Estimation · Business Analysis Notes

What it is#

In the BABOK® Guide, estimation is the technique for forecasting the time and effort a task will take. The BABOK Guide applies it in Plan Business Analysis Approach, where you estimate your own analysis work; in practice analysts also help the wider team estimate solution work.

The training materials make a practical point: the worst estimate is the one never made. A rough number with a stated range lets people plan and lets problems show up early.

Common methods:

Analogy — compare with a similar past task. It is the quickest and roughest method, and works best when the context is unchanged: same methodology, stakeholders, team and tools.
Decomposition (bottom-up) — break a large task into small pieces, estimate each, and add them up. Small pieces are easier to compare with past work.
Parametric — multiply a count by an average effort per unit, for example number of use cases times days per use case.
Three-point (PERT) — take optimistic (O), most likely (M) and pessimistic (P) estimates and combine them: (O + 4M + P) / 6. The spread between O and P also tells you how uncertain the work is.
Expert judgment — ask someone with more experience of this kind of work.
Delphi — several experts estimate independently, compare results, discuss the reasons for differences and re-estimate until the numbers converge. The training materials describe Planning Poker in Scrum as a form of Delphi.
Rolling wave — estimate the whole effort roughly, plan the next period in detail, and refine the overall estimate after each stage. The training materials note the downside, that the total usually grows, and the upside, that you learn about the growth early rather than at the end.

When to use it#

Use estimation whenever someone needs to plan, budget, prioritize or commit to a date, including your own analysis work. Choose the method by what you know:

early, with little information: analogy, expert judgment, rolling wave;
with a clear count of similar items: parametric;
for high-stakes or uncertain items: three-point or Delphi.

Do not present an estimate as a commitment or as a single precise number. If the work is too uncertain to estimate, say what you need to learn first and estimate that discovery work instead.

How to do it#

Define what you are estimating: scope, deliverables, what “done” means, and the unit (hours, days, cost).
Break the work down until pieces are small enough to compare with experience. Use functional decomposition if needed.
Pick a method for each piece, based on available data.
Involve the people who will do the work. Independent estimates followed by discussion reduce individual bias.
State assumptions and a range, not just a single figure.
Add explicit allowances for known risks rather than padding every item silently.
Record the estimate with its date, method and assumptions.
Track actuals against estimates and look for patterns in the differences.
Re-estimate at agreed points and when assumptions change.

The training materials stress step 8. When a team tracks actual against planned effort for a while, systematic bias tends to appear. Once you know your typical underestimate, you can apply a correction factor to future estimates.

Example#

The Workshop (a fictional chain of bike-repair workshops) asks how much analysis effort its online booking and notification project needs. All figures are invented.

Parametric. The analyst expects about 12 user-facing use cases. Past projects suggest roughly 1.5 analyst-days each to elicit, specify and review: 12 x 1.5 = 18 days.

Three-point for the stakeholder workshops on parts ordering, which depend on the supplier’s availability:

Text

Optimistic   O = 3 days
Most likely  M = 5 days
Pessimistic  P = 10 days
PERT estimate = (3 + 4 x 5 + 10) / 6 = 33 / 6 = 5.5 days

Rolling wave. The analyst commits to a detailed plan for the next two weeks (current-state interviews, document analysis), and gives a rough range for the rest: 25 to 35 days in total. The range is revisited after the first wave.

Tracking. Each week the analyst logs actual hours against each task. After the first wave, use-case work is taking about a third longer than planned, so the remaining parametric estimate is adjusted upward and the reason (unclear parts-ordering rules) is added to the risk register.

Common mistakes#

Giving one number → give a range and the assumptions behind it.
Estimating work you have not broken down → decompose until each piece is familiar.
Anchoring on the first number someone says → collect independent estimates before discussing.
Silent padding → make contingency visible and tie it to specific risks.
Estimating for others without asking them → involve the people who will do the work.
Never checking actuals → compare, learn, and adjust future estimates.
Treating estimates as promises → separate the estimate from the commitment a manager makes based on it.

Checklist#

Scope, deliverables and unit of estimation are clear.
Work is broken down into comparable pieces.
The method for each piece is chosen deliberately.
The estimate is a range with stated assumptions.
Risks have explicit allowances.
The estimate is recorded with date and method.
Actuals are tracked and compared.
Re-estimation points are agreed.

Sources and further reading#

IIBA. A Guide to the Business Analysis Body of Knowledge (BABOK® Guide), version 3. International Institute of Business Analysis, 2015, technique 10.19 “Estimation”.
Business analysis training materials (article on estimating business analysis tasks: analogy, parametric, three-point, expert, Delphi and rolling wave methods).