Thomson Reuters tackles marketing email fatigue with AI

Information conglomerate, Thomson Reuters has implemented AI to help combat emailing and marketing message fatigue.

The multinational conglomerate implemented AI tools from Oracle Eloqua to assist with operational and campaign processes to create improved marketing decisions, campaign engagement and ROI.

In the last year, Thomson Reuters announced a strategic initiative to add an additional investment into AI of US$100 million covering current products and also spend into upskilling the AI knowledge for internal employees through internal digital learning platforms and development.

Speaking at a recent Oracle CloudWorld event in Las Vegas Amanda Chatkin, manager for marketing operations at Thomson Reuters told the audience the company has since seen a direct increase in open rates and engagement following the implementation of fatigue analyses.

Regarding fatigue levels, its contacts “have moved from being oversaturated to either ‘saturated’, ‘just right’ or ‘undersaturated’.”

One key learning for the team through its fatigue analysis work, was that “fatigue is not send volume”.

“A contact can receive 10 emails every month and actively engage with the majority.

“Conversely, a contact can get three emails a month for several months and engage with none of them. This non-engaged person is more likely to be oversaturated.”

Chatkin further explained that a contact engaging on a semi to regular basis with emails will range from “saturated to just right to undersaturated.”

“My first approach fatigue analysis, we started broad and then more targeted,

“We were extremely specific on what newsletters to include and what email groups were being utilized.

“We educated internal teams on what they needed to be aware of, such as decreased list counts. We also created a plan to share results so the entire marketing team would be in the loop,” Chatkin said.

During the talk, Chatkin said that despite being a “digitally mature marketing organisation” it still faces “challenges persisting in our current work” leading to AI solutions.

“The first problem was international email sending. As a global company, we are sending campaigns across multiple time zones, even just within the United States,” Chatkin said.

Instead of creating multiple campaigns to activate in different time zones or other processes, “we wanted to simplify the campaign creation process”.

Chatkin said the challenge of over-emailing and marketing message fatigue stemmed from “a lack of engagement with truly relevant marketing emails”.

“This was due to our customer and prospect inboxes being clouded by irrelevant messages.

“In conjunction with this, we were also seeing inflated send lists as an organisation,” Chatkin explained.

“We worked across marketing to reduce over-emailing in a fully automated way. Our goal was to exclude over-saturated context in order to increase deliverability.”

According to Chatkin, the company wanted to send the right message, amount and timing to its customers.

However, it proved challenging to know which “contacts are ready to engage more and which are already fatigued from too many emails.”

“Fatigue levels matter because repeatedly emailing contacts that are not engaging with your content impacts your email deliverability.

“It can lower your email reputation and decrease inbox delivery rates.

“Fatigue levels are calculated by the number of emails opened and the number of email messages sent to them in the last 180 days. It also considers the recency and the frequency of emails sent to opens.

Chatkin while the fatigue model is refreshed each week, the send volume and engagement numbers may not change fatigue levels between model updates.

“There are nine possible fatigue levels. The first is inactive and there are a few different definitions of an inactive contact.”

This could mean a customer hasn’t received an email in the last 45 days, lacks an established engagement history, or has been assigned the default value.

“Next is under saturated, which means that emails have been sent to this contact but there is not enough activity to classify them at a different fatigue level.

“The next fatigue level is ‘just right’ and this is the optimal level of engagement.

“In an idealistic perfect world, a pure and tighter database will be classified as just right. And then we get into saturated, broken down into low, medium and high buckets.”

“Finally, we have oversaturated, which is low, medium and high. This means that there is low engagement that could lead to possible disengagement,” Chatkin said.

From here, the fatigue model is applied with globally subscribed contact or those without emails are assigned a fatigue level.

“Fatigue levels are contact fields. They can be used in dynamic content segments, campaigns, decision steps, program steps and 'leave scoring' as well as insights for custom reports,

“We have created shared filters to add it to segments, and have also been using decision steps on the campaign campus.”

Chatkin added that prior to assigning fatigue levels, ‘frequency filters’ were used however, these only enable nine sends a month.

“In early AI, we continued to do nine sends a month, but excluded all over-saturated contacts.

“We used to rely on email counts, but now we solely use the fatigue level to dictate what our customers should be emailed.

“Currently, we use the fatigue level to drive how frequently we email a contact.”

With the refreshed approach implemented at the tail end of 2022, the company can better spot how weary its contacts were from receiving emails, according to Chatkin.

“With almost two years of data for this specific segment … we are able to successfully decrease the amount of oversaturated active context,” Chatkin said.

Kate Weber travelled to OracleWorld and NetSuite SuiteWorld in Las Vegas as a guest of Oracle.