Part 4: Ai Marketer – Profusion’s response to AutoML
Emma Woodward, Senior Strategy Director at Profusion.
Welcome back, to the fourth and final part of our Ai Marketer (AiM) series.
To briefly recap, this is Profusion’s response to AutoML – predictive segmentation models designed to supercharge your marketing strategy.
And remember, AiM integrates directly into your preferred environment to enhance segmentation and, ultimately, increase retention and sales.
I invite you to review the pervious instalments here:
Last week we looked at how the AiM predictive algorithms will improve targeting and drive more sales for your business through the Customer Churn, Lifetime value and Propensity to buy models. This week we will turn our focus to the benefits of AiM in terms of our email health and overall engagement.
- Send Time Optimiser (ST0)
Optimise individual send times for your customers across all your email initiatives: BAU, behavioural triggers and lifecycle campaigns.
Our STO reinforcement learning algorithm will tell you when the best time is to send customers an email at an individual level. This makes sure it lands at the top of their inbox when they are most likely to engage with it – be that 6am or 7pm. This increases your engagement and conversion rates and maintains or improves your deliverability as a brand.
About this model: reinforcement learning is an area of Artificial Intelligence concerned with how ‘bots’ can maximise return. This superior technique has been used for many years to train automated bots to play chess, videogames, and test self-driving cars.
We have adapted this model for our very own Send Time Optimiser to enhance the probabilities for each customer engagement state with excellent results.
A major retail bank saw an increase in engagement using our Send Time Optimiser bya 1/3rd to 42% open rates.
- Inbox Accelerator
Our next generation email delivery technology protects and improves your sender reputation.
This AI email delivery solution ensures emails can be sent faster while making them more likely to reach the inbox. Its core objective is to achieve faster delivery of email to your best and most engaged customers first.
Speedy delivery to highly engaged customers naturally builds a good sender reputation with the mailbox providers who will place emails into the inbox and not the spam folder.
This means that the ‘email is less engaged’ customers can piggyback off the good reputation established from your most engaged customers and ensure an inbox delivery, even if they haven’t been opening emails from you for a while!
Not only does Inbox Accelerator support your best customers and ensure the less engaged receive their emails directly to their inbox, but it’s also a perfect algorithm for warming up a new IP super quickly and the perfect strategy to repair a damaged IP’s reputation.
A large catalogue retailer suffered a spam folder crisis due to bad list hygiene. Profusion successfully reestablished a good sender reputation using the Inbox Accelerator method in just 4 weeks when a normal recovery period can take up to 6 months!
- Frequency Capping
As your email 121 relationships grow, customers will naturally fall into multiple selection segments to receive a variety of different, yet important messages, from you.
This personalised experience is too important to be tarnished by sending them too many emails and yet tracking and controlling the number of emails you send to every individual contact can be complex.
Email send frequency capping allows you to set the maximum number of emails every customer should receive in a specific timeframe and most importantly it will be based on a hierarchy of message importance and can also be overridden at any moment.
For a major retail bank, a combination of 2 x BAU following by 1 x service message per week was optimal in terms of maintaining good engagement (minimal opt-outs) with the brand.
- Email Engagement Prediction
In the same way we predict customers at risk of churn, we will adapt the algorithm to prevent email churn, keep your email database clean and healthy in order to support and maintain your great sender reputation.
As described above, the churn algorithm looks at an individual customer purchase cycle. The Email Engagement Prediction will do the same but the key metric for this model is an email open. The model will analyse customer engagement patterns at an individual level and when a customer drops out of their normal pattern, we will test an email win back approach and roll out the winner.
Sales churn prediction is a fundamental part of any marketing strategy. However, this clever move to adapt it in order to support other metrics – in this case ‘email opens’ – will trigger an incentive at the right time to keep ensure customers continue to find value in receiving your emails.
A simple thank you and reminder of the benefits of banking with a major retail bank successfully decreased their email unengaged customers by 13% over a period of a year.
To summarise, AiM is the (initial) Profusion response to the exciting growth of AutoML and offers tremendous potential to accelerate your return on data investments with easy installation and ongoing support – you will rapidly see your customers in a new light. AiM frees up resource to creatively focus on improving customer value, engagement and retention so we recommend you make AiM part of your marketing team today…
I plan to turn this series into a single guide, if you’re interested in receiving a full copy, please let me know: email@example.com