“How do I get to a digital-twin KPI-system from my current KPIs?”

I made the case in another post that a digital twin business model IS the KPI system that explains how the business is performing, and will likely perform in future. But how to get there? I also mentioned that the model can start as a high-level summary, but add as many levels of detail as you want. 

“But I already have a KPI system! How do I switch to this better digital-twin version?”

First – let’s be careful! You probably thought the current KPI system was fine until you realised you could have something better. So …

… don’t throw anything out just yet

… change over step by step

What NOT to do. It’s tempting to say “Well, I’m mostly happy with our current KPIs, so let’s just put those into the kind of visual, time-charted diagram Kim tells us about.” That won’t work

Why not? … I explained in another recent post that the process for building a digital twin model is based on rigorous theory, so will, by definition be correct (At least, if you follow the rules!). That will not be the case for your existing KPI system. Not because anyone was foolish or careless – simply because that system was built without such a solid foundation.

So, your existing KPIs will likely:

… include many items that are not necessary, or that may be misleading,

leave out some items that are critical, especially some concerning rates-of-change to key elements,

… not capture accurately the what-causes-what dependencies that drive the system’s performance.

Existing KPIs won’t be hopeless, of course! They will include many correct items too, like sales volume and revenue, and the very simple relationship between the two, probably split by product, customer-segment, region etc. But too many other items will be missing or wrong to make those KPIs a good start point.

Start simple – follow the digital-twin process. The 4-step digital twin modelling process will initially seem not to show up anything new, but hang in there …

Starting from step 1 – the outcomes of interest, such as sales and operating profit – the first few causal links will likely trace out the causality back from operating profit through revenue and costs to sales volume. (For anyone quite new to this, here’s a tiny section of such a causal model – history and alternative futures for a small service provider. As I say, nothing new yet, except that you can see the trajectories and dependencies involved.)

Then step 2 will add how changing revenue and sales rates reflect changes to customer numbers and their purchase rate. (Up to this point, the model will likely match some of your current KPIs exactly.)  

Step 3 will then add an explanation of how changing customer numbers reflect how customer win- and loss-rates have changed. (For many cases, this will be the first critical piece of causality that their existing KPI systems don’t capture. Even where a KPI system does capture these items, it will likely not display them in a visual image that makes clear how and why customer numbers and sales have changed as they have over time – and it certainly will not show how those numbers will likely change in future.)

On the cost-side of the business, the same steps 1-3 will give a clear, visual explanation for how rates-of-change for staff and maybe capacity explain the time-path behaviour of operating costs.

Pause your transition here!

Cross-check with existing KPIs. So now you have the first few small groups of KPIs in a rigorous causal structure – essentially, the Income Statement plus how changing customers, staff and other resources have been driving changing revenues and costs. So:

Check that this first stage of the digital twin model is right, and displaying all the time-charted causality management needs, in order to understand how the business has been performing at the highest level.

Cross-check the items in this first stage model against the equivalent items in your old KPI system. If there is anything in the old KPIs that is not in the digital twin, ask if that item is actually telling you something important.

(A whole class of missing items will likely be ratios, like gross-margin-% and labour-cost-%. That’s because the digital twin process traces out causality with absolute values. By all means add such ratios to the digital twin model – but DON’T allow them to dominate decision-making! I have explained elsewhere how dysfunctional this can be. E.g. revenue falls, so labour-cost-% rises, so we cut staff, which hits quality, which loses customers, which loses yet-more revenue …)

Start using this digital twin model instead of the corresponding old KPIs. When management is happy, drop that part of the old KPI system.

… and REALLY use it!  Unlike your old KPI system, your digital twin model will show you how the business will likely behave and perform into the future. That scenario will reflect assumptions about how key items like price, customer win/loss rates, hiring etc. will change into the future. So now you can vary those assumptions and test alternative future decisions, before landing on what exactly is best to do.   

Repeat the same cycle. That process of setting up the first, high-level digital twin gave you THE time-based explanation for why operating profit has changed as it has. (Since the digital twin matches everything in the real-world system, there cannot be any ‘other’ explanation.)

Now you can move on to step 4 – build the digital-twin explanations for how numbers of customers, staff, intermediaries, product range/functionality, capacity and other factors have interacted and changed over time.

This process will give you the few additional causal structures needed to complete the high-level digital twin model. Commonly, these pieces include:

… how customer-growth and retention reflect marketing, sales, price, customer service. (A whole new set of factors often missing from existing KPIs on this issue concerns these same items for competitors! E.g. your digital twin may show a relationship between your price and customer purchase rates and/or customer win-rates … but those items cannot possibly be independent of the price your competitors charge.)

… and, if relevant, the same for intermediaries (your direct customers) and their impact on end-consumers and sales

… how and why staff numbers have been changing, due to hiring, promotion and attrition rates, and the impact on changing staff costs – across all teams.

… how effort and spending on product development have been changing the range and functionality of products and services you offer.

Dig deeper. Unless yours is an especially simple business, your existing KPI system will almost certainly have layers of detail. It may split out, for example, the performance of different regions, customer-segments or products, different service teams, the development of distinct functional staff teams and so on.

Your digital twin dynamic business model can do exactly the same. (A current project of mine, for example, shows the customer acquisition and on-boarding performance and resulting revenue streams in total – but underneath that are matching models for each of three customer segments).

So … simply replicate any pieces of the high-level model for any such segments or sub-divisions. Those become the KPI systems for the executive or team responsible for each of those pieces.

This article can only be a short summary of the process for getting from an existing set of KPIs to a clearer, more rigorous and forward-looking digital-twin KPI system.

To properly implement that transition, you should go through the process we teach in the Dynamic Business Modeling courses.

Leave a Reply

Your email address will not be published. Required fields are marked *