Bring The Noise! Our growing, multi-national client lists throws out new challenges each day for our technology team, which is constantly coming up with new, bigger, better, faster solutions to deal with enormous quantities of data coming in and sending email invitations out. Last year our bigger clients were getting up to the tens of thousands of emails a day, and while our infrastructure was not exactly creaking, it seemed a good time to go to the next level. And then came a request from a new client that wanted to get into the “Million Emails a Month” club.
So to tackle this challenge, our newest programming asset Anthony was tasked with creating a sub-system to send vast quantities of email, something that scaled beyond our existing system and could take us into new high volume areas. Anthony, who has a background in cutting edge 3D-graphics (something called “augmented reality”) wasted no time on coming up with a name for the project: The CustomerGauge Mail Monster - specified as a system so monstrous it could get through any volume of email we could throw at it. It was enough to strike fear into the other members of the team…
For any geeks still with us, our email system is quite complicated. Getting a consistent Net Promoter® Score survey result needs rigour in contacting customers. Most CustomerGauge clients survey transactions continuously, so we ensure that end-customers receive a survey invitation after every customer service contact or e-commerce transaction. Our system automatically sends emails every day. There are various rules applicable for each client – for example, 1) De-duplicate similar order numbers (in case feed was sent twice), 2) Send email invitation exactly 7 days after transaction, but not if we have contacted customer in the last 45 days (never good to over-contact them). 3) Then send a reminder email after another 7 days if customer has not answered.
Creating the HTML invitation email itself is no simpler – we have 20+ languages for some clients (with western and asian characters), and conditional-merge items (like product images, signatures and order information) to increase completion rates. Delivering emails successfully means avoiding spam filters, so we also remove bad addresses, and carefully “throttle” email sending to same domains.
This has resulted in great success for us and our clients. Survey completion for CustomerGauge clients is among the highest in the industry – we regularly top 25% for B2C and approach 50% average for B2B.
What’s that coming over the hill?
Back in the lab Anthony’s Monster had started to take shape. Building on our existing highly advanced Amazon EC2 structure, he generated special server instances that could scale up as needed. He demonstrated it to an eager technology team. With an impressive combination of programming techniques Anthony had built a system that in his words “…is always looking for food [in the form of data to turn into email], and when it finds some, grows a super-monster to manage the problem, which then clones smaller, zombie-like monsters that work on the queues, turning raw data into finished emails. When there is no more email to process, the mini-monsters simply die, and the super-monster goes back to sleep, ready to awake again…”.
“So a bit like Godzilla in Destroy All Monsters?” suggested Camilla, who surprised us more with her knowledge of daikaijū movies than of the sophisticated system we had just seen. Roy asked: Why? “Well, at the end of the film, Godzilla turns out to be one of the good guys, and kills the evil Fire Dragon, and lives to protect Earth once again”. “Yes” Said Anthony. “Exactly like our new Mail Monster”.
Fight The Power
The Mail Monster has already chomped it’s way through thousands of test emails before being unleashed on the real thing – and has been successfully running for several weeks. It’s now officially out of beta and will be a part of new CustomerGauge installations from this week.
If your organisation has many customers to survey, we humbly invite you to compare your existing emailing animal with the awesome experience of our email-eating Mail Monster. Grrrrrrr.
Reading customer comments is a serious business. Ultimately, it has to be the best way of understanding what your customers think about your organisation – and it is essential that comments get to the right people – the ones that can actually change or improve customer experience.
However, there are some drawbacks to manually classifying the comments:
- It’s not scalable: Some of our clients have hundreds of customer comments a day. It takes only seconds to read and classify each one, but it still takes resources
- Interpretation varies by person: People classify comments in different ways. And what may be urgent to one person is mundane to another.
- Customer comments are variable: It can take some head-scratching to understand what customers are referring to. We regularly see multiple issues in a single comment. Sometimes positive and negative sentiments are expressed together. Or positive comment, “zero” Net Promoter score.
This unpredictability means that it’s almost impossible with the technology of today to do pattern recognition, and automate the task (although we are working on this in CustomerGauge Labs).
So we have a better method, recently productised in CustomerGauge. We call it by various names: “Customer Self Select” or “Automatic Root Cause Analysis” but like all good solutions it’s simple and customer focused. The trick is to make the survey smarter. We ask the customer to highlight the issues that bother them (or delight them) the most. With a few simple clicks, the customer chooses the issues you should prioritise. We then direct the results into a series of “buckets” containing issues relating to Logistics, Product, Service etc. We call these “Level 1″ as they are the high level matters to focus on.
Below each Level 1 issue are more detailed reasons that provide more detail. For example, Delivery issues might be related to lateness, damaged goods or returns – and it’s important to separate out what drives the customer score. These are the Level 2 issues. Customer comments provide context and detail.
Depending on the initial customer rating the reasons can change, and are easily edited in the CustomerGauge administration tools.
The real magic comes in reporting. We now have a quantitative number of issues, selected by customers themselves. We then organise into Level 1 and Level 2 reports, and add Net Promoter Scores together with other customer data – for example order value or segment information.
We build up the data and allocate to individuals, so the Logistics manager automatically receives a report on issues relating to his/her department. These can later be assigned to projects (more in a later post).
The newest report is our Waterfall Analysis. This shows a bridge between Promoters and Detractors to break down the Net Promoter Score by Level 1 issues. It works by aggregating Net Promoter Scores in each issue, and weighting them on by the number of issues selected. It’s a simple way of visualising the impact of key elements in your business on your Net Promoter Score.
And then try the survey to see how our Self Select system works.
We are happy to explain how it is all put together, and demonstrate how our clients are already making great progress with this tool. Revolution is a strong word, but we see this making significant improvements in some sites due to its simplicity of implementation and interpretation. Contact us to arrange a webex and see more.