7 Tips for Transitioning Data Insights into Business as Usual

“Progress is impossible without change, and those who cannot change their minds cannot change anything” George Bernard Shaw

The Irish playwright was on the money – no matter what way you look at it, attacking an old problem with a new solution is going to require change. In particular, attacking an old problem with advanced predictive analytics is going to spur an array of changes to business processes and with this, a necessary change of perspective. Staff will now be tasked with using the results of models to prioritise who they target with marketing materials, who they call and which products they promote to them.

In times of transition, staff can feel overwhelmed, confused and frustrated if communication is poor. In fact lack of staff buy-in is one of the top reasons for the failure of newly adopted predictive analytics solutions. Consider the following scenario (it might be all too familiar) –

Charlie, the Head of Innovation, had been tasked with finding a new way to boost sales and drive efficiency in the marketing department. With the pressure of the management board to use the ever dwindling budget to get major return on investment, he hired Angela to head up the new Data Insights division.

Angela, an expert in handling big data, has built an impressive customer churn model to get the ball rolling. She fired a quick email over to Dave in Account Management to get him up to speed with the new scored lists of customer to target.

Dave opened the attached document from Angela and his heart sank, “decision trees… boosting… contingency tables…” – none of these words meant anything to him. It was only last Tuesday he found out that the department would be required to use a new system to determine which customers were likely to leave the company. Dave didn’t know how he was going to fit this in around his already incredibly long call list and quarterly reports.

It’s not a pretty picture – anxiety, uncertainty, limited communication.

Each party must be informed, bought in and committed to the new system for it to be fully effective. But how can this be achieved? How can the sting be taken out of change management? The following are seven tips to enable proactive engagement with change and to bypass stress at the outset of a predictive analytics project.

1. Start Small, Talk Business

Has the company used predictive technologies before? What is their appetite for innovation? If the answers are no and small, consider the speed with which you plan to make operational changes, the scale you adopt for the first model roll-out and even the impact of buzzwords or jargon during discussion. When tailored to the audience, a clear presentation of the new approach and its benefits will go a long way to mitigate most fears and concerns. 

2. Include the business from the beginning

The buy in from department heads and those who will be implementing the results is absolutely key for success. Change is frightening – if key staff are included in the decision making and planning for deployment their greater sense of control will make them much easier adopters of any changes to the way that business will be done from now on.

3. Identify obstacles

Without fail there will be obstacles, which if left unscoped or unaddressed, can derail months of excellent analytical work. The best way to identify major roadblocks is to engage with the experts in your teams – the analysts, senior management and staff who will use the final results. Some will immediately bring up known data quality issues, others will stress the importance of not only identifying the most disgruntled customers, but the fact that a retention pitch that does not address their specific pain point would be disastrous. Time spent at the outset anticipating risks will pay dividends during the deployment phase of a project.

4. Make a clear deployment plan

Deadlines for deployment can bring with them a sense of anticipation – or in some cases dread! A clearly outlined plan before project kick-off is essential for co-ordination between teams and avoiding deployment during an unsuitable time e.g. during the Christmas rush. The project plan will need to outline dependencies, contingencies and time to upskill key staff for the most effective passage through deployment.

5. Keep the analysts involved

As in Angela’s situation above, analysts are often scheduled back to back on projects with little room to check in with the deployment of their model until it has to be updated and refreshed. Scheduling time for them to check in, assess pain points and learn of any unforeseen obstacles once the project has ended could be just the thing to keep the deployment strategy on course. Even if model insights are right on point, when they are misunderstood and used ineffectively sales targets will not be met. This could be the death of predictive analytics in an organisation.

6. Prioritise good communication

Openness, clarity and trust are all essential for effective communication between stakeholders in a predictive analytics engagement. Each party must understand the power of their contributions to the success of a project and its roll out in an organisation. This may take the form of the clear outlining of obstacles (as outlined above), the definition of roles and responsibilities and the most efficient use of the modelling insights. This is where well-chaired meetings and documentation tailored to the right audience will allow all parties to understand the key processes and allow a platform for discourse about best practices going forward.

7. Know when to stop

Good is good enough – don’t wait until models are perfect to deploy the results. It is often tempting to get it ‘exactly right’ the first time, but this can be at the cost of lost momentum. Go for quick wins now and opportunities to improve model accuracy or to broaden the scope will likely present themselves down the line.

What it really comes down to is partnership. Predictive analytics is not a solo sport. For a data driven solution to begin well and take root in an organisation it requires the skills of a wide array of staff. The Charlie’s will see the vision, the Angela’s will bring the analytics to life and the Dave’s will put the insights to work. With a multi-skilled team on board to tackle the changes that inevitably come with the deployment of predictive analytics the results can be powerful, and the possibilities for the future broaden exponentially.

Version 1’s experienced consultants are on hand to help you understand your SPSS needs – from consultancy and training to finding the best software and license type for your analytical and usage requirements. Contact us to discuss your requirement and identify the best SPSS solution for you.