Identifying the cracks early, and dealing with them before they grow into pain points, is a priority for most companies. But identifying them is hard. Companies are endlessly surveying customers and analysing existing data to identify leaks in the customer journey, but the results are usually hard to trust, too high-level to be actionable, or too late to make a difference.
Maddyness spoke to Sharad Khandelwal, cofounder and CEO of SentiSum, customer analytics software that helps users get to the root cause of customer friction. Sharad outlines how his academic background in machine learning helped him found a startup, highlights the perks of a simple interface and shares his experience working with British Airways.
[Maddyness] Tell us what your startup does and how it came about in your own words. Did you have expertise and experience in the field you chose?
[Sharad] We built SentiSum in 2016 to bring the latest developments in artificial intelligence to this problem. Our AI makes vast quantities of qualitative customer data accessible to brands—whether that’s customer support conversations, customer reviews, or free-text fields in feedback surveys.
We style ourselves as a customer analytics platform because we unify every source of customer feedback into one accessible place—our customers can then dig into the data to do root cause analysis of customer friction. But, our major focus is customer support tickets.
I studied Machine Learning at UCL, worked in the field for a while and we’ve now built one of the most talented NLP teams out there. As a team, we’ve found that bringing this talent to support tickets is what helps our customers unveil friction points in the most detail. The speed and volume of customer complaints through this channel ensure companies can work proactively to tackle the root causes of those complaints and drive better business outcomes.
I’d never worked in customer service or experience myself, but thanks to a lot of research and now almost five years in the business, we’ve built a great product that’s ready to scale.
What kind of insights can SentiSum provide, and why are these important?
SentiSum uncovers two core data points from qualitative text: sentiment and meaning (or topic). But we apply both of these in a variety of ways.
When a customer leaves a review, fills out a feedback survey, or writes into a customer support department, they’re usually talking about lots of different things. At scale (say 100,000 reviews or 10,000 monthly support tickets), that can be extremely complex to understand.
Our AI model (once trained on the customer’s data), picks out multiple topics in each piece of feedback data and categorises them in a hierarchical way. As a hypothetical example, when looking at an individual support ticket for a food box delivery company like Riverford Organic Farmers, we could find an overall theme like ‘missing ingredient’ is causing the complaint. Most analytics stop there and require a human to sift through the conversation to find out more details.
Our AI pulls out sub-topics, so the Director of Support or Experience will know that the conversation was driven by missing carrots in particular, and there was a whole host of other problems causing the customer to be annoyed.
On our platform, we’ll collate this data so that our customer will know the volume of support tickets, reviews or surveys that mentioned each specific issue. That gives a really clear idea of what’s driving friction and which issues should be fixed first.
Knowing these insights and also the level of customer emotion they create is important for a number of other reasons, too: from identifying and responding to potentially harmful social media reviews and optimising conversion rates, to measuring the performance of a new product or feature in real-time.
What makes a company truly ‘customer-centric’ and why is this a good thing?
Customers are everything to a business. Most businesses are set up to acquire new customers, as they should be, but often fail when it comes to the retention of existing customers.
The data proves that it’s cheaper and more profitable in the long-term to retain customers than to acquire new ones—and that’s why being customer-centric is important.
We actually address how to create a customer-centric culture in the first chapter of our first ever eBook (which you can find here). To simplify it massively, it boils down to:
- Enabling employees with tools and software that allow them to put the customer first. One example would be removing admin work in the customer support department so they can spend more time with customers.
- Empowering employees with the authority needed to put the customer first. Operational procedure so often gets in the way of customer-first. Whether that’s the unnecessary need to escalate complaints or that only managers can issue refunds—both create mistrust and demotivation amongst customer-facing teams.
- Encouraging employees to be customer-centric. There are so many ways leaders can show the customer matters most. The impact of someone in the C-suite walking in the customer’s shoes or jumping on a support call can be large.
How did you develop your software - what psychological considerations did you bear in mind?
I built the MVP myself a while back. We were part of a 500 Startups cohort so we hit the ground running. We’ve had a lot of iterations since then by our excellent product and engineering teams who should be getting all the credit for what the software does today.
We build for simplicity because, although the AI and background engineering is complex, our users don’t need complexity. By focusing on the most important information first and the ability to deep dive into it, we let users get high level and go deeper when they need to.
Simplicity is important because everyone across a company can benefit from listening to customer insights—whether they’re technical or not.
We’ve had clients roll out users from customer support leadership through to product through to marketing and operations. It needs to work for each of them.
You work with big names from Schuh to British Airways. How have you helped them?
We’ve worked with both for a while now and they have been really valuable partnerships for us. They took early risks working with a startup when we launched and that has meant we were able to build our software to better serve them at every turn.
For both, we’ve made it simple for them to make sense of 100,000s of survey responses and reviews in minutes. A task which took them hours of manual work—and even then they would typically take a handful of data points as a sample.
We’ve also brought them objectivity. Manual data tagging is not only time-consuming but it’s subjective and sometimes biased. Automating that process allowed them to take a singular, objective look at customer pain points.
Our first anomaly detection use case was built for one of these two. We were able to quickly identify when a customer was having a bad experience with a third-party vendor, so they could change it for future customers.
What have been the biggest professional challenges during lockdown? (and have there been unexpected rays of sunshine?)
It’s been a challenge not being able to hustle alongside the team in an office. Working from home (even internationally) has always been something we do, but this year we had ambitious targets that, at times, could have moved faster in an office. Especially onboarding new hires!
Other than that, we had to change our customer focus during lockdown. As everyone will already be aware, some industries are really struggling at the moment and we thought it best not to come knocking on their door right now.
On the positive side, we learnt to work well from anywhere giving everyone the ultimate flexibility, which also opened options to hire talent from anywhere in the world.
What’s in store for the future?
Happiness, prosperity, growth, success — that’s how any founder sees the future 🙂
But really, we have ambitious goals to be the category leader when it comes to customer insights and automation. 2021 will be all about laying the foundations for growth by investing in customer success, brand building and product innovation. We’ll continue to build our use cases and build a client base with our partners over at Zendesk (who have an awesome customer service blog by the way!)
And finally, a more personal question! We’ve started asking everyone we interview about their daily routine and the rules they live by. Is it up at 4am for yoga, or something a little more traditional?
This picture sums it up pretty well!
I have a pretty normal routine really. Every day is different depending on the different meetings, I try to then juggle cooking, eating, taking a long walk, and watching some TV in between. My life rules are: be kind and pursue happiness. It might be a bit cliché but I think there’s a clear ‘why’ behind building a successful company, and it’s to be able to be free and happy.
