A Structured Approach to Estimating Costs
I need to estimate the cost to complete a project we’re considering. It won’t be a large project, but there will be mixes of operational and capital costs, work to be completed by internal and vendor staff, and purchases of new equipment. Basically, I’m looking for an approach that will help me identify and estimate all the different cost areas, and is robust enough to stand up to scrutiny. I’m not familiar with statistical modelling and don’t have any relevant software tools.
We developed an approach we call ‘80/80 Estimation’. It’s a fairly light but structured method that focuses on building range estimates around what’s most likely to happen. It’s best used to estimate costs, but can be used for other purposes, such as modelling a timeline.
80/80 Estimation aims to create accuracy right from the start, though within a range. Because it’s a light approach, it’s well suited to being revised as better information becomes available so that range estimates can be tightened. Consider using it at initial scoping stages, then revisiting it when developing a business case, when doing detailed planning, and/or after procurement work is completed.
80/80 Estimation has similarities with the popular Monte Carlo methods, but excludes the latter’s intensive probability modelling. This means it doesn’t need specialist software or knowledge. If your investment was to be substantial, with significant areas of uncertainty, then we’d generally recommend performing a more thorough quantitative analysis. We can help with that.
We strongly recommend using a spreadsheet to create your estimation tables. Let’s look at how it works.
First create a table as per Figure 1.
You’ll need to spend some time thinking about how you’ll structure the information you enter into the Project Component column. If you have existing project definition information then for clarity and consistency it might be a good idea to start with that structure – all the information you’re aware of at this stage. That’s likely to result in a mixed list of project deliverables, activities and purchases. If your costs are going to be mainly based on personnel time, then you might want to take a different (or even combined) approach and list roles instead.
Now complete the remaining columns for each component. You will seldom need to enter information into every column for any one line item.
- Days of Effort. Enter effort estimates for the people you’re going to need. The ’80 Low’ and ’80 High’ figures should define a range of reasonably possible effort. At this stage ignore ‘outliers’ that might move you outside that range, but are highly unlikely to. Note that throughout this process the ‘Likely’ effort estimate won’t necessarily be the mid-point between the high and low figures – it could be any point between them.
- Personnel. Treat these columns as optional for now, though they’re useful to kick off resource planning.
- Capital Expenditure. Enter cost estimates for all the capital items you can think of, including for personnel identified under ‘Days of Effort’ when you’ll be capitalising their work. If you need to add granularity (e.g. for different resources or capital item purchases) then insert extra rows within/for individual ‘Project Components’.
- One-off Operating Expenditure. Enter operational cost estimates in the same way as you did for capital costs. Note that this version of the spreadsheet focuses on the cost to implement change and doesn’t cover ongoing operational costs. If you want to model those as well then make sure you keep one-off and ongoing costs separate.
- Assumptions. Note any assumptions you’re making when identifying and estimating cost items. You’ll test these assumptions when you look at outlying items later.
When you have totalled your columns you’ll be starting to build a picture of the likely costs of your project. Now we need to make sure we’re considering what contingency might be needed.
We’ll accommodate risk in our estimates. This works best if you’ve already spent time identifying and assessing risks. You can skip this stage for the moment, but you should come back to it later. Add a second table under the first that looks like Figure 2. Note the column headings are a little different.
For each risk, estimate the effort and cost required to mitigate it (if mitigation is possible). Also estimate the ‘Hit Cost’, i.e. the impact on the project if the risk is realised. Decide what the ‘Likely’ effort and cost figures should be; they might be the ‘Mitigation’ or ‘Hit Cost’ figures, or a middle ground. Use the ‘Personnel’ columns if you plan to invest in mitigation activity.
Next we’ll work in a list of Outlier contingency items. These are things you don’t expect to spend any time or money on, but could conceivably come up. They could be informed by your initial set of assumptions, or even be extreme risk-related events that aren’t covered by risk ‘Hit Cost’ figures.
Add another table as shown in Figure 3. When you see how these columns work you’ll realise why Risks and Outliers are treated separately. This is where there’s a little emulation of Monte Carlo style modelling.
‘Low’ and ‘High’ expense figures should define the range of possible costs if the outlier comes within the project’s scope. So don’t simply enter a ‘Low’ figure of $0 by default
‘Likelihood’ should be expressed as a percentage. Ideally baseline percentage figures will be mapped to a set of definitions, probably based on a standard risk management approach. So for instance if your organisation defines ‘rare’ likelihood risks as occurring between 0% and 10% of the time, then you might choose to enter 5% for all outliers you consider will occur that often.
The simplest way to calculate ‘$ Impact’ is to create a formula that multiplies the mid-point of the ‘Low’ and ‘High’ figures by the ‘Likelihood’ percentage. (We sometimes need to make this a little more complicated.)
Here’s an example. For Project X we don’t expect there will be any need to pay recruitment agency fees, but there’s a small chance recruitment will be needed (yes this particular example might instead be treated as a risk, but let's ignore that). We estimate that if these fees do end up being incurred then they’ll be between $10,000 and $30,000. We’ll say this outlier is ‘unlikely’ to occur, which by our organisation’s definition equates to a 20% ‘Likelihood’ value. The impact of this outlier should be $4,000, i.e. 20% of $20,000 ($20,000 being the midpoint between $10,000 and $30,000).
By totalling all the ‘$ Impact’ figures you’ll generate a project contingency allowance that’s a little more scientific than simply adding an overall flat percentage margin.
Finally it’s time to total the figures, as shown in Figure 4. For transparency’s sake we recommend keeping the outlier / contingency component separate, as shown.
Calculate the two sets of subtotal figures by adding the corresponding ‘Likely’ totals from the Risk table to the ‘Low’, ‘High’ and ‘Likely’ Project Component totals in turn.
In this example we’ll use the same Outlier ‘$ Impact’ figure across each of the total columns (one figure across the capital expenditure row, one across operating).
You might be able to see other ways to generate estimate totals that could be used depending on the circumstances. For instance you might change how risk impacts are incorporated. If you have outliers that are dramatically different in size then the ‘swings and roundabouts’ approach used in this example might not be appropriate.
Be realistic when you communicate the results of your estimation work. When you’re at the early stages of scoping an initiative then you should keep in mind that an educated, scientific, structured guess… is still a guess.
This estimation approach scales well, particularly to combine individual project estimates to create traceable programme-wide views.
For more information on project cost estimation method, email us at Maven@consultmaven.co.nz or call us on +64 4 801 6026.