Project estimation methods use scope and evidence to estimate effort, cost or duration. Analogy adjusts comparable work; parametric models use calibrated drivers; three-point methods model uncertainty; Delphi structures expert input; bottom-up sums work packages. Choose by evidence and the decision, not a universal ranking of speed or accuracy.
Terms used in this guide. fee: the amount charged to the client; cost: the expense included in the stated project model; margin: revenue left after the included costs; margin percentage = that amount / revenue × 100; baseline: the approved original plan kept for comparison.
In plain words
An estimation method is a way of working out how much effort, cost or time a piece of work needs. The five common ones are analogous (compare with a similar past project), parametric (multiply a count by a known rate, such as hours per screen), bottom-up (add up every task), three-point (use best, likely and worst cases) and expert judgment such as Delphi. The right method depends on how much you know at the moment you are asked.
Who this helps: New presales analysts and project managers who need to choose a method, and anyone preparing for a project management exam question about estimation techniques.
Questions this guide answers:
- What are the main project estimation methods?
- How do analogous and parametric estimating differ?
- What is bottom-up estimating, with an example?
- What is three-point estimating?
- Which estimation method should I use, and when?
Words used in this guide
| Word | Plain meaning |
|---|---|
| ROM (rough order of magnitude) | An early range, labelled cost or price, given before the scope is defined |
| Budget estimate | A firmer range once the scope is written down in outline; used to plan a budget |
| Definitive estimate | A detailed estimate built from agreed scope and a task list; can support a quote |
| Delphi | Several experts estimate separately, compare, and repeat until their estimates converge |
| Baseline | The approved estimate kept unchanged so later results can be compared with it |
| Work package | A piece of work small enough that its owner can size it and say what done means |
How do you estimate a project step by step, and what should you consider at each step?
An estimate is a reasoned answer to “how much effort and cost will this work take?”. The method (analogy, parametric, bottom-up, three-point) is only one step. A newcomer should know the whole path, who owns each part and what to check before moving on.
| Step | What you do | What to consider | Owner | Checkpoint |
|---|---|---|---|---|
| 1. Understand the request | Write down what the client wants and the decision the number supports | Is it a budget conversation or a commitment? How rough can it be? | Account manager and presales consultant | The decision, the date and the user of the number are named |
| 2. Define the scope | List deliverables, boundaries and acceptance in plain words | What is included, what is not, and what the client must provide | Presales consultant | A scope note with a version and an exclusions list |
| 3. Break the work down | Split it into work packages you can size | Testing, release, coordination and handover are work too | Presales consultant with a delivery lead | No deliverable without effort; no effort without a deliverable |
| 4. Choose a method per package | Analogy, parametric, bottom-up or three-point | What evidence you have: past projects, counts, a task list | Presales consultant | The method and its evidence are written beside each number |
| 5. Size effort by role | Hours for each role in each package | Skill level, availability, review time, rework | Delivery lead | Hours reconcile between packages and roles |
| 6. List assumptions, exclusions and risks | Write what you are assuming and what could go wrong | The one unknown that moves the estimate most | Presales consultant and delivery lead | Each assumption has an owner and a test |
| 7. Cost it | Hours times approved cost rates, plus other costs | Rates from the rate card; costs counted once | Presales consultant and finance | Cost is labelled cost, not price |
| 8. Review | Someone other than the author tests it | Trace one deliverable through the estimate end to end | Delivery lead or peer | Findings are fixed before the number leaves |
| 9. Price and state the status | Add margin and say what kind of number it is | ROM, budget estimate or definitive estimate; range and review point | Presales lead and finance | The label and the next evidence step are in the message |
A short worked example
A client wants three reporting dashboards built on two data sources. The presales consultant works through the steps.
- Scope: three dashboards, two sources, one test environment. Exclusions: mobile access and historical data clean-up.
- Effort by role: analyst 60 hours, developer 140, tester 40, project manager 30: 270 hours.
- Cost: 60 × 55 + 140 × 60 + 40 × 45 + 30 × 70 = 3,300 + 8,400 + 1,800 + 2,100 = USD 15,600.
- Risk: the main unknown is data quality; a 30% chance of USD 3,000 of rework is an expected USD 900, which becomes a labelled reserve: USD 16,500.
- Review: the delivery lead traces the second dashboard through the estimate and asks where the access set-up hours are. They are already inside the developer’s 140 hours (about 10 of them), so the totals do not change, but she asks for a separate line so the next reader can see them before the number goes out.
- Price: at a 30% target margin and no revenue-based charges, 16,500 ÷ 0.70 = USD 23,571.43. This keeps the example simple by leaving out commission and an overhead share; if your policy has them, they go in the divisor and the price is higher. Because data quality is unresolved, the planning range runs from USD 15,600 (no rework) to USD 18,600 (rework happens), or about USD 22,286 to USD 26,571 as a price, with the unknown named.
Where to go deeper
- What the number means and how to word it: what a ROM estimate is and ROM vs quote vs definitive estimate.
- How to cost the team: resource-based estimating and the project rate card.
- Method detail: three-point estimating and parametric estimating.
- Turning the cost into a price: price to a target margin.
What are the main project estimation methods?
| Method | What it does | Evidence and limitation |
|---|---|---|
| Analogous | Adjusts comparable completed work | Useful when differences can be explained; weak comparators mislead |
| Parametric | Uses a calibrated relationship such as hours per data source | Needs a supported driver and comparable observations |
| Three-point / PERT | Models alternative task outcomes | Bounds and distribution assumptions need evidence |
| Delphi / expert judgment | Uses structured independent expert input and review rounds | Can challenge uncertain assumptions; agreement is not proof |
| Bottom-up | Estimates work packages and sums them | Needs enough scope detail and a check for omitted or duplicated work |
These methods can estimate effort, cost or duration depending on the defined output. Converting reviewed role hours to cost is a further calculation; multiplying hours by a rate does not, by itself, predict the hours.
Which estimation method fits which stage of presales?
At qualification, use the available reference evidence to explain a planning range. During discovery, investigate the dependencies with the greatest effect on effort. Before commitment, require traceable scope, role effort, cost policy and risk treatment, using whichever validated method or combination fits the work.
For repeatable data loads, a calibrated hours-per-source model may support the estimate. A novel interface in the same engagement may need expert review and a task breakdown. The stage determines the decision and required evidence, not a compulsory method.
When is each estimation method the wrong choice?
A method is unsuitable when its inputs are unsupported: an analogy without comparable scope, a parametric driver outside its calibration range, three-point bounds chosen arbitrarily, expert consensus without relevant knowledge, or a detailed breakdown that invents requirements. Correct the evidence gap rather than switching labels to make the estimate sound more certain.
Independent expert review can be useful even when history exists, particularly to challenge whether that history applies. Use enough detail to support the decision without disguising unknowns as precise task hours.
Which estimation method should you use, and why?
No method is the most accurate on its own: precision comes from how well the scope is defined and how good the evidence is. Answer these in order.
- Do you have at least one genuinely comparable past project? If yes and the scope is still undefined, use analogous estimating.
- Is there a reliable historical rate or formula that fits this type of work? If yes, use parametric estimating — but verify the rate data first (see parametric estimating for services).
- Is a specific task or component genuinely uncertain, with real optimistic and pessimistic bounds, inside a project that is otherwise scoped? Use three-point/PERT for that component (see three-point estimating: the PERT method).
- Is the work genuinely unprecedented, with no historical data and no reliable comparator? Use Delphi/expert judgment.
- For a commitment, verify scope, cost inputs, staffing feasibility and remaining risk. A detailed breakdown is useful where packages are understood; a validated parametric or hybrid estimate can also support repeatable work.
How do analogous estimating and T-shirt sizing work in practice?
Analogous estimating starts from a similar finished project and adjusts it. A past customer portal took 500 hours; the new one has 20% more integrations, so the starting point is about 600 hours, which a delivery lead then challenges. It is fast and needs a comparable project and honest adjustments.
T-shirt sizing gives each backlog item a size, then attaches hours to each size. With invented sizes of S = 40 h, M = 80 h and L = 160 h, a backlog of three S, four M and two L items is 120 + 320 + 320 = 760 hours; at USD 60 an hour that is USD 45,600 of labor cost. Agree the hours per size once, with the team, and revisit them after each project. Sizing is for early ranges; it is not a substitute for a task-level estimate before a fixed price.
Example: the same task estimated with three methods
Consider 40 data sources. An illustrative comparable project used 220 development hours; known additional mapping requires 20 hours. The adjusted analogy is 240 hours. At USD 85 per hour, labor cost is USD 20,400. The 20-hour adjustment needs supporting scope evidence rather than a percentage chosen to reach a price.
A separate calibrated model at six hours per comparable source gives 40 × 6 = 240 hours and the same USD 20,400. Agreement is a useful cross-check only if both methods have credible evidence; they are not independent confirmation if both simply reuse the same assumption.
For three-point inputs O = 180, M = 240 and P = 360 hours, the PERT mean is (180 + 4 × 240 + 360) / 6 = 250 hours, or USD 21,250 at the same rate. The difference reflects the uncertainty model. Investigate it before choosing the role plan; none of these cost figures is a client fee.
How do you turn an estimate into a price?
An estimate of cost is not yet a price. In Managed Margin, a bottom-up estimate becomes an estimation strategy: stages and roles, each role's loading per stage, costed at your planning rates, then priced to the service order's target Managed Margin. Alternatives can be compared side by side, and the one you propose is finalized as the baseline delivery is measured against. See how to price a project to a target margin and project pricing software. It does not run PERT, Delphi or analogous estimates for you: those methods inform the hours you enter.
Estimation methods: common questions
Which method should I use with very little scope detail?
Analogous or parametric estimating, presented as a range. A ROM estimate is the usual form.
Is resource-based estimating the same as bottom-up?
Not necessarily. Resource-based costing multiplies role effort by cost rates. The effort may come from work-package estimates, an analogy, a calibrated driver model or a combination. Bottom-up describes how the work is estimated; resource-based describes how the proposed team is costed.
Is a cost estimate the same as a price?
No. The estimate says what the work costs; the price adds the margin you need and the costs charged on revenue, such as commission.
- Project Margin Management: What It Is and How to Do It
- Cone of Uncertainty in Project Estimation, Explained
Real example: a presales analyst uses four methods in one week
An illustrative story with invented figures, to show how the idea works in a real office.
Ravi is a new presales analyst. On Monday a client wants a rough budget, so he compares with a similar past project: an analogous estimate. On Wednesday an RFP lists 30 screens, and his team’s history is about 24 hours per screen, so 30 times 24 gives 720 hours: a parametric estimate.
On Thursday a fixed-price quote is due. Ravi and the team list every task and the hours each needs, then add them up: a bottom-up estimate. On Friday one integration worries everyone. The team says best case 40 hours, most likely 60, worst case 110. The three-point formula gives (40 + 4 x 60 + 110) divided by 6, which is 65 hours.
Ravi learned that the methods are not rivals. Each fits a moment: a rough guess, a repeatable count, a number to commit to, and a risky piece of work.
Where do estimation methods fit in an IT project?
Choose an estimation method whenever a new estimate starts: each opportunity has a different amount of scope detail, history and time, so the method should follow the evidence you have.
Across the whole estimation lifecycle: analogy and ROM at qualification, parametric and three-point in presales, bottom-up before pricing; then calibration after delivery.
| Before | This step | After |
|---|---|---|
| Upstream: the opportunity, available scope detail, history of similar work and the decision the estimate supports. | A method chosen for the stage, then an effort estimate, then cost and price. | Downstream: pricing, approval, the delivery baseline and calibration against actuals. |
Analogous, parametric, bottom-up or three-point: how do you choose?
The five methods in this guide are alternatives and complements.
| Approach | Use it when | Watch for |
|---|---|---|
| Analogous | A close past project exists | Weak match |
| Parametric | A reliable driver and history exist | Wrong driver |
| Three-point | Uncertainty sits in a few tasks | Averages read as promises |
| Bottom-up | Scope is detailed | False precision early |
| Delphi or expert group | No history; experts disagree | Group bias |
Use the method supported by the relevant component’s evidence and cross-check where an independent basis is available. Do not treat a method name as proof that a fixed fee is ready to approve.
Project estimation methods: frequently asked questions
Which estimation method should I use?
The one your evidence supports at this stage. If you have a close past project, start with analogy; if scope is detailed, go bottom-up.
Can I combine methods?
Yes. Combining and comparing is good practice, and a large gap between methods is a signal to investigate.
How do methods connect to price?
Methods may estimate effort, duration or cost. For a role-based price, first establish supported role hours, apply cost rates and other project costs, then solve for the fee including revenue-based charges and target margin.
How do you estimate a project step by step?
Understand the decision the number supports, define the scope and exclusions, break the work into packages, choose a method for each, size hours by role, list assumptions and risks, cost the hours at approved rates, have someone else review it, then add margin and state what kind of estimate it is.
What should you consider when estimating a project?
Scope and exclusions, the evidence behind each number, hours by role (including testing, release and coordination), assumptions and the one unknown that moves the estimate most, approved cost rates, a second reviewer, and whether the number is a ROM, a budget estimate or a definitive estimate.
Continue the sequence: guides to read next
- Three-Point Estimating and the PERT Formula, With an Example: optimistic, likely and pessimistic estimates combined with PERT
- What Is a ROM Estimate? Meaning, Cost and Example: the earliest, widest estimate
- How to Price a Project to a Target Margin: Formula and Example: turning any estimate into a price
Sources and methodology
- U.S. Government Accountability Office, Cost Estimating and Assessment Guide (GAO-20-195G). Used for the analogy, parametric, and build-up method comparison and when each applies.
- Project Management Institute, Lexicon of Project Management Terms, Version 5.0. Used for analogous, parametric, three-point, and bottom-up definitions.
- RAND Corporation, The Delphi Method: An Experimental Study of Group Opinion. Used for the structured, anonymous expert-consensus method and its origin.
- The worked data-migration figures are original and illustrative, not a customer result.