arrives. | Put a date on it. Raise the rate or stop selling it. |
| Seniority priced at the wrong grade | Delivered hours show director time on work quoted at a mid-level rate. | Price the grade the work actually needs, or change who does it. |
| Scope that is generous in writing | Rounds of revision, site visits or meetings that are unlimited in the proposal and finite in your plan. | Put a number in the scope, and a price on the one after it. |
What changes when the loop runs
One agency installed weekly project margin, priced work beyond scope, and re-rated two job types over the following year.
Firms think they have a pricing problem and go looking for a number. They have a memory problem. Nobody wrote down what the last one cost.David Schofield
Common questions
How many jobs do I need before the data means anything?
Three of the same type gives you a signal worth acting on, and eight to ten gives you a rate you can defend. Start with the type you sell most often. Waiting for a full year of clean data across every job type means never changing a price.
Should I raise prices or cut cost?
Look at the gap by job type first. A type that is consistently underpriced needs the rate changed, since cutting cost on work that was mispriced at the outset moves the loss into quality. A type priced correctly but delivered badly needs the delivery fixed, and the two require different people to act.
My clients will not accept a rise. What then?
Change the scope rather than the headline rate: fewer included revisions, a defined number of visits, a price on the round after the last one. Clients who resist a rate increase accept a boundary, and a boundary protects the same margin without a negotiation about your worth.
Do I need estimating software?
No. A recorded planned cost on the quote, hours tagged to the job, and a quarterly comparison run in a spreadsheet will change your prices within two quarters. Reach for software once the manual version is working and the number of jobs makes it slow.
Related reading: Project margin that arrives too late · Busier than ever, and the bank balance disagrees · Financial Visibility · We won the work and can't staff it
Find the job type carrying the rest
Bring three jobs of the same kind and what you think they cost to deliver. One call is usually enough to tell whether the problem is the rate, the scope, or who did the work.
Book an Operational Clarity Call →