Analytics Inspiration Tour part 2: Start small, finish big

Posté le

Auteur
Anja Dhondt

Anja Dhondt

Hi folks,

We’d like to let y’all know what happened two weeks ago in Nashville. It all started with over 6.000 professional and academic attendees, different booths and an enormous amount of presentations. To give you an idea of the latter: 2-3 technical sessions per day times 80 topics per session times 3-4-5 presentations per topic (and this for 4 days!). Easy to conclude this is a lot. On top of this, Nashville hosted 5 sport events (including basketball, hockey and American football) and about 100 concerts. And this was just the weekend! Can you imagine this was a bit overwhelming for two humble Belgian analytical girls?!

The Annual INFORMS conference focussed on operational research and management science. With the wide variety of sessions to choose from, there was something for everyone. This also meant we had to come prepared and had to put together the perfect schedule… And as if this was not challenging enough, the agenda was topped with inspiring key notes and networking opportunities.

Hoping to inspire you as well, we would like to share our main take-aways:

1. Starting small can sometimes yield the biggest difference

As Gartner shows in their analytics journey, there are several steps a company can take when it comes to applying analytics and retaining insights. Yet it all starts at the bottom: get your data right, get your definitions and KPI’s straight and know what your aspirations are. After this crucial step, you can start with implementing the ‘fancy’ analytics (i.e. lifecycle management, time series analysis, game theory and even machine learning). It is important to not want to go too fast. Do it step by step, first things first.

2. Artificial intelligence needs human intelligence (cognitive intelligence)

As machines are getting smarter and can help us make better decisions on more elaborate data, including human characteristics such as compassion and value judgement will always lead to better results. We should not see or treat the two separately from each other, as both parts have their advantages and can add value to the other in order to reach the desired result. Even though we can get more and more powerful machines delivering better outcomes, the input of a human remains of great value.

3. There will always be data limitations

Data will never be perfect and will always have constraints. It is not about ignoring these limitations or running away from them but about understanding and mitigating. By putting sufficient time in exploring and moderating, they can even have an added value by bringing more insights into the business. Some limitations will always be there so ignoring them can only bring (more) harm. A perfect example of this is map data. Imagine you work for a delivery service and you have to do a delivery at a store by making use of your GPS to get to that store. Your GPS will give you the most optimized route to send you to the front door of that store. It doesn’t know there is a backdoor where deliveries have to take place. And that this back door can only be accessed by taking a complete different road.

4. Let your analytics be hands on

Usually analytics are applied by the data scientists in the headquarters of a company. And although these analytical models can be very helpful further in the organisation, they often do not get into the hands of the actual customer handlers (i.e. employees in your department store). Yet, providing them these interesting insights can help them become more successful in their jobs, as they can suggest a perfectly matching scarf with the blouse a customer is trying on. If you make sure your resulting models are workable for everyone in the company (for example implementing a user interface), you get a win-win situation: your employees feel empowered and you see your profits increasing. And on top of that, you will have a satisfied customer leaving your store.

5. Embrace change management

Your program/solution needs to be feasible, optimal and implementable. This requires often a new way of working from your employees and colleagues. It is important to foresee sufficient training, education, communication and time for your employees to accept and embrace the system. Make sure that the black box becomes a glass box. Change management can sometimes be the major challenge but also a huge opportunity so it is important to spend sufficient resources and time to manage this.

6. Current models can always be more optimized after a certain time

The above is a conclusion we ran into multiple times at INFORMS. Some examples:

  • Optimizing your category management by including reviews shows huge improvement (and thus higher sales/profits).
  • Predicting return abuse can be done very accurate by just including data on returns. However, the prediction gets much better by including sales data as well (meaning higher savings).
  • When assigning your shelf space, you should include the store lay-out as well as the product allocation (impulse shopping increases and thus profits).

As you can read, the research on all kinds of optimization problems is continuously growing and making models better, by adding existing or newly available data. Soon, predicting election outcomes will become more accurate as well! ;-)

7. It’s a small world

We hear it often, but clearly it is! Even if you are over 7000 kilometres away from home, you can run into your former analytics professor :-) Prof. Dr. Dirk Van den Poel and his PhD students were also present at the INFORMS conference, helping us to represent the Belgians at this huge event! They even presented, focussing on business applications in social media analytics.

Next to the constant love of music enthusiasts, hockey supporters and football fans, sunny Nashville also managed to inspire the nerdy last week ;-) We definitely hope to go back some day.

If you have not done so already, check out Roos & Lotte’s blog of last week, on the Big Data and Marketing Innovation Summit in Miami!

Prêt à devenir une “customer company” ?