Consumers are bombarded with hundreds of marketing messages on a daily basis and these messages are served via multiple channels. It may be email, Internet banner ads, coupons or any other combination of media used to engage with consumers. As well as these channels there are numerous other treatments to consider, such as offers, timing and pricing.

The ability to identify which combinations contribute towards an organisation creating and maintaining competitive advantage through measurement of the efficacy of its marketing initiatives is critical to success. Controlled testing is the most effective means of learning.

Organisations need to adopt a culture in which experimentation plays a leading role. Experimentation implies testing and learning!

Testing marketing has traditionally not been executed or planned in a disciplined way. More often than not SAS’s experience has seen the following behaviours:

• There is a lot of talk about testing but very little action because there is little understanding of the business reasons on why we should test
• Testing is not executed rigorously and this is often because there are not many skilled testing practitioners
• Testing has not been seen as a methodology to learn about what drives customer responses or offer and treatment effectiveness
• Traditional A/B Testing methods are cumbersome.

Traditionally when building direct marketing campaigns, marketers have used splits or control groups (A/B Testing) to evaluate how a consumer will respond to an offer but these techniques become cumbersome when you start testing more than a handful of alternatives.

Consider the following, if you were marketing mobile handsets and you wanted to test how a consumer would react to 2 price points, 3 different handsets and 2 different contract terms then using the traditional approach to marketing would require control cells and numerous test cells. This can be quite cumbersome to monitor and track and you would end up with a Campaign that can resemble something along these lines:

As you can see A/B Testing makes your campaigns more complex and also may not result in meaningful cell sizes to conduct meaningful tests.

A/B Testing also makes it harder in complex campaigns to determine which variables are driving consumer behaviour. Is it price, contract term or a combination of variables. In addition as you want to test more variables, for example, subject line and creative, then the number of test cells can grow significantly.

Given this complex environment how can we test which variables drive consumer behaviour in a more effective manner?

By taking a lead from other industries, SAS has developed an approach to help organisations design and evaluate robust marketing tests which can be used in conjunction with your A/B Testing methodology.

The following schematic shows the essential steps in a well formed and managed Marketing Testing Strategy.

The above methodology is based on the following paradigms, a test is an experiment:

• To answer a business question
• To acquire important customer knowledge
• To validate a belief or judgement
• To establish potential cost-effectiveness
• To quantify risks with a proposed approach

The testing methodology is based on the following insights developed by SAS globally:

1. Understand your business goals and problem

2. Determine the KPI to be achieved to make the marketing campaign or test successful. Is it Return on Marketing Investment (ROMI), Leads Generated or Sales?

3. Ensure your sample sizes for testing are not too large or too small. It is important to note that you may not need 10% of your customer base to test treatments, or, certain comparisons may require many more customers that is realistic. It is important to also ensure sampling and allocation of customers to cells is representative of the customer base.

4. With modern analytical techniques you can test more than one variable at a time and this is where you can leverage your investment and partnership with SAS by applying the analytical technique referred to as ‘Experiment Design’.

Your investment in SAS means that you have the technology available at your fingertips and the ability to utilise Experimental Design.

While not a new concept, Experimental Design allows you, as a marketer, to significantly increase the variables which can be tested in a single campaign and more importantly test multiple in market offers to consumers simultaneously. This allows you to see which offers and combinations of variables are driving consumer responses and ultimately improve your campaign response rates.

Typically approaches utilizing Experiment Design also reveal unexpected results around which offers or variables drives the higher responses.

Test more than a message as many variables can influence a campaigns success. Background color, subject line, salutations or any other combination of variables can be tested.

5. If the results look bad do not be afraid to “pull” a bad campaign from the market. This is probably one of the hardest disciplines to enforce but will provide invaluable benefits if done properly.

6. Perhaps the most important message in testing is the journey. You should keep testing as the lessons you learn from your tests will feed into your future campaigns.

As a marketer you strive to get better insight in to your customers and how they will react to your campaigns. A well-structured testing methodology is an invaluable tool in your arsenal so contact SAS to find out how you can improve the effectiveness of your marketing campaigns.

Campaign Execution Improvements in new Marketing Automation releases

Customers using SAS Marketing Automation would see situations where Contact History would have been overwritten for campaigns that are executed without a recurrence being set.

Consequently Contact History would be lost and the only fix to prevent this was to set a recurrence:

If a recurrence is not set Contact History will be overwritten. If a recurrence is set then Contact History will append with a new Occurrence number. This can be validated in the SAS CDM table CI_COMMUNICATION and you would see a Communication Occurrence number being set and ultimately Contact History being preserved.

This issue continues in Marketing Automation releases 5.40/41.

However, all Marketing Automation releases 6.1 and beyond have had this issue fixed. This means you can execute the same Campaign repeatedly without a Recurrence and Contact History will not be overwritten.

In addition in Marketing Automation 6.1 the user can now execute and then close the campaign and the campaign will continue to execute without results being lost. The user can close a campaign and the campaign will continue to execute in the background and also log out of CI Studio while a campaign continues to execute in the background.

Contact SAS if you would like more information on improvements in future releases of Marketing Automation and how they can help improve your teams productivity.

Information Map Filters

A common question asked by our Marketing Automation customers is how do we query multiple data records stored in rows rather than columns. While the data may have common attributes, the question remains what is the best option for surfacing the data in Marketing Automation?

Almost without fail the optimal way to solve this problem is when designing the MA data environment (your marketing database), always organise data into high-performing DBMS structures. For reasons of performance, data governance, and ease of use and maintenance for the SAS user community, solving data issues in your database is always the best bet. If you have the option, reorganizing data for end-user consumption via DBMS tables and views is almost always the right course of action.

If reconfiguring the MA data mart isn’t possible, a second option would be the use of advanced information map joins. By creating multiple aliases (logical copies) of our physical data structures in the MA Information Map, we have the opportunity to configure multiple join paths to the same physical data and to represent that data accordingly to the marketing end user. The solution works, but can be cumbersome to configure and maintain, as the number of data items.

Finally the MA information Map comes in the form of information map filters. The MA Information Map map custom properties "IsFilterItem" and "VarUsedWithFilter" may be used to generate a virtual folder structure based on row-level attributes of a non-subject table. The map administrator configures data items for each column of data to be surfaced to the end user, then a filter to correspond to each way in which we want to present that data.

This feature means if you have model data spread across a number of tables, which cannot easily be joined because data is in rows and not columns you can use filters to display information as follows in the Information Map:

It is important to note that configuring your information map in this way may have an adverse performance impact on query performance, especially when used on very large tables. As all filter expressions go in the SQL joins and not into a WHERE clause, it's possible that this type of query could produce skewed processing on your DBMS.

Be sure to test the behavior and performance of this configuration prior to putting it into production and as always this option should always be considered if you cannot reorganize your marketing database to display information correct.

Contact SAS today to find out more on this Information Map feature.