Definition
Estimating is about forecasting the resources (time, finance, effort) needed to act.
It is useful for both business analysts and stakeholders.
The result of this method can sometimes take the form of a simple number. In other circumstances, it is expressed as a scale with minimum and maximum values and expressing data such as probabilities. This scale is also called a “confidence interval”. It is used to establish a level of uncertainty, calculated according to the amount of information available. A large confidence interval means that the person in charge has little information. Each estimate can have its level of uncertainty.
Estimation is done in an interminable way. Business analysts review (and revise if necessary) their estimates whenever new information becomes available. Estimates can be made more accurate through the use of past data (e.g. actual hours spent on an activity).
Why use an estimation?
Strengths
- Relevance: Estimation helps in making decisions about financial means, time frames, or the importance of a group of elements. It also helps to justify these decisions.
- Planning: This technique enables change to be managed in an effective and controlled way.
- Accuracy: A team of competent people can provide an accurate indicator that is close to the real value.
- Adaptation: The repeated implementation of an estimate allows it to be supplemented with various knowledge and to be refined regularly.
Features
Methods
There is a multitude of methods used to estimate the cost and effort of an initiative. Business analysts usually develop their estimates in collaboration with project managers and other team members. Business analysts can also use the many estimation tools and techniques available.
Each situation has its own method.
- Top-down approach: The analysis starts with the higher-level elements and then breaks them down further in the hierarchy.
- Bottom-up approach: This approach focuses on the elements at the bottom of the hierarchy. This allows the details of the activity to be examined and aspects such as costs or effort to be estimated at an individual level. These elements are then summarised at a more aggregate level.
- Parametric estimation: The organisation creates a parametric model based on its own history. This allows the model to be calibrated to the competencies and skills of the teams and processes. Thus, the model provides values and information related to these parameters.
- Rough Order of Magnitude (ROM): This method is used when there is little information available. This usually happens in the early stages of creating a solution. It has a large set of values, but with a small confidence interval. In other words, this method favours quantity over quality. A ROM estimate usually has a maximum accuracy of between +50% and -50%.
- Cascade: This method consists of a sequence of estimates that accompany an initiative. These estimates provide detailed information frequently, particularly about the short-term activities that make up an initiative.
- Delphi: This method combines the company’s history with expert opinion. Through several sessions, individual estimates are discussed with the experts to reach a consensus.
- PERT (Program Evaluation Review Technique): This method assigns three values to each element of the estimate.
- Optimistic value (OV): Represents the most beneficial scenario.
- Pessimistic value (PV): Represents the worst-case scenario.
- Realistic Value (RV): Represents the most likely scenario.
The PERT value for each element is calculated as follows: [VO+VP+(4xVR)]/6
- Final estimate: If you have enough information, you can then conduct a final estimate. This estimate is more accurate than a ROM and is used to forecast more substantial elements such as timelines, final budgets, or resource requirements.
Ensure that these estimates are acceptably accurate (margin of error of 10% or less).
Some of the techniques seen above are inspired by others in the list. So, your teams can also combine several of these methods during the development of an initiative. For example, a final estimate can be cascaded for each phase of the initiative where there is sufficient information. If not, a ROM estimate can be done.
Tip :
Ideally, systematically use several techniques for each estimate. This may require more time and resources at the time but will save you time and resources later on because of the reliable results.
Note :
These methods require participants to have precise descriptions of the elements in question. Make sure you structure the analysis and the elements in order to divide the work optimally.
Tip :
When you complete the estimation process, remember to incorporate constraints and assumptions.
Accuracy of information
This is the uncertainty margin. It measures the difference between the estimate and the true value (assessed later). It is calculated by dividing the magnitude by the mean value of the confidence interval, which gives a percentage.
ROM estimates often have a maximum accuracy of +50% to -50%. Final estimates should ideally have an accuracy of 10% or less.
Information sources
There are many sources of information that can be used depending on the context and their level of accuracy.
- Comparable situations
- Similar elements regarding the assessment.
- Includes projects, initiatives, risks, events, etc.
- Background of the organisation
- Experiences of the organisation on previous work.
- Useful when the techniques used, or the skills of the team are similar between the previous and current experience.
- Expert judgement
- Use of expert knowledge.
- Allows estimation to be based on key knowledge.
- An assessment conducted by people familiar with the subject is often more accurate.
- Experts can be external or internal to the organisation.
Accuracy and reliability of estimates
When a specific attribute is estimated, the accuracy of the estimate is derived from a consensus between the various estimates made. The level of consensus results from the assessment of imprecision (e.g. the difference or standard deviation).
To assess the reliability of an estimate, you can repeat it with other methods or people.
The level of reliability and precision of an estimate is represented as a range of values and its associated confidence level. The estimates made and their potential values thus allow the creation of this range of values.
Example:
One team estimates that one activity requires 40 hours of work.
- A 90% confidence interval suggests that the number of hours required is between 36 and 44, depending on individual estimates.
- A 95% confidence interval suggests a range between 38 and 42 hours.
Thus, the range of values decreases as the confidence interval increases.
Tip :
Estimators can use the PERT technique. This allows an estimate with a wide confidence interval. If they use different methods for each element considered, they can distribute the probabilities. This produces an overall estimate including all the elements, with a confidence interval and a range of values.
The personnel responsible
Responsibility
- In general, the responsibility for an estimate lies with the people responsible for the item being estimated.
Teams
- An assessment works best when it is carried out by a team. This is because it takes advantage of the diverse knowledge of those present.
Experts
- Experts provide a higher level of accuracy.
- An external expert can help when the organisation requires a high level of trust.
- The expert’s estimate is then compared with the internal results.
- This allows gaps to be found and decisions to be made on possible changes.
Tip :
Make sure that the people doing the estimates have the right knowledge. The better the knowledge, the more accurate the estimation results will be to the actual values (calculated later).
Tip :
Your teams can (and should!) combine these different methods.
How to use an estimation?
Read more by subscribing to the BA Toolkit
On top of the brief description of the method, access to:
- an explanation of when and how to use it,
- a practical example,
- tips and recommendations for use,
- downloadable checklists and templates,
- references for further reading,
- …
