Customer Analytics Gone Wrong – Eight Common Mistakes to Avoid When Deploying Customer Analytics Models

Designing customer analytics models is only half the battle. Equally, if not more difficult, is deploying them, such that actions triggered by the model outputs are being taken on a daily basis. In this follow-up article, we highlight some of the most commonly made mistakes that prevent companies from succeeding at deploying models…

As stated in a previous article, we found that more than 65% of customer analytics models companies in Turkey designed (or had a consulting company design for them) were never deployed. The amount of time and resources wasted on the design of such models is only one part of the loss here – more disappointing is that the wealth of opportunities such models would have presented to the companies was never tapped into. Millions of opportunities to cross-sell to customers, to up-sell to customers, to retain customers, all never acted on. Add to this that those companies which failed to deploy the models have likely lost confidence in the concept of customer analytics for good; all in all the negative impact of failing to deploy is significant.

When looking at the reasons why companies fail in deploying segmentation, retention, next best activity, and similar customer analytics models, we see a common set of mistakes they make, errors that provide ample reason for detractors to push the models onto a shelf. To help companies avoid making these mistakes, we highlight eight of the most commonly made ones in this article.

Around Ownership & Sponsorship

Often the single key reason why models fail to deploy is the lack of ownership around them, with no one group of individuals responsible for driving their uptake across the business. Moreover, a lack of a sponsor (ideally at the CMO level) creates a situation wherein no one feels responsible for utilizing the models (and no one is told to be responsible).

The Business Intelligence unit feels their area of responsibility is around designing and updating the models; the Marketing Department feels they are swamped with their day to day business – in such environments data mining models get ignored. With strong sponsorship at the C level along with a clear set of roles and responsibilities and targets around the models (explained below) such barriers that prevent models from being fully deployed can be overcome.

Around Target Setting

Failure to set targets around what is expected to be achieved from the utilization of the models is another commonly made mistake we see companies make. Proper deployment of marketing analytics models will impact revenues positively, be it via increase in acquisition, increase in cross-sales / share-of-wallet, or decrease in churn. This impact, once estimated, needs to be assigned to responsible teams and individuals, such that their focus area includes ensuring the impact is realized. To that end, model-related targets need to be linked to potential bonuses of employees; this is the single most successful way to ensure models are fully deployed.

Around the Offers

Lack of customized offers to drive consumers to take the desired action is another shortcoming we often see around model deployment. Simply knowing who is likely to churn, or who is likely to accept a given proposition is not enough – what is required is a customized offer to drive the desired action. In the case of churn at an operator, it’s free or discounted local & international minutes, SMS, data, handsets, privileges, etc., designed and available to offer to the customer when conducting proactive retention efforts. Around selling new products and services at a bank, it’s fee waivers, trial free periods, lower interest rates, repayment period extensions, etc., – sweeteners, if you will.

Just telling a customer to buy a new product or service or not to sever their relationship is not enough; the right offers need to be designed to match what the client wants or needs, requiring the Marketing business unit to actively support such development efforts.

Around the Channels

Another shortcoming around the deployment of models is around the utilization of all available channels for conducting the model-driven campaigns. What we often see is that companies rely too heavily on using SMS to cross-sell, up-sell, retain, etc., with a sprinkle of outbound calling thrown in. These are the channels most commonly used for conducting traditional mass campaign communications, and all too often marketers stick with what they know, what they are comfortable with.

Channels barely utilized include auto-dialer, inbound contact center, inbound dial-in, shops / dealers / branches, self-care portals, and auto-dialers. Rare is the opportunity to get to interact with a customer; a company should capitalize on each opportunity to make a pitch. As such, companies need to work on including some (if not all) of the above mentioned channels in their plans around contacting customers to take model-related actions.

Around the People

Selling is not easy; it is a skill that is traditionally obtained through lots of practice, with a great deal of falling and failing involved. Dealing with rejection over and over again is difficult, leading to many people who take a shot at selling to give up and walk away. Selling to irate people is even more so, such as is the case in terms of those contact center agents who try to retain disgruntled customers.

Companies often look for the low cost solution when it comes to building their retention and sales inbound and outbound desks, sometimes outsourcing it to offshore contact center agencies, sometimes building it themselves but hiring employees at minimum wage. In almost all cases, training is underutilized, guidance and monitoring is minimal, incentives are lacking, as are systems (with manual processes in play). The success (or lack thereof) of contact center efforts to grow or retain customers is tied at the hip to the value and importance placed on the contact center.

Around the Pitch

Even one key word said at the right time in the right manner can trigger a sale of a given product or service, can trigger a successful retention of a relationship. A great example from an engagement we had with an operator in the MENA region – letting the customer who intended to close his or her fixed line account know that in the case of a natural disaster (such as an earthquake), that the fixed line network almost always stays on line vs. the mobile network. This simple script being used alone resulted in a 10% decrease in reactive churn rates.

Companies too often fail to put the time and effort into making sure the right messages are used with the right customers, at the right time, via the right channel. Optimizing when a message is conveyed to a customer of a specific customer segment at what time with what words will significantly affect the outcome of the pitch, be it a success or failure. From contact center scripts to SMS, companies need to pilot all their communications across all channels to optimize performance, to maximize ROI.

Around Performance

Failure to measure performance is another short-coming we often see around the deployment of analytics models. Too often, growth and retention target lists are acted upon, but not measured, to understand not only the success around actions, but more importantly, around understanding the impact on the bottom line. Lack of measurement often results in campaigns being conducted that should in fact be cancelled (due to their failure to achieve a positive ROI), leads to a failure to recognize exceptional or poor performers (in the contact center, for example), as well as prevents efforts from being optimized (shifting customer contacts to the most successful contact channel, for example).

Companies need to measure a variety of aspects around analytical models when they are deployed – these areas include, but are not limited to model accuracy, channel conversion rate, agent performance, offer effectiveness, timing effectiveness, pitch effectiveness, etc.

Around Results

One final area to mention that companies often fail around is the sharing and celebration of results from the deployment of analytical models. Rather, the effort of companies to grow and retain customers often becomes one that takes place behind the scenes, below-the-line, not communicated or shared with employees of the company. As is the opening of a dealer or a branch often the cause for celebration in a company (with a formal communication made to the employee base, a ribbon-cutting ceremony, etc.), the success realized from analytical models being deployed should also be recognized.

Outstanding performers in the contact center should be rewarded for their performance; milestones around reducing churn should be announced via email; an uplift in revenues attributable to a specific campaign launched thanks to an analytical model should be shared – failure to give recognition to the impact the deployment of analytical models has on the bottom line will ensure they remain something that is an after-thought in the employees’ minds. Not everyone here has to be communicated with necessarily – key stakeholders who may be showing resistance to the models may be the primary targets (i.e. segment management that is yet to buy in to below-the-line methods).

We believe that if companies take notice of the areas we have advised around in this as well as the prior article in this series, they are likely to generate a significant ROI from the deployment of analytical models. 

Tags : agents, analytics, business intelligence, call center, channel, customer analytics, customer value management, data mining, intelligence, marketing, marketing campaigns, retention, sales, segmentation