Preparing for automation: 3 practical steps.

I’ve noticed a great many articles about automation recently. They are often based on dire predictions regarding jobs that robots will do, but fail to make useful suggestions regarding the practical steps that companies need to take. Instead you're left to wonder at what point an army of sentient automatons will arrive to do your work - or if you should start talking to some IT people about 'developing an app'.

I do believe there is a logical sequence of steps to take if you want your company to exist in 2050. Here they are:

Step 1: shift from courses to resources.

Most organisations don’t know how their employees do their jobs. Literally. I know this because I spend a lot of time with organisations trying to build performance support. These organisations have job descriptions, standard operating procedures, and courses to train people in these procedures - all of which bear little or no relation to how people actually get the job done, which you only discover when you talk to people doing the jobs in question (although you probably suspected as much). My point is this: if you don’t actually know how work is getting done today in your business, you’ve got little chance of ever automating it successfully.

By building resources, we discover how people are working today; by building performance support we enable inexperienced people (or machines) to do the job by codifying capability. The truth is that most of this knowledge is currently tacit and hidden within your company culture. This is going to be a problem for you with or without robots - for example as people change jobs more rapidly, have less capability, or as you struggle to maintain competitive advantage*.

Step 2: shift from resources to guidance:

GPS is much better than a map. A map is a great resource, but you have to regularly stop and figure out where you are. A GPS tells you what to do next because it knows the map, and it knows where you are.

Once you have created a set of resources that enable someone with little or no experience to do their job well, you have already taken a big step forwards - but (as we have discovered) you are still dependent on people finding the right resource to use at the right time. This is how things stand with Google today: it’s a great resource, but you still have to look things up - wouldn’t it be great if it had enough information about you to know what you need without you having to look it up?

The addition of contextual information to resources enables the creation of performance guidance systems - like GPS - that in turn bring about a second dramatic reduction in the level of competence required to do something well.

Uber is not disruptive: GPS is disruptive - Uber could not exist without GPS.

Just like GPS, performance guidance systems don’t have to be mind-bogglingly sophisticated AI; just a few data points are enough to give leaders guidance on how to improve the performance and engagement of their teams, for example. Once you have the resources that an employee needs, that’s a good point to start looking for data that might tell you which resource, when. (Note that this is not ‘personalised learning’ which is a red herring, but ‘performance guidance’).

Step 3: shift from guidance to automation

Hopefully the progression is now pretty clear: by the time you’ve figured out the rules for what needs to be done, when, in your organisation (starting with the realisation that these are almost completely different to the rules you have in place today) you are in a good position to consider automation: for example you know what an HR-bot should say in response to 95% of inbound requests, and could share these with an app developer. You know what makes people successful in a technical role, and could program a machine to work similarly. Instead of saying 'congratulations, you're a leader, here's some stuff on leadership styles', you know what it is that a leader needs to say and do at various points to improve performance and engagement. There is of course a risk that you are replicating sub-optimal ways of working - but equally there is often a good reason why people do things the way they do today.

In summary, there are things you can do today to prepare for automation in the future: such as the creation of simple one-page guides and checklists. Often, when we do this kind of work, we find that people have already started doing this for themselves (as a symptom of the redundancy of SOPs and training) - and these resources are already circulating informally across desks and hard drives.

The starting point is this question: ‘how can we capture what people are doing today, as simple instructions?’

*Many people think that automation starts with a massive data-mining exercise which essentially surfaces implicit algorithms. Whilst this might work for Google or Amazon, it probably won't work for your company: you don't have enough data, it's too costly, and the interpretation is too complex. You'd be better off just asking people.