How to Maximise the Benefits of a Decision Engine for Marketing Activities

Marketing and POPIA

With the introduction of POPIA (Protection of Personal Information Act) in July 2021 in South Africa and the prevalence of similar legislation globally it has become crucial for any business to target customers in the most effective and efficient way in order to optimise marketing activities and marketing budgets. A very effective way to implement marketing campaigns is by using a decision engine which will be discussed in detail in this article.

Decision Engine is here to assist

To start exploring the benefits of using a decision engine for marketing activities, it is important to first understand what a decision engine is in comparison to the traditional campaign spreadsheet plans.  In layman’s terms a decision engine is a strategy automation tool that can be used to implement scorecards, rules, decision trees and matrices to automate campaign selections using a large variety of criteria and data.

Optimise plans in order to optimise results

During the planning phase of any marketing campaign, a well-defined and thought through process and campaign plan should be drawn up to optimise available prospects and marketing budgets. To ensure sustainable campaigns, it is important to target the right customers at the right time through the right communication channel.

Balanced outcome

Marketing activities/campaign plans are scheduled for a specific period (usually 12 months or for the new financial year), budgets are allocated, and response rates are predicted for each campaign.  In order to maximise budgets and to ensure that response rates remain stable and sustainable over the full campaign period, it is advisable to have a balanced selection. Marketing managers should not skew the population to only market to the customers that offer the highest potential profit as this will affect the volumes left for the other campaigns, the response rate, and the capacity to fulfil on the campaigns. This will ultimately impact the budget and the achievable ROI (Return on Investment).

A full features decision engine will help organisations to distribute the available prospects between various campaigns to ensure a consistent result and ROI, without exhausting the prospect base.  A variety of campaigns can be processed at any given interval (daily, weekly, monthly) and may consist of a blend between new to file prospects, previously targeted prospects, or marketing activities to the existing customer base.

When creating a well-defined marketing plan, marketing managers should consider factors such as:

• Budgets – target the most profitable, and highest propensity to respond customers, while taking into consideration that response rates will decrease over time when prospects are targeted multiple times.
• Risk – target customers that will pass the credit decisioning criteria and select the best paying customers to minimise application processing costs for prospects that will be declined, thus reducing bad debt and provisioning costs for the accepted population.
• Retention – target customers who are more likely to remain on the books and will be profitable over time.
• Sales Targets – align marketing activities to set and achieve sales targets (taking things such as seasonality into account).
• Communication channels – target customers through the most profitable but also their preferred communications channel (SMS, email, outbound calls, etc.).
• Channel capacity – align marketing activities and expected response rates with the available capacity of each communication channel in order to fulfil and optimise each campaign.
• Contact Intervals – there are often contact interval rules in play that determine the minimum time between repeating a message to an individual, or the number of messages an individual should receive over a rolling time period (often called the resting period). This will avoid over-solicitation of prospects and protect the company brand.

The ROI for each campaign should be calculated upfront using the following formula:

(Propensity to respond) x (value of the response) – (cost of the communication).

Target the right customer, with the right product, at the right price, at the right time, via the right channel with the right call-to-action.

Taking all of the above factors into consideration, it is easy to see that optimising marketing activities can be a very complex exercise if done manually and constructing these decision flows on general purpose tools (such as a spreadsheet). Decision engines are designed to help make these selections as seamless and automated as possible.

Self serve or managed services decision engine?

Companies who offer decisioning engine software usually have two options for marketing activities, namely a self-serve option where the marketing managers plan and configure the process flow themselves or the managed services option where the decision engine software provider will assist with the design and do the full configuration on behalf of the client. The only difference between these options is who will be responsible for the full end-to-end process flow (from planning to fulfilment).

How is a decision engine used in the automation of these marketing activities? The following process flow is best practice and used in many businesses, especially in the financial services and lending industries:

1. The integration starts with discussions between the vendor and the client (marketing manager) covering qualifying questions to specify particular criteria that will be applied in the decisioning process for the selection of leads. These questions are used to include specific target markets such as:Age (exclusions can be applied).
   • Gender (this can be useful if different marketing content or creatives are being used).
   • Geographical (useful if the call to action is to visit a store or delivery of goods is restricted to certain areas).
   • Estimated Income (minimum income criteria can be applied or it can be used to calculate affordability and the likelihood to be approved).
   • Payment Profile (there are various fields that can be used if a credit bureau is used during the decisioning process to control risk).
   • Historical campaign data – a decision engine can use the outcomes of previous marketing activities as feed to build or refine models, strategies or exclusion rules which can improve the response rates and application success over time.
   • Existing customer database – this is used for cross-selling, up-selling or ‘next best product’ campaigns, or for selection exclusion if the marketing activities are focused on acquiring new accounts.
   • General exclusions – these exclusion rules will be applied to the available population at the beginning of the process to eliminate prospects who do not match the criteria as discussed in the points above.
2. The next step in the process is to assign different portfolios. A portfolio designates a group of records or applications which share a common product or source characteristic.  In insurance industries for example, the portfolios could be split between customers who already have vehicle insurance versus customers who do not.
3. Strategies are now assigned to each portfolio and business owners can decide to run A/B testing and allocate a certain percentage of the available prospects to either the champion or the challenger strategy. With a decision engine various different A/B tests can be run simultaneously and the best performing strategy can then become the champion with more tests to keep refining policy rules and/or models.
4. If the marketing campaign is to offer a lending product, then policy rules will be applied to each strategy. Policy rules can include criteria such as estimated income, staff, age, etc.  Policy rules can also be split into pre-bureau policy rules and post-bureau policy rules, if a call is made to a credit bureau to obtain more payment information and is used during the decisioning process.
5. Scorecards are now applied (risk and/or response scorecards) and the relevant scores calculated. These scores are typically used in what is called a dual score matrix and risk-response grades are applied for different score band combinations.  These risk-response grades will be used in a decision tree where different action plans are assigned. The risk-response grades can also be used to assign different credit limits or pricing/fees (risk-based pricing).
6. Optimal communication channel decision trees are applied to assign the prospects to different communication strategies such as SMS marketing, email communications or outbound calls. This is usually determined by the contact information available for each prospect, the age of the contact details (if available) and for previously targeted prospects whether it was delivered successfully.
7. Once the communication channel has been assigned with the relevant message (personalised or generic), volumes can then be selected according to the 12-month campaign plan, resting periods assigned for future communication and the various campaign files extracted and sent to the different platforms for execution. File formats for the different platforms are set to ensure easy and seamless extraction and execution of files.
8. Probably the most important phase of any marketing activity is the tracking and reporting which should feed back into the decision engine to refine the models, exclude prospects whose contact details are outdated due to failed SMS messages, bounced email addresses and invalid telephone numbers used for outbound calling. Since the inception of POPIA (or other personal data protection legislation) it is a requirement to flag any customer who has unsubscribed from marketing activities and ensure that they are excluded from future campaigns.
9. When making use of decision engines for marketing activities, reporting can be automated and the results can inform whether to make any changes to the exclusion rules, strategies (A/B tests) and whether to include different score bands (risk-response grades). More prospects could be targeted If the risk and/or response rates justify including higher risk or lower response grades.

Conclusion: 

It is evident from the process flow above that decisioning and segmentation can be very complex and potentially error prone if not automated or making use of a decision engine.

Decision engines usually have audit trails for easy reference to track changes made to each campaign. They will also produce reports to track performance and compare the results to the initial ROI calculations to determine whether a campaign was successful or not.

In an ever-changing environment it is advisable that marketing managers consider making use of decision engines, weighing up the cost of such a decision engine versus the benefits. The main benefits of a decision engine are the ease of making changes or assigning A/B testing strategies.

As mentioned previously, marketing activities can be a very complex process and the ability to perform A/B testing in an easy and structured manner is of significant value.

Some initial test strategies could include the following:

• Testing a new demographic of customer with a different offering
• Offering preferential terms in order to incentivise response from better risk customers
• Fine tuning the risk-response cut-off to maximise returns

These are only a few examples and there is no limit to the number of challenger strategies that can be performed to optimise each campaign. Each round of A/B testing should have clearly defined success criteria, so that the organisation can easily track the success of each strategy tested.

Subsequent A/B testing generations can build on the learnings from previous generations, continually improving the expected results from marketing campaigns. All of these actions are easier to configure and control using a decision engine.

We have referenced The POPIA Act previously when we wrote about ‘How to grow and secure your customer base under the new POPIA regulations’, Part 1 and Part 2

Part 1 1 in this series focused on new customers acquisition within the context of the new POPIA/GDPR regulations. This second article addressed how companies should optimise their marketing strategies for their existing customers.

About ADEPT Decisions

We disrupt the status quo in the lending industry by providing lenders with customer decisioning, credit risk consulting and advanced analytics to level the playing field, promote financial inclusion and support a new generation of financial products.