Often projects are presented with the presumption that it will have a particular impact, outcome, and results. These presumptions may have a history of being inaccurate, and had there been an accurate prediction of the impact, outcome, and result, the project might not have gone ahead, or might have been re-thought.
Accurate outcome modelling should be the foundation of good decision making.
Currently if outcome modelling does take place, it is either by parties that have a vested interest in starting or preventing the project. Or it is done by consultants engaged by these parties. The consultant knows that they’ll get more work if they deliver a prediction that suits the needs of the party paying them. They also know that an inaccurate prediction is unlikely to be reported on or have an impact on their business. This results in any modelling that is done being intentionally flawed to meet the desires of those that require it.
To prove this, a comparison between client desires, the modeled outcome, and the actual outcome would be revealing.
If instead the financial reward was linked to accuracy, then outcome modelling would be much more accurate. As a result, better and more transparent decisions would be made.
Outcome modelling is also done much less frequently than most people would consider prudent. Knowing how likely it is that you’ll get the result you’ve paid for seems like necessary information. But acquiring it is expensive. And information that might prevent a project from proceeding is often not desired. The barrier to initiate outcome modelling needs to be lower.
There are already a number of businesses that have a lot of skill and experience turning available data and information into predictions. These include Deloitte, EY, KPMG and PwC, but there are many other businesses in this space. Naturally businesses will be more accurate in areas where they specialise, such as behavioural change, public health, or transport, and housing.
There should be a focus on modelling outcomes that Benefit points are linked to. If benefit points are based on “number of active transport users”, “number of people who change that behaviour”, etc, then this should be modeled.
Other modeling could also predict impact areas such as
Often predictions will need to be connected or grouped. For example predictions might be for 1 year, 3 years, and 5 years into the future. It should not be possible to hedge predictions by making them contradictory. Predictions for cost, and time to deliver are naturally interdependent and should be grouped. An outline for how this is to be managed should be part of the process and uniform for all participants.
A set of rules that encourage accurate predictions, financially reward those who are most accurate, discourage gaming the system, and entice sufficient participation by capable entities is required. They could include things like: