Formula-Based vs. Experience-Based Estimation: When to Use Each

What experience-based estimating is good at

Judgment handles novelty. When an engagement has an unusual shape, a difficult stakeholder environment, or a technology nobody has scoped before, a practitioner's read is the best instrument available.

Its weaknesses are consistency and transferability. Two estimators produce different numbers for the same work, the reasoning lives in one person's head, and a client challenge is answered with seniority rather than structure.

What formula-based estimating is good at

Formulas convert repeatable delivery patterns into arithmetic. Once you know that each integration costs a predictable amount of effort, the estimate stops depending on who is available to build it.

The benefits compound: estimates become comparable across the firm, junior staff can produce credible first drafts, and a client conversation about price becomes a conversation about drivers.

The risk is false precision. A formula built on thin experience produces a confident number with no more information behind it than a guess — and it is harder to question because it looks like math.

Signals that you are ready for formulas

• You recognise the shape of an engagement within the first discovery call.
• The same one or two variables explain most of the effort difference between projects.
• Different estimators produce noticeably different numbers for similar work.
• You have delivered enough comparable engagements to know where the last ones landed.
• Reviewing estimates has become a bottleneck for your senior people.

Signals that judgment should stay in front

• The engagement type is new to the firm.
• Effort is dominated by stakeholder dynamics rather than countable units.
• Requirements are still moving while the estimate is being built.
• Your comparable history is small enough that a rule would encode one project's luck.

Finding your first drivers

Start by looking backwards, not forwards. Take the last several comparable engagements and ask what actually explained the difference in effort between the smallest and the largest.

In most practices two or three drivers explain most of the variance. Integrations, environments, user or site counts, workshop counts, review cycles, and regulated-review requirements are common candidates. The goal is a small number of drivers you can ask about during discovery.

• Pick drivers you can determine in a discovery conversation.
• Express each rule in plain language before encoding it.
• Keep management and review effort as their own rule rather than a hidden margin.

The hybrid most firms end up with

In practice, mature estimating is layered. Formulas produce the baseline for the repeatable parts of the engagement, and judgment adjusts for the parts that make this client different.

That layering is also the most defensible thing to present. You can show the client the rule that produced the baseline and the specific reasons you adjusted it — which is a far stronger position than either a bare number or an unexplained model.

Keep the rules alive

Estimation rules describe how your firm delivered in the past. Delivery changes: tooling improves, methodology tightens, team composition shifts. Rules that are never revisited slowly become a record of a firm that no longer exists.

Reviewing estimated versus actual effort after delivery is what turns a formula from a guess into an asset. It is also the least glamorous habit in professional services, and the one that separates firms that estimate well from firms that estimate confidently.