How to apply machine learning to assist digital transformation projects

Large digital programs usually tend to focus on the technology side of the solutions, particularly IT transformation programs.

Image by Sourajit Sengupta 

The focus is predominantly on implementations of new platforms to support the business while forgetting other important aspects, such as people, organisational structure, business processes and platforms, what is going to change and how are they going to change in response.

Attention should be given to the role platforms continue to play in new solutions. Some of the focus needs to be on the domain of the solution rather than just on the tech. For example, the business architecture (a top-level view of how a business will operate), and how new systems will operate in conjunction with existing infrastructure.

System testing for quality assurance is often an afterthought. There tends not to be sufficient focus in the early stages on how a solution will be deployed on the business IT environment.

Consideration must also be given to the organisational change in the wake of a new deployment. This should be structured and planned from the outset so that change is properly managed and becomes relatively seamless.

Only by focusing less on the technology aspect and more on the business side can enterprises successfully implement any organisational rules changes, new business processes and strategies to manage the business transformation.

If the business is not set up to manage the transformation and staff are not prepared, the whole program can be jeopardised.

Other factors to consider are more akin to standard project management practices, such as dealing with risks at multiple levels – from technology to the risk of marketing changing due to adversity in the business environment. Given the average timeframe of programs, it is more likely than not that such issues will arise at some point so it pays to be prepared.

Structuring and planning for program effectiveness

During large transformation programs, operatives are spread far and wide so incidents tend to looked at in isolation, rather than examining the root causes. There needs to be more focus on a smaller number of factors that are crucial to the project.

It is very important to have a mechanism to collect information and data that provides a view of project success based on characteristics. With a small amount of high-quality data, you can build a model to interpret new data sets and correlate with existing datasets to draw reliable conclusions into a project’s success or failure, as well as gaining a better understanding of the characteristics of a project – what have we learned and what insights and recommendations should we glean from it?

So what are all the considerations and characteristics of projects? By collecting and assessing high-quality data, through machine learning, the system acquires knowledge and becomes better equipped to provide recommendations on how to achieve success.

This means that humans don’t need to do too much. Organisations do need to invest effort in understanding outputs though and be ready to be able to accept and understand data to take recommended actions.

Use machine learning to focus on practical, measurable outcomes. It is not simply about change management but rather making sure you are using tools and approaches in a way that concentrates on what is important to achieve successful business outcomes.

Article by Michael Devlin published on IT Brief Australia

Measuring people performance with ML


Eek! A machine is going to tell me I’m not good enough?…

We hear a lot about artificial intelligence and machine learning as if they are future concepts with no practical applications. But in the project management and business transformation space there are real-world applications of AI and ML showing very positive results.

One of the challenges of assessing the effectiveness of a transformation project is measuring the effectiveness of people. ML can be applied to measuring and isolating the people characteristics of these temporary organisations that make up a project to determine if people are performing well or not.

When something goes wrong in a big, complex transformation project it can be difficult to pinpoint the exact cause and effect, but it’s imperative to determine the root cause. It is quite a complex environment to even measure these problems and that is one reason why traditional project management methodologies and tools haven’t attacked the problem.

“Implementation of targeted ML-generated advice enabled a complete project turn-around in a very short period of time.”

We are using ML for this, as part of our AI assurance approach, and it is proving very useful. ML enables us to isolate and measure the people performance characteristics of a transformation project, that are predictive of success. With ML we can very accurately determine the probability of success, to derive improvement actions.

Why machine learning? The old approach to solving the problem is to gather data, see what occurred most often and draw conclusions as to why the project failed. The problem with that method is in such complex systems trying to draw cause-and-effect it is difficult. You would need an enormous data set that is much larger than most organisations can effectively gather and process.

ML can make quite accurate predictions on smaller, but higher-quality, data sets. There is a a need to train the machine learning system and we have done this via the characteristics we have isolated and measured in our challenged project recovery work. Over time we trained the ML to learn from each project what the characteristics of success look like.

It has taken eight years to train, but now our ML system learns on its own and is very accurate at predicting success or failure and advising how to improve your odds.

The practical outcomes? We have worked on financial services projects and one client ran our machine learning system a number of times across a recovery project. It gathered environmental data about the project and ran it back though the algorithm which correlated the characteristics of the target project with past projects. The implementation of the targeted advice generated by the ML system enabled a complete project turn-around in a very short period of time. We could then provide very specific advice on how to reform or improve the project to improve the probability of successful delivery.

Transformation success or failure almost always links back to people. This is a fairly well accepted proposition – everyone knows people are a problem and how they behave impacts projects significantly. AI is the only effective way to accurately measure something as complex as behaviour in a group context. The silver bullet is finding a connection back to people being the biggest risk.

For multiple parties to come together and be successful with a transformation project there needs to be good risk management. For example, a banking project will bring together a number of big suppliers and all those parties would be managing their own program risks and outcomes.

One example of what we can do with ML is precisely measure how much these groups are sharing common risks. You can’t expect groups to abandon their own interests, but business leaders must have visibility to the degree of risk alignment in order to manage a successful outcome.

Project and transformation success is a perennial challenge and sometimes improvement is as simple as shining a light on a problem, measuring the gap and raising the conversation at senior levels. For the first time ML is enabling this conversation.

AI and ML are great examples of emerging technologies helping to improve an age-old challenge like transformation project success. I encourage you to look at how your organisation is challenged by projects and processes, and consider how a data-based solution like machine learning can be applied for better outcomes.

Contributing opinion by Michael Devlin originally published on iStart on 12 September 2018

Are people the biggest risk to transformation success?

Australian companies are undergoing a hive of digital transformation activity with some projects more successful than others. Is enough being done to review the impact of people on the success of a project? Let’s take a look at how the “people risk” of a transformation can be identified and mitigated.

Start with the building blocks

Large transformation projects are seen as any other temporary organisation, with functional specialisations, partnerships and goals. The huge difference between an organisational unit and a transformation project is the temporary organisation is formed in a hurry – sometimes in as little as three to four months – and they are put together with not too much science.

There are expectations the teams will be a high performing unit without any real nurturing or caring. Compare that with any other large organisational unit and the disparity is clear.

A CFO building a high-performance function will start by developing a strategy, designing the organisational structure and carefully selecting staff and partners. Once that is put in place it goes through a process of one to two years to get to a level of operational excellence.

Too often the characteristics of big transformation projects lack this level of awareness assembling a person-driven function. This is the highest risk part because not all organisations form transformation programs with this care and many contract this risk away by engaging with a large IT services firm.

Temporary organisations are just as strategic to a business as permanent ones and having the right people in the right role, performing at a high level is a foundation for success.

Know your differences

As with any project, a transformation organisation has different skills and cultures that have to work together to deliver an outcome.

There are differences between people, suppliers, and even sub-cultures within a transformation program, and, due to their temporary nature, don’t really have the opportunity to go through a cultural change management exercise.

Your project environment, therefore, has to compensate for these cultural differences. You have to create a temporary project culture that is strong enough to balance the individual cultures. For example, there are particular technologies, such as collaboration tools, which can help deal with that.

Why is this a big risk? The problem is nobody is talking about it, let alone measuring it. This leads to a lack of awareness of what needs to be done to mitigate the people risk associated with cultural differences. One supplier might have a completely different way of delivering the same work package compared with in-house staff. These differences must be identified, measured and managed, not swept under the carpet of the transformation program.

Consider this scenario. An insurance company will find a big heavyweight IT services player and then try to transfer the internal scope of the transformation to the supplier. In most cases, the new supplier doesn’t align or change itself to meet the in-house people engine.

None of the focus is on forming a function of people that will operate in a high-performance manner – this is crucial when you need to do a significant amount of delivery in a finite time.

By shifting the delivery to a third-party, you need to be prepared to identify and assess the differences from the outset and manage them for success.

A path to better people-driven outcomes

Is it all misalignment and the invariable doom and gloom? The good news is a lot can be done to reduce people-related risk to ensure people continue to be your most important asset. Time and time again, our clients demonstrate that once identified and understood, these risks can be corrected using standard project management practices – sometimes in a matter of weeks.

A lack of role clarity, structure and lack of accountability are all found in dysfunctional transformation organisations. When we look at a failed or poor-performing project we often find the root cause back to these problems and we can use those as a predictor of whether the project will be successful.

Most of today’s project management collateral and tools deal with communication and operational challenges, but not at the same breadth and depth as what is possible by analysing the people-related risk factors.

Our advice to clients is to think about the people stuff upfront; set yourself up for success; measure the performance, and incrementally improve.

The old cliché “people are your best asset” unfortunately gets lost in many transformation projects. People are your best asset, but they can also be a risk if they are not managed properly from the outset.

Contributing opinion piece by Michael Devlin originally published in IT Brief Australia on 24 September 2018