Project Estimating Practices


Posted on 17th July, by JimYoung in Blog. Comments Off on Project Estimating Practices

im1

Estimating is one of the most controversial subjects in project management. Project estimation can even make the best project management experts pull their hair out. We all know that cost and time overruns can lead to project failure. The challenge with estimating is that it always involves some uncertainty. Some of the factors that contribute most to this uncertainty are:

  1. Experience with similar projects: The less experience we have with similar projects, the greater the uncertainty. If we’ve managed similar projects, we will be able to better estimate the costs.
  2. Project duration: The longer the project, the greater the uncertainty. If our project is of a short duration we are more likely to account for most of the costs. Also, we will be able to better estimate costs for the time periods that are closer to the present.
  3. People: The quantity, aptitude and attitude of our project workers will be a huge factor in estimating their costs. Early in the project, we may not even know the specific people that will be on our project. That will increase the uncertainty of our cost estimates.
  4. Poor design: While no one wants to admit it, poor design is time and again the root-cause that plays spoilsport even when the best efforts are taken. Often, people conjure up a tight design on the basis of a requirements document, that is misconstrued to be frozen and do not show the forethought to make the design scalable.
  5. Not splitting the tasks enough: Most projects have a WBS (Work breakdown structure), but sometimes they are not broken enough to be conceptualised with clarity and there is then no proper basis for estimation and then well, it slips out into a subjective guess.
  6. How much contingency is the right amount: What is the right contingency and management reserve once we arrive at an estimate?
  7. Top to bottom scheduling: Instead of doing bottom-up estimation, some projects start with – “I need the budget set at X dollars” and then a work breakdown is done where the task estimates are retrofitted.
  8. The risk of analogous estimation: Often, project estimates are done based on an expert judgment or from past projects’ experience. While picking an analogy and mapping the estimate might seem like an intuitive thing to do, it’s often risky because of the numerous variables in a project, the unique elements, and the people involved and their skillset. The job of estimation turns out to be enormously exigent owing to the diverse nature of every project.
  9. Ignoring team capacity: There is a lot of debate about what unit, estimates need to be provided in – should we measure complexity, time or effort? Irrespective of what unit is followed, some project managers tend to ignore their team’s capacity. This inherently adds a risk layer that makes the estimate unreliable.
  10. Management reserve: The contingency reserve is used to manage identified risks, while management reserve is used to manage unidentified risks. A common method for determining the management reserve is to simply add 5-10% of the cost baseline, but this of course could be wildly wrong and depends on the accuracy of the cost baseline and risk analysis.

Fortunately, there are some tools and techniques that we can use to develop more accurate cost estimates. Once a WBS is prepared we can then analyse the identified work to produce detailed “bottom-up” estimates at work package level.

im2

Estimating is difficult and the key thing to remember is that an estimate is only an estimate. The project’s estimated work effort, duration and cost are living things throughout our project life and should be revisited when anything significant changes in our project or when we know sufficiently more to further refine the estimate.

The NZ Wellington town hall strengthening cost has gone from $43m to a huge $130m excluding on-going maintenance costs, recognising too that no doubt loan interest rates will rise and the debt will still need to be serviced. Perhaps the original guesstimate was deliberately held down to obtain approval and placate ratepayers, or the business case, if there was one, that justified this investment, was never updated in case it demonstrated that this vanity project was not viable, particularly given that Wellington City already has a surfeit of suitable venues – TSB Arena, Michael Fowler Centre, St James Theatre, The Opera House and Academy Galleries.  The Council could have sold the building for conversion into much needed apartments and perhaps used the proceeds to roof the Cake Tin to thus create yet another all-weather venue.

There is no such thing as an exact estimate, whereas an accurate estimate or 50:50 estimate has the same likelihood of being too little as it has of being too much. We usually only know exactly how long our project takes and how much it costs when it’s done. Yet as project managers we’re often seen to be as good as our estimates, illustrated by this diagram analysing time and cost: 

Estimating Failure

im3

There are many different ways to predict the duration, work effort and cost of a project. Here are some of the most common techniques:

  1. WAG Estimation. A WAG or “wild-arse guess” is an off-the-cuff or order-of- magnitude guesstimate for the cost or completion date for a project, used when there is insufficient data available to support a more informed estimate. The SWAG (scientific or stupid wild-arse guess) variation probably involves the use of a math’s calculator.
  2. Top-down Estimation. Top-down estimation is based on a high-level WBS whereby the entire project is estimated to obtain a rough result.
  3. Bottom-up Estimation. Bottom-up estimating or WBS estimating involves estimating each piece of project work at work package level and then summing the estimates. It’s usually the most time-consuming project estimation technique, but also the most accurate.
  4. Delphi Technique. Also known as expert judgment, the Delphi technique relies on surveying experts independently and then averaging their responses. It’s as good as the experts we ask. The name “Delphi” derives from the Oracle of Delphi who according to Greek mythology was a priestess who could predict the future.
  5. Parametric Estimation. Parametric estimating uses mathematical models, such as cost per square metre or cost per tonne to develop an estimate.
  6. Comparative or Analogy Estimation. When similar projects have been undertaken in the past, these may provide us with the data needed to produce an estimate. This technique clearly can’t be used for unprecedented projects.
  7. Three-point Estimation. Often, three-point estimates can be combined with other estimating techniques to produce a more rounded set of values. As the expression suggests, this technique takes three inputs – the best case, the worst-case and the most likely, and then averages these three figures: (O + M + P) / 3.
  8. PERT Formula. PERT (Programme Evaluation and Review Technique), developed during the US Polaris Missile Project 1960, provides a weighted average estimate given estimated Optimistic (O), Most Likely (L) and Pessimistic (P) durations or costs: (O + 4L + P) / 6. PERT aims to account for uncertainty. This typically beta (β) frequency distribution recognises that here is a limit to how quickly a job might be completed, but very little limit to how long it could take. Imagine that an Optimistic estimate is 6 days, a Pessimistic estimate is 20 days, and the Most Likely is 10 days.

PERT Example. If we use the PERT estimation technique, that gives us a Best Estimate of 11 days and a standard deviation (σ) of 2.33 days calculated by the formula (P-O)/6. 
Here’s another PERT example. Say we estimate that writing and designing an e-book will take 20 hours. That’s our best guess. Now we consider the risks. If the client approves the e-book upon the first delivery, that’s the best-case scenario. Our optimistic estimate might then be 15 hours. If the designer is late delivering the files, the expert is not immediately available for an interview, and the client requires three rounds of revisions, the pessimistic estimate might be more like 30 hours. In this example the result is {15 + (4 x 20) + 30} / 6 = 21 hours. The following table shows PERT calculations for a project of four sequential critical path tasks, were project duration is calculated at 48.5 days – a single figure due to the Central Limit Formula (CLT) that says even though the individual task durations follow the beta distribution (or any other distribution), when they are added up, the resulting distribution (of project duration) is always Normal.

 

im4

Chunking the project enables those people with the requisite experience and skillsets to more accurately estimate resource needs, work effort, duration and cost at WP level, to produce a project bottom-up estimate against which our financial performance might be assessed. Some common estimating issues include being too optimistic, not being clear about what we’re estimating, not factoring in everything that needs doing, and not distinguishing between effort and duration. Some key points are these:

  1. Check we’ve identified all the work to be done, including all those jobs we do as PM – managerial overheads.
  2. Ask two team members to separately produce estimates for the project. If their numbers are close, we can be more confident about their accuracy. If they differ widely, we should discuss why.
  3. Ask the experts for their estimates, but take into account any tendency they may have to be overly bullish (ie, an optimism bias) or too cautious. Estimates provided by those without hands-on expertise will be educated guesses at best. Ideally, the numbers will come from the team members who will be doing the actual work. After all, each person has their own work speed and people will be more motivated to make finish dates and budgets that are based on their own assessments. But watch for padding – that extra amount randomly added as a safeguard against the unknown. To pad is bad – it’s unethical and may deprive other projects of funding.
  4. Don’t confuse elapsed time, duration and work-effort. Elapsed time includes weekends and statutory holidays, duration is measured in workdays, and work- effort is measured in person-hours or person-days. Work effort is the usual basis for costing labour.

Effort, Duration and Elapsed Times

im5

  1. Check with those who have done similar projects recently, review any relevant historical data, reports, lessons learned, and consult published productivity data. However, estimates based on memory are subject to the cognitive bias of the estimator, therefore involvement of others usually helps provide some balance.
  2. Use normal work-package team sizes for estimates. Sometimes it’s not practicable to add more people or do a particular task with fewer people. And if a task takes one person two hours, don’t think that 120 people could do it in one minute.
  3. Where appropriate, conduct trials or dummy runs or build prototypes.
  4. Focus particularly on the more expensive tasks (Pareto Principle or 80:20 Rule) given that if we’re 10% out with a $100 task it usually doesn’t matter too much, whereas 10% out with a $100,000 task does matter. Focus most of our effort on the important few, rather than the trivial many.
  5. Adopt phased estimates (rolling wave strategy) to re-estimate in waves as our project proceeds and details become clearer. Work that is imminent is estimated in detail, while work that’s some time off is estimated at a higher level. All estimates need to be re-assessed as more details become available.
  6. Allow for the unexpected – a contingency sum to cater for identified risks and approved scope changes, and a management reserve to allow for unidentified risks.
  7. Learn with experience – update our estimating database.
  8. Identify and allow for all factors that may affect time such as setup time, learning curves, weather, weekends, work hours, holidays, skill levels, machine variations, industrial action, sickness, fatigue, and staff turnover. Climate change means that weather and “wet days” are now less predictable.
  9. Find out current labour and material costs, insurance rates, interest rates and exchange rates for costing purposes.
  10. Assume people will be productive for no more than 80 percent of their work time, and recognise that those working on multiple projects take longer to complete tasks because of the time lost switching between them. Even when people are assigned full time to our project, they will not spend all day every day working on it. Most PM software tools allow us to set up an availability profile for our resources, with task durations automatically adjusted to reflect this availability.
  11. Be cautious about scheduling overtime since an eight-hour day might easily become a 16-hour day depending on how project work proceeds. Working overtime is a contingency we might need to use during execution.
  12. Develop a spreadsheet, document estimate assumptions, and include the estimate date and its author. When generating estimates, we’ll often have to base them on assumptions, especially when requirements are not well defined. Thus, we need to record these assumptions with our estimates so that this information is available to others who might need to know how our estimates were derived or need to update our estimates.
  13. Strenuously resist calls to reduce our estimate without a commensurate scope reduction. If it’s proposed to reduce our budget we need to recommend how this might be achieved – perhaps by using cheaper materials or by reducing product functionality.
  14. Provide estimators with feedback on variance – the difference between their estimates and actual costs and durations.
  15. Consider using the Delphi technique whereby we gather estimates from several experts and take an average.
  16. Use PERT (Programme Evaluation and Review Technique) to obtain a weighted average estimate of time or cost.
  17. PMs and team members should feel free to challenge estimates. If something doesn’t feel right, ask questions and offer solutions. If disparities crop up they may be due to different understandings of what’s required.
  18. Recognise Parkinson’s Law that states “work expands to fill the time (and budget) available,” and Murphy’s Law that states “if anything can go wrong it will,” which requires we manage the risk and may include contingency money or time for known but accepted risks.

The total project budget is the baseline estimate plus the contingency reserve plus the management reserve, where the baseline is the cost of all WBS tasks including PM overheads. A quote received for any outsourced work will include a profit margin and GST. The contingency reserve is for anticipated or known risks, whereas the management reserve is an allowance for unanticipated or unknown risks and variations, but without detailed history of costs the management reserve is difficult to estimate and if established is held by the project sponsor.

The contingency reserve is not a slush fund for bailing out the project and may be determined by costing identified risks and is often reduced as the project proceeds should these risks not occur. Given the table below, probability theory tells us that the likelihood of say both risks A and C occurring is 10% x 40% = 4%. Risk D is an opportunity – money saved. 
Another way to determine our project’s contingency is to first decide what level of confidence we would like for on-time project completion. The PERT formula provides a 50:50 duration estimate whereby there is the same likelihood of the actual duration being greater or less.

Contingency Calculations

im6

Reserves

Added to the project contingency is the management reserve. Assuming a cost baseline of $121,000 and a 5% management reserve, the project manager would calculate the management reserve as $6,050 (i.e., $121,000 x 5%). The $6,050 management reserve would be added to the $121,000 resulting in a total project budget of $127,050.

The range of reserves and confidence levels change over time. As project managers progressively elaborate the project, confidence in the budget estimates should increase. During the Initiating Process, a Rough Order of Magnitude Estimate could be +/- 50% from actual. During the Planning Process, a Budget Estimate might decrease to a range of –10 to +25% from actual. Later in the project, the Definitive Estimate might be a range of +/- 10% from actual.

Project managers should determine if reserves will be reduced during the project as work moves beyond identified risks without the need to use contingency. Reserves should not be used for gold plating a project.

Project Duration Confidence Level Calculations

If the BET = (O + 4L + P)/6 = (22 + 160 + 100) = 47 days, the Standard Deviation = (P – O)/6 = 13 days, and Contingency = Standard Factor x Standard Deviation. The Standard Factor is the same for all projects. As with all estimates, the accuracy of these answers depends on the accuracy of the input data. To save us from these laborious calculations, if they are really needed, we could invest in some pricey software such as @Risk or Crystal Ball to run a Monte Carlo simulation to determine for example what project cost or duration would provide us with the confidence level we require, or what are the chances of finishing our project within budget and on time.

im7

Doing time and cost estimating accurately is often the difference between consistent project success and frequent failure. Organisations often judge whether our PM has succeeded or failed depending on whether or not our project has been delivered on time and on budget. To have a chance of being successful as a PM we need to negotiate realistic budgets and achievable deadlines. We might check our estimating prowess with the following questions:

Estimating Checklist

im8

We should avoid giving our sponsor and other stakeholders any firm commitments about our project cost and duration until we complete our research and calculations, and when we do, range estimates are much more realistic than are single figure estimates, that’s say $4,750 to $5,500, or $5,000 – 5% or +10%, rather than a $5,000 single figure. The final budget is the total cost of the project, including reserves, which is the amount of funds the organisation needs to have available for our project. If that number exceeds the cost constraints imposed by the project charter, we must suggest options to our sponsor such as removing some scope or obtaining more budget.

Our project budget includes both direct and indirect costs. Direct project costs include team members’ wages, travel, project materials, supplies and equipment. Indirect project costs include overhead costs such as office space, heating, lighting, furniture, fixtures and equipment, and the project’s share of our organisation’s administrative costs. From a cost viewpoint, the ideal project duration would be when total costs are lowest.

 

Project Costs

im9

There is no easy way to produce accurate estimates, except with experience. Remember, it’s also prudent to re-estimate periodically throughout the project, and always record the actual costs so that we and other PMs may benefit from the experience. Importantly, don’t be bullied into a budget or timeframe we cannot achieve and don’t bully team members into committing to something they cannot achieve. Seeking more funding may be perfectly valid, particularly if we agree the requirement for increased funding and the reasons for this in advance with our sponsor. Similarly, we should declare any surplus. At the start of a project it is worth identifying and agreeing what type of issues could increase expenditure beyond the budget and warrant additional funds or reduced functionality.

Project Bottom-up Estimate for Recruiting Project

im10

Once the cost estimate and contingency are agreed, these become our budget. The simplest way of graphing the planned use of this estimate against time is an “S”curve that shows the anticipated cumulative expenditure over our project’s duration. The name derives from the S-like shape of the curve – flatter at the beginning and end and steeper in the middle, which is typical of project expenditure. Analysis of S-curves allows us to identify project growth, slippage, and potential problems that could adversely impact the project if no remedial action is taken.

Project Expenditure Curve

A graph of expenditure allows a cash flow forecast to be developed and a drawdown of project funds to be agreed. Actual expenditure inevitably varies somewhat from planned expenditure. A report showing an S-curve for the original budget alongside a curve of actual spend to date, clearly shows how actual expenditure varies from that originally predicted. Also, this baseline prediction can be used as the basis for Earned Value Management (EVM).

When project implementation is underway, tracking actual time and cost as we work is very important for two reasons. First of all, that data is useful input for future estimates. Second, we compare our estimated costs to actual cost, and our planned schedule to our actual schedule, as an important source of lessons learned that may include better methods of time and cost estimation for our next project.





Comments are closed.



Blog

My latest thoughts on Project Management and life.

The Project Audit

Project auditing is a formal type of “project review”, most often designed to evaluate the extent to which project management standards are being followed...

Monitoring Project Progress

We project managers need to be constantly on our toes to ensure that our project is on the right track. Monitoring includes all the...

Typical Key Project Roles & Responsibilities

Professional relationships between us project managers (PMs) and our sponsors are organic – they are born, they grow, they mature, and ultimately they conclude....