2021-05-19

Notes on The Analytics Enterprise

Extracted from the public Notion original. Notion record created 2021-05-19. Last edited 2021-05-19.

The Analytics Enterprise

Beyond the competency centre

Building a team of skills analytics professions is relatively easy. You just have to train them in a range of analytics and visualisation tools - or let them choose their own - and then give then access to as much data as possible.

If you build an analytics team in this way you will have some success. Once people realise they can engage with this team to develop reports, and dashboards, and provide data to regulators, and provide data to the executive team, they will come from far and wide to make requests.

Faced with a broad range of stakeholders in different business units, and with access to more data than they can understand, the analytics team will reach a crisis point. The team will have more work than they can handle, for starters. They will also be doing a lot of work that isn’t “real analytics”.

Because they are some of the heaviest users of data in the entire organisation they will be handling data that nobody else does. This means it’s not well integrated, has poor quality, and doesn’t interoperate well with other data sets. Also, when people come to them for help it’s never because things are easy - the analytics team will by definition get the difficult tasks.

The commonly quoted statistics about the amount of time analytics team spend in data preparation a result of this approach to shift difficult and unsupported activities to the analytics team. This is also the reason why analytics teams have to service data requests, reports (rather then a “real analytics”) and all sorts of ad hoc activity.

At the same time, centralised analytics teams are the exception to many IT policies and service management processes, as well having unique support requirements. Every time the organisation tries to strength the protection of their data it makes it harder for analytics teams to get the data they need to do their job. Every time the IT team try to become more “agile” it becomes more difficult for the analytics team to understand where to get the data they need, or even what data is available.

Some analytics teams are left with the job of explaining why “the business” needs certain information - and then treated like “for analytics” isn’t a sufficient answer.

So building an analytics competency centre - once the key challenge in improving an organisation’s analytics maturity - is now the cause of modern organisational woes relating to analytics.

Analytics is the elevator

We have flattened our organisations over the last quarter century - but they still have at least three distinct layers of structure. The flattening of our organisations was always a follow on impact from the increase in peer-to-peer direct, direct, unbounded communication between all individuals, layers, and business units.

If you picture an organisation with three layers you have the executive layer on the top, a management layer in the middle, and an operations layer at the bottom. Even this view of the organisation itself is challenged by those who seek to make our organisations for customer oriented - so should be supplemented by a customer orientation that cuts through all layers at will.

[Image unavailable in the original Notion source]

Analytics teams are often challenged to make their analysis more business focused, simpler, more relevant. Using only data, the right analytics and visualisation tools, and an understanding of the decisions that might be made from the analysis - these teams are challenged to do what every other team can’t.

Analytics, when don’t well, traverses all three layers of the organisation, as well as incorporating the customer perspective. Just imagine all of the effort expended to keep those layers in place: executive teams doing anything they can to delegate work to management, management leaving the technical details to others lest they be accused of micro-managing, customer service teams compensating for broken back-end processes with work around some to maintain a positive customer experience.

Analytics, when it success, is cutting through all of that. It is selecting the right data from across operations to create a body of facts and evidence to underpin it’s analysis. It is ensuring that the information is summarised and presented in a way that management can deem as “actionable”, it is including the customer perspective to provide a lens of value across all analysis, and it is balancing this with the often functional, and siloed structure of most executive teams.

We think of analytics as an elevator between the different levels of your organisation. It’s not the only way to get between levels - but its the easiest, safest, and when you use it you get the most well comfortable and well designed ride.

An elevator is also a good metaphor because elevators come back down, they are constantly in motion and they don’t just carry information up. Analytics, as the elevator, bridges these organisational layers and provides the organisation with the continuous feedback it needs to perform and improve.

Analytics team also have special access - because not everybody gets a pass to both the executive suite and the basement.

Analytics is the intelligent agents brain

When you use Suri on your iPhone you might notice some responses are provided by Wolfram. Wolfram is a knowledge and computational engine developed by Stephen Wolfram’s company. It structured knowledge and provides sophisticated mechanisms to manipulate knowledge and data.

It may not always be the case the Apple needs to rely on the services of Wolfram to service Suri requests - but for now it does. Perhaps the future is that Apple has it’s own equivalent. Perhaps their is an entire market of such services that Apple can tap into. We’ll see.

This distinction between Suri and Wolfram shows the value of separating the intelligent agent (Suri) that interacts with the user, from the brain used to think about requests.

When you think about Suri you might think about it as an individual servicing you personally, or you might think about her (him?) as a single global intelligence that is only giving the impression of servicing you personally- and this is more accurate.

You might also think about Suri as either being personally super-intelligent and knowledgeable or you might think of her as having access to a variety of tools and resources that she can access as required - and again, this is more accurate.

Analytics teams has these distinctions too. When they are providing analytics services on request they are acting like they are servicing just your request - but they are also a single function - and will take that into account to tell you things you didn’t ask for. In some organisations they have an obligation to inject a customer perspective into all of their analytics, for example.

Also, the vast repository of resources and references the analytics team builds up must be managed proactively so that the illusion of a fast, intelligence response can be presented when analytics is required.

Lastly, there doesn’t have to be a personal carefully managing each request. Your analytics team is too important for you - however important you might think you are. When you interact with your analytics team use need to use Suri - or whatever primitive approximation of a Suri that your organisation has invested to produce.

If you find it inconvenient to find your own information, insights, and data that doesn’t mean that inconvenience is sufficient reason for your to ask the analytics team to do it. You might value from a personal assistant to for you but that doesn’t mean you get extra access to precious analytics time.

If your organisation wants body to be able to get analytics without effort on their part this has to be designed into the organisation. Either the analytics team needs to be expanded, and tiered with both high-cost resources and cheaper resources that can do things that are inconvenient to others, or a full suite of discovery, push analytics, and intelligent agents must be deployed.

Analytics is middle management

Once all the data has been presented and visualised, and once the potential impacts of each decision have been modelled to predict the alternative futures, then the actual decision becomes easy. You might prefer to “use your gut”, or you might not “trust the data”, or you might not want to make a decision at all, and just sit this one out.

Humans are the most important part of an organisation (maybe). Analytics will always be considered subordinate to the people in the organisation. That’s something we can all agree on. But once we have all the information at our fingertips we need to make a decision, why do we need to deligate that decision-making to somebody else?

We used to have a concept of middle management that would manage workflow, make decisions, prioritise work, and deal with exceptions. Culturally, our organisations have rebelled against middle management. This concept of middle management doesn’t really exist in modern organisation anymore and if you reach for a middle manager for help you either wont find one, or they will no longer have the ability to help. They are caught up in other things, change initiatives, and trying to understand what it happening across the organisation in this new network connected world.

Individuals must collaborate and take ownership for their own decisions. But individuals don’t have the unit perspective that the old middle management class had. To get this perspective they need it pushed to them - through analytics.

Those who feel like the companies they work for have deteriorated, or that the rug has been pulled out from under them over the last few years, are typically suffering from the effect of organisations becoming more flat, agile, and self-directed - but without the appropriate investment in analytics to allow individuals to get the context they need to make decisions, and be effective.

This dynamic will come back to bite many organisations. They are trying to reap the benefits of dynamic, distributed workforces without putting in the real work.

Analytics is the voice of the customer

Analytics is risk management