How happy are you? AI tool scores happiness based on your emails
How do you know when your customers are happy? It can be tough- or impossible in some cases – to gauge on face value alone. Can you tell from a smile? Or a gut feeling? Perhaps you rely on customers proactively telling you or completing a feedback form.
“Only 1 out of 26 unhappy customers complain. The rest churn. A lesson here is that companies should not view absence of feedback as a sign of satisfaction.” -Huffington Post
As you can imagine as an IT services company, we tend to involve technology where we can to get our results as accurate as possible. We genuinely care about our customers happiness, and ultimately, we want our customers to succeed. Our aim is to empower and support that success through deploying technology solutions that fit and deliver real value.
So it’s not surprising that we’re looking to go a step further in our quest to understand how happy our customers are by using artificial intelligence.
Azure AI works by reading the language sent via email from our customers. It then produces a score or grade which reflects the happiness or equally unhappiness of a customer by the language they have used.
This is HUGE. It means that we aren’t just relying on a smile and a handshake to determine how a customer feels about our service, we can see exactly how happy our customers are in black and white.
Azure AI allows you to tag emails to customer accounts and projects, so for example when an email comes into your inbox about or referring to ‘project digital transformation’ TSG can tag this email to the same project in CRM. From here we can see a score which indicates the feeling about a specific project.
For the purposes of this blog we’ve used the testing platform to see what type of scores we get based on our emails. As you can see below, a positive email has given a high sentiment score.
You can see the email message to the left of the screen which has been analysed. The score is situated to the right along with the words that have been selected for analysis.
That’s great Grant, thank you very much for your help on this, we’re really thrilled with how the project is progressing. Please keep me in the loop with anything relating to the Office 365 roll out moving forward.
Now let’s not pretend it’s all rainbows and unicorns in the world of customer feedback… however an indication of unhappiness can be incredibly helpful. It means you can devote time to understanding why a sentiment score is lower than you’d like and put measures in place to improve the situation.
Azure AI also lets you tag emails to support calls or just to an account overall to see how the customer is doing.
You can determine trends from this sentiment and then action accordingly.
You can see an example of a low sentiment score below, this message was negative landing this email with a sentiment score of 8%.
I’m not happy with how the project is going, we’re seeing slow progression and we must get this moving quickly. I was disappointed to see that we didn’t meet the deadline as agreed.
The messages differ quite significantly in language and this is clearly reflected in the scoring.
How can we use this information?
Large movements in aggregate sentiment will ultimately lead to proactive alerts. This means that if the sentiment score on a project suddenly drops to low then Project Managers can act. If the sentiment score of the customer overall drops, then an account manager can act.
In contrast to the sharp changes in score we can also plot monthly trends and analyse these via BI tools such as Qlik and Power BI which highlight possible slow burning issues. Again, we’re afforded insight that means we can proactively improve a situation.
Why not take a look at the platform in action?