September 1962 saw the release of a gritty UK film entitled ‘The Loneliness of the Long-Distance Runner’ and I have often considered plagiarizing that for a ‘to-be-written’ book for my retirement ‘The Loneliness of the ‘Whatever Job I am doing at the moment which involves a lot of air travel’.
Having spent that last 20+ years on regular business travel during that time working across complex ‘Computer-based’ software Projects – I thought I would share some of my recent experiences as a Customer Success (& Account) Manager – and what makes from my/our view – a Project successful – especially in Big Data, Analytics & Machine Learning at Predikto.
A previous Sales Director/Manager of mine had an old saying :
‘Keep to the plan, follow the steps – if you skip steps, BAD THINGS HAPPEN’.
I truly believe he was right.
Here, at Predikto, we have similar ‘mantra’- and I’ll share the Top Five which keep us focused on delivering successful Projects – namely:
- This is a journey – enjoy the trip…
If you are expecting to flick the ‘Predikto switch’ and turn the darkness into some new evangelical ‘Light’, then I am afraid you will be sadly disillusioned. Machine Learning projects are just that – they take time to implement (allow the engine to initially learn) and then to allow the engine to learn and optimize its answers (predictions) over time.
- This is a Partnership – not Vendor /Customer Relationship
Partnerships require time, trust & the right skills to develop properly. Sure, we provide the vehicle and the customer provides the data – but it is much more than that. Weekly face-to-face meetings, utilising the software, reviewing the notifications, working out what the top ten features are indicating, etc. It’s a joint effort for a team of subject matter experts – Predikto on their Enterprise Suite (MAX, Recon GUI, etc), Customer on their data, assets, etc.
- We provide Notifications – not Root Cause Analysis (RCA)
A notification in our GUI (Predikto Recon) is MAX tapping the user on the shoulder saying ‘hey there Is an issue here’ and MAX will provide as much information as he can as to the reason for the developing issue(s). This notification gives the user a place to start looking for the issue but may not be the root cause.
What do I mean exactly? Recently, on a fleet of locomotives, we saw an alert on one asset for a potential motor drive failure – and (one of) the main feature indicated was high temperature in the windscreen wiper system. What – you might ask – does wiper system have anything to do with driver motor failures? Simple, the motor driver wiring passing around the locomotive body was near the wiper system and was, locally, getting hotter than normal – the hotspot developing nearest the wiper system where only one temperature sensor was available…
So, consider what Predikto MAX is telling you – he might be using a ‘proxy’….
- Now I have a ‘helper’ who has ‘got my back’ – so what?
So – then you need to consider regular calls with your Asset Management Team and jointly reviewing the notifications with your Customer Success Team and identify which ones are most relevant and why?
The questions you now need to be asking are:
- ‘I have this new information – how does it help me?
- ‘Does this information help me look at the problem differently?’
- ‘What other information do I have to help me contextualize this alert properly?’
- ‘Can I also ingest this data to help me visualize, then contextualize – hence improve my predictions & notifications?’
- ‘Who else needs to know about this information – workshops, depots, logistics, scheduling, contracts & procurement – should we be writing our contracts differently?’
- Finally, how do I monetize this Solution properly – get ROI?
Once you have the ‘technology’ – monetization requires an analysis of the current process (as-is) and the potential required changes (the to-be) are in place to ensure that, not only are the people but also the processes are available and that both are ready to implement the required changes.
Typically, for business process change, Predikto recommends baby-steps initially, keep the changes small, manageable and track-able. Perhaps employ a business process change consultant to help with the smooth transition from as-is to to-be!
The changes are likely to be measured over longer periods of time (remember it is not a ‘flick the switch’ activity) and those processes need to be sufficiently flexible so that changes may be made ‘on the fly’.
And, very often, there unforeseen benefits that arise through making such changes, and are often difficult to ‘monetize’ accurately, example below.
A locomotive breaks down pulling a cargo, blocking a track. Not only is there the issue of fixing or recovering the loco, potentially replacing the broken loco or pushing the entire cargo to a siding for loco change, or renting a new loco, etc BUT the track upon which the cargo has been siting for several hours is also blocked. So rail operators do not record these ‘lost loco’ opportunities and almost all do not know the exact cost of the ‘loco(s) behind’ either having to be rerouted, cancelled, delayed, etc.
So for Customer Success – as a customer – please follow us on a journey with your data to make your business processes predictive. We are the experts in our business, you are the experts in yours but together this journey could be AWESOME!