Two projects can have the same deliverable and very different economics. Two consulting jobs may each produce a report, but one needs extensive negotiation with several stakeholders. Two fabrication jobs may produce similar items, but one needs extensive inspection, scarce specialists and careful rework control. Two software rollouts may cover the same number of users, but differ sharply in data quality, integration and the effort needed to get people to use the new system.
When a business pushes such work through one standard estimating template, the estimates look consistent while the underlying costs are not. Some jobs are quietly overpriced and lost. Others are underpriced and won, and the business only discovers the loss when the job is finished. The problem is especially visible in service and project businesses, where the product is created through people’s effort and judgement rather than a stable list of parts.
This article explains how to estimate from cost drivers, the factors that actually cause cost to arise in a particular job. It covers why templates and rate cards mislead, which groups of drivers matter most, how to separate uncertainty from margin, when to charge for a discovery phase and how to learn from every finished job.
What the research shows
A 2018 master’s thesis by Azadi Fırat Kaya examined project costing in a company providing translation and localisation services. The evidence comes from a single case, so it should not be generalised too far, but the findings illustrate a common problem. Participants identified the rates of the specialists used, the terms agreed with each client and the number of process steps required as major cost factors. A job needing translation only had a very different cost structure from one needing translation, editing, quality assurance and proofreading. Clients’ quality expectations also influenced which specialists were chosen, and therefore the cost.
The general lesson applies far beyond translation: cost follows the work system that produces the outcome, not the label on the deliverable.
Common misreadings
- A rate card is a cost model. Rates are inputs. They say nothing about how many hours, activities, specialists or iterations a job will need.
- Average past cost predicts the next job. Averages work when the process is stable. They mislead when complexity, customer requirements or the mix of resources varies.
- The customer’s target price sets the estimate. Commercial pressure may justify redesigning how the work is delivered, but lowering an estimate because the customer will not pay more does not lower the actual cost.
- Only direct labour matters. Coordination, review, rework, data preparation, change handling and project management can be large costs even when they are less visible.
- Every surprise is an estimating failure. Sometimes the business has taken on work for which it does not yet have a stable way of delivering. The estimate should then show the uncertainty rather than hide it.
From requirement to margin
A cost-driver estimate connects each step:
customer requirement → activities needed → resources needed → timing → cost → margin
That makes the estimate explainable. If cost rises, the business can see whether the cause is greater volume, more complexity, scarce specialists, extra quality checks, a tighter schedule, more interfaces, uncertainty or changed requirements. That is far more useful than knowing a job is 15% over budget.
Seven groups of cost drivers
| Driver group | Questions to ask |
|---|---|
| Demand | How much work is actually required? |
| Complexity | How many variants, interfaces, exceptions or unknowns are there? |
| Quality | What checks, reviews, testing and acceptance steps are required? |
| Resources | Which skills, tools, suppliers and facilities are needed? |
| Process | How many steps, handoffs and iterations create the deliverable? |
| Timing | Does the schedule force overtime, premium resources or work in parallel? |
| Commercial conditions | What approvals, changes, warranties, payment terms or customer-specific obligations add cost? |
Not every group matters for every job. The point is to ask the questions deliberately rather than assume the answers from a template.
Customer requirements create process requirements
What a customer expects shapes how the work must be done. A customer requiring independent verification creates review work. A government client requiring detailed records creates documentation and audit effort. A customer with strict brand standards creates proofing rounds. These are not sales details. They are cost drivers, and the estimate should trace cost back to the promises made in the proposal.
Process count can matter more than volume
Volume is easy to measure. Process complexity is harder. A job with 1,000 units and one stable workflow may cost less than one with 500 units requiring several reviews, exceptions and handoffs. Useful drivers to count include design iterations, review rounds, approval authorities, stakeholder groups, test cycles, product variants, sites and expected change requests.
Cheaper resources are not always cheaper
Scarce expertise is expensive, but cheaper resources can increase hours, supervision, rework and schedule risk. A less experienced person may take twice as long. A senior specialist may cost more per hour but solve the problem with fewer iterations. The relevant measure is not the hourly rate but the cost of achieving the required outcome at the required quality.
This is the thinking behind activity-based costing: tracing costs to the activities that create them and the factors that drive those activities, rather than spreading costs across jobs by broad averages.
Separate uncertainty from margin
A good estimate distinguishes five layers:
- Base work: what is known and required.
- Known additions: specific extra requirements for this customer or job.
- Uncertain work: activities that may or may not be needed.
- Contingency: an allowance for uncertain work and risks, sized to the uncertainty.
- Margin: the profit the business intends to earn.
Blending contingency into margin is dangerous. It makes it impossible to tell whether a job was profitable or simply consumed hidden contingency, and it encourages cutting “margin” in negotiation when the money was really covering risk.
Charge for discovery when uncertainty is high
For non-standard work, a credible estimate may not be possible until the work is better understood. A short, paid discovery phase, in which the business investigates the requirements, site, data or design before committing to a fixed price, can protect both parties. The customer gets a more reliable price and plan. The business avoids committing to a number it cannot defend. If a fixed price is required anyway, the uncertainty should appear as an explicit contingency rather than being absorbed silently.
Review estimates before they go out
A second pair of eyes catches many estimating errors. Before a significant quote is submitted, ask someone other than the estimator to review it against a short checklist: which drivers make this job different from a typical one, which assumptions does the estimate rely on, what is excluded, how large is the contingency and why, and does the margin sit separately from the contingency? Writing the assumptions and exclusions on the quote itself also protects the business later, because changes in scope can be identified and priced as variations rather than absorbed.
Explain the drivers to customers
Customers often compare quotes on the bottom line alone. A cost-driver estimate gives the business a better conversation. Instead of defending a higher price, it can show what creates the cost, such as multiple review rounds, after-hours work or several sites, and offer options that change those drivers: fewer proof rounds, installation in normal hours, staged delivery or a simpler finish. Many customers appreciate understanding where their money goes, and some will choose a cheaper option that still meets their needs. Others will accept the higher price because they can see why it is justified.
Look beyond project cost
Where a project creates something that will be used for years, such as equipment, a building or a software system, low project cost can create high operating cost. A cheaper machine may need more maintenance. A quick software integration may create years of manual workarounds. Where it matters to the customer, show the whole-life consequences of options alongside the project price.
Learn from every finished job
The most valuable estimating improvement is a habit of comparing estimate and actual cost after each significant job, and identifying what drove the difference: rates, volume, rework, quality checks, changes, delays, interfaces or poor assumptions. Over time, build a cost-driver library: for each type of job, the small number of variables that explain most of the cost variation, with typical values. New estimates then start from evidence rather than habit.
A worked example
This is an illustration. A commercial signage business is quoting two jobs, each for 20 illuminated signs. Its template prices signs at $1,800 each, so both quotes would be $36,000.
The estimator builds a cost-driver estimate for each, using a fabrication cost of $1,300 per sign and an internal rate of $95 an hour for design and project management.
Job A repeats an existing design for a single site, with the customer’s artwork ready:
- Fabrication: 20 × $1,300 = $26,000.
- Artwork adjustment: 4 hours, $380.
- Installation at one site: $2,400.
- Project management: 6 hours, $570.
- Estimated cost: $29,350. At the template price, the margin is about $6,650, around 18% of the price.
Job B needs a new design from a concept sketch, three rounds of proofs, council permit coordination, and after-hours installation across five sites:
- Fabrication: $26,000.
- Design from concept: 24 hours, $2,280.
- Three proof rounds: 9 hours, $855.
- Permit coordination: 10 hours, $950.
- Installation: five sites at $1,200 plus $300 travel each, $7,500.
- After-hours premium of 50% on installation labour: $3,000.
- Project management: 20 hours, $1,900.
- Estimated cost: $42,485, plus a separate contingency of about $2,000 in case the council requires design changes.
At the template price of $36,000, Job B would lose about $6,500 before any margin. With the cost-driver estimate, the business prices Job B to cover its real cost, contingency and margin, and explains the difference to the customer by pointing to the design, permit, multi-site and after-hours requirements. It also offers a cheaper option with installation during normal hours.
How this applies to a small Australian business
Small project and service businesses often estimate from experience and a few rates, which works for repeat work and fails on unusual jobs. Practical steps:
- List the cost drivers for your main types of job.
- Ask the seven driver questions for any job that differs from the norm.
- Trace customer requirements to the activities they create.
- Show contingency separately from margin.
- Offer paid discovery for uncertain or complex work.
- Compare estimate and actual on every significant job.
- Build a simple cost-driver library from completed jobs.
- Check GST and other tax treatment of your pricing with your accountant.
The articles on pricing your product or service and should-cost modelling cover related methods.
Signals worth watching
- Wide, unexplained cost variation between jobs of the same type.
- Sales routinely overriding estimating assumptions to hit a target price.
- Lower-cost resources associated with more rework or supervision.
- Quality activities added after the job is won.
- Contingency used up early in jobs.
- Non-standard work priced using standard productivity rates.
- Jobs that look profitable until rework and support costs are counted.
Common mistakes
- Quoting from a rate card alone.
- Using averages for jobs that differ in complexity.
- Letting the target price set the estimate.
- Ignoring coordination, review and rework.
- Blending contingency into margin.
- Never comparing estimates with actuals.
Frequently asked questions
Will cost-driver estimating make us more expensive than competitors? Sometimes on complex jobs, and that is often correct, because those jobs cost more to deliver. On simple jobs it may make you more competitive. Either way, you will know where you make money.
How detailed should estimates be? Detailed enough to capture the drivers that vary between jobs. Repeat work can use simple unit rates. Unusual work needs more thought.
What if the customer will not pay for discovery? Offer a smaller discovery step, price the uncertainty as an explicit contingency or consider whether the job is right for your business.
Who should do the estimating? Someone who understands how the work is actually done, supported by records of past jobs. Where sales staff prepare quotes, involve the people who will deliver the work in checking the drivers and assumptions for anything non-standard.
How long until a cost-driver library is useful? A handful of well-reviewed jobs of each type often reveals the main drivers. Keep adding to it over time.
What if our accounting system cannot report costs by driver? Start outside it. Ask the people running jobs to record a few simple facts on each job sheet, such as hours on site, number of revisions, travel and waiting time, and review them in a spreadsheet each month. A rough record kept consistently for a year is usually enough to show which drivers matter, and it tells you what the accounting system should capture later.
Questions to ask
- Which three variables explain most of the cost variation in our jobs?
- Are we pricing customer-specific complexity or absorbing it?
- Which activities create cost but are invisible in our current estimates?
- Where does cheaper labour increase total cost through rework or delay?
- Which types of job need a discovery phase before a credible price?
- Do we compare estimate and actual on every significant job?
Bringing it together
A good estimate is not a number produced by a template. It is a model of how the work will create cost. Trace customer requirements to activities, resources and timing, ask the driver questions deliberately, separate base work, known additions, uncertainty, contingency and margin, charge for discovery when uncertainty is high and learn from every finished job. Standardise the questions and the discipline, and let each job’s economics reflect its real drivers.
Source: KEVOS notes, drawing on a 2018 master’s thesis by Azadi Fırat Kaya on project costing in a localisation services company. Examples and figures in this article are illustrations.
