I recently gave a talk on using geospatial modelling & visualisation to aid decision making I introduced the idea that people don’t want ‘the answer’, but need ‘insights and options’. My premise is that the people making the final decision need to be involved in the process of making that decision. The challenge is that many senior decision makers are time-poor and cannot participate in the actual analysis. This is where the skill of using good data visualisation can make all the difference.
In the ten points listed below I have tried to bring out some of the approaches that can be used to communicate better to all stakeholders and explain to them the options before them. That way they are more likely to understand and accept the final solution. As any reorganisation requires effort then this is crucial, otherwise they won’t trust the results enough to drive forward with the change.
Our brain+eye are extremely good at picking out patterns in visual data if the data is presented well (see book ref 1,3 & 4 at the end). This means you can convey complex information in a way that people will understand if you use good visuals. And if people understand something they are much more willing to accept the conclusions that come from that information.
Our brain interprets information against its existing knowledge base. Therefore try to use images and colours that relate to people’s experiences, e.g. use traffic light colours for good, warning, bad. Also consider the audience: if it consists of people who are very involved then you can cover more complex issues.
The ultimate diagram is one that is instantly understandable to the intended audience without the need for an explanation.
Generally people, especially senior staff, are intelligent but don’t have the time to investigate something in detail. While you can work on the assumption that they trust you with the problem there is great benefit in conveying the insights you have found in a form that they can grasp. Then they become part of the analysis team and can process the findings themselves.
If at all possible try to visualise the ‘now’, i.e. current, situation and show this to everybody involved. Display the current data against the criteria being used on the new requirements being introduced. This will help stakeholders see that you understand things properly and, especially if the current situation is ‘bad’, it will focus the attention of key stakeholders on the need for change.
There is often a mismatch between the term an outside person and the client would use, e.g. is it a clinic, hospital, facility, demand point etc… This can easily lead to confusion. I try to adopt the client’s terms in all my documents, even ones that are not seen by the client. That forces me to really understand the term I am using and not just assume it is something I already know about.
All the studies (see book ref. 3 & 4 at the end) show that our brains work mainly by comparing things. Therefore producing a range of answers with different criteria helps people see the options and trade-offs better. Try and produce options that relate well to the client, including extremes if possible. Scenario planning is a very good approach, as is ‘what would it look like in x years’ time if we did this?’
It is very revealing to listen to the views of all the stakeholders. Why, because they will be evaluating any results produced against their own views. Some of their ideas might be excellent and some might be based on out-of-date information. However including the predominant stakeholder ideas in your final results can be very useful in allowing people to see their ideas in comparison with others.
Many problems are incredibly complex and it isn’t possible or economic to model all the interactions. In these cases you need to make some critical judgements on what and how you might simplify the model. It is important to ensure that all the various stakeholders will accept the model with the proposed simplifications.
In real-life situations there are factors that are very hard to include in a model. For instance a simple location analysis might suggest closing service point A, but actually the lead person there is excellent and won’t move for family reasons. Therefore try to add flexibility to any modelling you do to allow a comparison of client preferences against a no-preferences base model.
With terms like metahuristics and genetic algorithms the client can feel disconnected from the analysis being done. However the final decision and responsibility lies with the client, not the algorithm. Therefore take time to explain how the algorithm makes its decisions and, if you can, point out why it has say ‘chosen point A’ or ‘predicted an x increase’. This can really help them trust the results.