Making a real-time decision, and presenting each prepaid customer with their individual Next Best Offer for their next purchase

THE CONTEXT

We all do it, don’t we? When we are shopping for something we need in our daily lives, we’re always on the look out for the best deal that will give us the most value for the price point we’re willing to pay. Is this any different when a Communications Service Provider’s (CSP’s) Prepaid customers are about to spend their Top-up credit and are looking at a list of options of Data, Voice and Messaging bundles?

In the Prepaid segment, customers purchase their bundle subscriptions in a variety of ways. Multiple options of different Data, Voice and Messaging volumes & validities are made available at different prices, as well as new permutations of bundles that include free data access for certain social media or games apps. It is becoming a hard choice for the consumer, especially with the competitive marketplace forcing CSPs into continuous change and release of new packages at new price points.

The timing as to when each prepaid customer might buy their next bundle is primarily driven either by the expiration of the validity period or the full depletion of the included volumes of their current package, but it is also impacted by cashflow of those living pay-check to pay-check, through forward planning of some customer being aware of future volumes they will require in an upcoming period or simply through habitual routine. Though this combination of permutations, it can be hard to anticipate what is ‘normal’ behaviour of the customers, which of them are ‘influenceable’ and what the right times and channels are to reach them with an up-sell offer.

As is best-practice, when measuring against statistically representative control groups, the trend in the results of push-campaigns promoting a particular next bundle purchase, is highlighting that a reducing number of customers are influenced by offers that are pushed to them by SMS or email, so what other ways are there to use Customer Value Management (CVM) practices to maximise the lifetime value of our customers?

IS THERE A ‘RIGHT-TIME’ TO SEND AN OFFER?

At Evolving Systems, we believe the answer to this is that the best time to make a CVM up-sell offer to a customer is right in the moment that they are there in the channel, engaged and making their purchasing decisions.

By all means, triggered SMS messages to remind customers that their bundle is expiring or that they have just a few megabytes of their data left, are great to have as part of a customer journey…but when attempting to supplement these informative messages with a specific bundle offer, it generally reaches the customer at a time when they are more concerned about what they have left, rather than what they’ll buy next.

The famous up-sell case study is that of Amazon, where it is claimed that 35% of their sales happen via the recommendation engine (source : McKinsey & Company).

The Amazon recommendation engine does not send you emails or SMSs with specific offers, but rather it takes advantage of the moment you are engaged on the site and have made your purchase choices. At this point it contextually makes other relevant product recommendations. Personalised emails and/or triggered push notifications are used only to drive the consumer to the purchasing channel, where multiple offer options can be presented with different incentives.

DRIVING CUSTOMERS TO DIGITAL CHANNELS

CSPs are increasingly aiming to promote the use of their digital channels and are encouraging their customers to top-up and to buy prepaid bundles in their Apps and Self-Care Portals. This is understandable as increased digital channels utilisation will lead to many business benefits for the CSP :

• Direct purchases in Self-Care channels reduces commissions paid through the dealer network.
• Online purchasing reduces the costs & logistics in the production, printing and distribution of scratch-card stocks linked to particular bundles.
• The customers can purchase whenever they need to without physically getting to a retailer, leading to more frequent and timely transactions.
• The use of the digital channels enables the collection of valuable customer data points.
• The CSPs App downloaded on the customer’s device opens up a new direct communication channel to market a variety of other products and services to the customer.

This initiative can be hard to get right though. The App must provide the customer with a good reason to download and keep it as one of their ‘go-to’ apps. Evolving Systems have supported several CSPs to improve this important App-downloads KPI, which we achieve through the use of Gamification Techniques in our rewards programs. “Gamification” is the application of game-design elements and principles in non-game contexts. We always advise applying such elements when designing any Loyalty Program, but in this use case, we can apply similar concepts to the CVM bundle purchasing process. For example, a fun and engaging reveal mechanic such as “Shake-to-reveal” can boost the customer experience, participation and therefore incremental revenue generated by the uplift that the personalised Next Best Offer can bring.

“Just For You” is a Digital Engagement use case that addresses the dilemma of the right time to decide and make a prepaid bundle offer to your customers and that introduces gamification into the prepaid bundle purchasing process. The goal is to incentivise the use of digital, self-service channels to purchase new prepaid bundles whilst using CVM principles to avoid revenue cannibalisation and generate incremental revenue from the CSPs existing customers.

Different from the ‘build-your-own-bundle’ approach that some CSPs have started to introduce, but that many believe puts too much control in the hands of the customer, “Just For You” is the presentation of a highest-ranked, individual or list of, bundle offer(s) from a marketing user-created catalogue. The catalogue can be configured to contain hundreds of different bundle offers, which are combinations of the CSPs products and services bundled together at different prices, each set up with different customer eligibility rules. The ranking and selection of the best offer(s) to make to the customer is carried out by the Evolution Real-Time Offer Manager using inbuilt proprietary and custom-embedded Machine-Learning algorithms.

The end-customer experience is in the form of a special menu that can be introduced into any customer-initiated engagement channel like a Mobile App, a Self-Care site, a USSD/UMB menu and can even integrate with your Chatbots and Smart Assistant ‘Skills’ to answer the “What’s offers does my CSP have for me today?” question on any chat channel or voice-controlled Home Assistant.

Most importantly, “Just For You” doesn’t just run on a static formula or preloaded matrix for offer-matching. The offers are decided and presented dynamically, with the decision being made in real-time by the Evolution Platform. This allows important context variables to be used as part of the decision rather than just the historical behaviour of the customer. For example:

• Has the customer recharged their main balance today before visiting the menu?
• What is the current balance of the customer – shall we present an affordable offer?
• What device is the customer using, is it compatible with 4G / 5G data packages?
• Have we shown the same offer to this customer earlier today on any other channels and they rejected it?
• What location is this customer in? Are they in a good coverage zone?

The key to the Next Best Bundle decision is the proprietary Machine Learning Algorithm that is used by the Evolution Platform. The algorithm calculates a ‘score’ for each offer that the customer visiting the channel is defined as being eligible for, and the resulting values rank order the offers into a priority list for that particular customer where the offer with the highest score is at the top of the list.

The fundamental principle of the Evolving Systems algorithm for the Next Best Bundle decision is based on:

The Propensity model represents the likelihood of the customer to take the offer. It is an integer value between 0 and 1 that is representative of the percentage chance. It stands to reason that the more attractive the offer (the bundles  with the largest discount or the cheapest price) shall have a higher propensity score whereas the more expensive offers are less likely to be accepted, but Evolving Systems have married this with the historical taker rate of the offer by customers that are similar to this customer (In the same segment). This way, the propensity score of an offer is a dynamically changing score, learning from the presentations and acceptances over time. 

The Value model represents the uplift in revenue that will be achieved through the acceptance of this offer. This is a calculation made based on the historical revenue of the individual customer using either the average, most recent, maximum or minimum amount from historical sample periods of the same duration as the validity period of the associated offer being scored. This is subtracted from the offer price to give us the forecast uplift that represents our value score. Evolving Systems can offer basic, intermediate and advanced ways of calculating this from the CSP’s data events and subscriber profile data, which is the ‘secret sauce’ to this recipe. It stands to reason that any offer that would cannibalise expected revenue would therefore have a negative score and shall be further down the priority list, from where it can be ignored by the system.

Combining the Propensity score with the Value score provides an optimal Offer Score that is a weighted balance to identify the most appropriate bundle offer that can be made to this customer at this moment in time.

The Evolution Platform acts as the ‘brain’ between the CSPs prepaid products, the usage & spending behaviour of customers as seen in the data and the channels that a CSP hooks up to ask for the bundles it should show the customer as part of a gamified user experience or purchasing menu. It is a ‘brain’ that marketing users can control and can instruct with business levers to apply when selecting the offers from ranked priority list that the algorithm provides. Here we can enforce elements like affordability by applying the rule that the offer price must be less than the current subscriber balance, or that as part of a business strategy for this last week of the quarter we want to prioritise sales of bundles that have a higher price and longer validity duration in order for earlier revenue recognition.

A major advantage of the “Just For You” approach powered by the Evolution Platform is that because it is ‘Below-The-Line’ and served as a special offer to individual customers based on their behaviours, patterns and forecasted needs, it means that the bundles used in the program need not be the same as those that are standard and listed on the website and advertisements of CSPs, which are predominantly designed to attract new customers.

Using the Evolving Systems Real-time Offer Manager’s Catalogue feature, marketers can create new bundles combining special combinations of products and services at special prices that are made available for consideration by the scoring algorithm only for certain customers or on certain channels.

This capability removes the need to set up all offers in the IN. The Evolving Platform shall only need a single charging wallet for each of the individual product/service elements of the bundle and can perform multiple provisioning commands to fulfil the offers upon purchase.

Implementing personalised Next Best Offer in the Bundle purchase process can be tuned to achieve a multitude of business objectives:

Primarily, generating incremental revenue from the program is the key objective and result to measure. Using an A/B testing model in the operationalisation of the program by the Evolution Platform, you are able to create a set-up for measurement that is using a static offer decision matrix of standard bundles and prices to a B-group, whilst the A-group are served with the real-time answer to the Next Best Offer from the PxV algorithm running against all configured BTL offers configured in the Evolution Real-time Offer Manager catalogue. By comparing the ARPU statistics over the next weeks and months between the A and B groups normalised to the same representative size, you are able to measure the impact. 

One customer using Evolution and this approach to power their Next Best Offer program has consistently measured that the PxV algorithm with a gamified Shake to reveal mechanic in the App is generating a 6% increment in net service revenue from the bundles it is selling, when compared to the customers in the B group purchasing standard offers. Not only this, but through a marketing push and appeal of the gamified purchasing process, this channel is now processing more than 25% of ALL bundle purchases made across all traditional channels, which is increasing the overall impact of the program visibly on the bottom line.