the_evolution_of_automation

The evolution of automation: A journey from A to B

We take a look at the rapid pace at which A/B testing practices have evolved, and explore the artificial intelligence software that is empowering modern marketers to embrace automation in their campaigns.

A/B Testing: A brief history

A/B testing has been an important technique in marketing for decades, allowing the best-performing creative to be determined statistically from a small-scale test, before being rolled out to the full audience. Each test takes time to run, since a particular volume of communications must be analysed before any conclusions can be considered valid.

Historically, when marketing communications were less frequent and broader in scope (when they involved sending the same product catalogue to all customers once every 3 months for example), it was feasible for A/B tests to be conducted manually.

Modern marketing involves sending more specialised communications with higher frequency, perhaps targeting specific customer segments with specific product ranges. This makes it difficult to perform A/B tests manually. It might take a few weeks to run a single A/B test, so doing this for a dozen product ranges is a considerable investment of effort. 

Segments may respond differently; for one segment A may be best, but for another segment B might be better; and what about trying alternatives C and D? The results of the test might indicate that B is best now, but for how long will this remain true? In a few months’ time, the conclusion could be different, especially if there are seasonal effects to consider.

Modern marketing needs to be automated. A marketing application needs to be able to monitor the performance of the communications being sent and make adjustments accordingly to improve their performance. This “Automated Improvement” could be seen as a form of “Artificial Intelligence”.

Artificial Intelligence = Automated Improvement

Modern customer expectations have led to exponential demand being placed on marketing applications. Multi-channel campaigns have become the norm, and marketing teams are expected to be capable of both performing and progressing in tandem. 

Achieving this means tracking campaigns with sophisticated AI tools. From the moment of launch, marketing teams need to be able to analyse relevant data in order to identify additional opportunities, or aspects of the campaign that could be improved for greater success. This is what will empower marketers and enable their projected performance figures to continually increase. 

There are reams of sophisticated tools available on today’s market that are more than capable of making campaign creation, analysis and management processes both simple and intuitive, even for traditionally-minded, tech-illiterate marketers. 

With the right technology, every marketer can be a digital marketer, capable of automating timely, event-driven, targeted and personalised campaigns using call centres, email, SMS, social, CRM, direct mail and mobile push. 

A Case in Point - PeopleStage

Apteco PeopleStage can now automatically improve the performance of campaigns through the use of Automated Alternatives. The user specifies a number of alternative content items (such as subject lines or offers). PeopleStage then monitors the performance of each over recurring runs of the campaign, and automatically adjusts the proportion of each alternative, in order to increase the overall response rate.

By simply ticking the “Automated Alternatives” option, PeopleStage monitors the performance of each alternative and adjusts the volume of each, using a “Champion-Challenger” algorithm, similar to a “winner stays on” pool competition. 

The initial run of the message is sent to all the alternatives. The best two become the champion/challenger pair in the first “test”, while the other alternatives will take it in turn to be the challenger in subsequent tests.

Ray Kirk's picture

Ray Kirk

Consultant Developer

Ray is part of the development team, but has a client facing role. His main focus currently is PeopleStage, from a training, best use and design perspective. Previously a SAS analyst, Ray's background is in statistics - expertise he used to develop the Decision Tree Module within FastStats®.