Year In Review
On February 6th 2016, Mike and I set out to reduce the cost of providing customer support at scale. We’re solving this problem by combining the best machine learning technology with the best live chat experiences.
With the end of the year nearing, I thought I would take some time to update our customers on where we’re at with our solution, Ada.
It was a tremendous year of growth. We built hundreds of features and improvements this year. Here are the highlights:
- A multi-platform layer targeting SMS, Facebook Messenger and our own web chat simultaneously
- Automatic translation of answers to many languages
- Stage one of offering authenticated, API-backed, personalized answers¹
- Conversation context and tracking of metadata from anywhere
- Deeper help desk integrations
- Richer analytics
- An improved conversation view with filtering and live updates
- Activity logging for auditing and reverting unintentional changes
- Improved user security, profiles and roles
- Accessibility compliance for our own web chat platform
- In-chat reviews of answers and better measurement of satisfaction
We also scaled our platform to handle 100 times more traffic and improved our learning engines confidence by 10%². At the beginning of this year we had to handle 20 conversations at any given moment. At peak, we now handle 3000 concurrent conversations and are set up to handle much more in 2018.
We're excited to begin 2018 with a renewed focus on projects that directly move us closer to our goal of reducing as much of your support as possible, while offering a better end-user experience³.
Our focus for 2018
- Deep, rich, tailored experiences for your customers, backed by your own APIs for delivering account balances, order tracking information and much more.
- Seamless agent hand off to live chat support backends that our customers’ teams already use so that when a person is needed, the resolution for the end user is simple and fast.
- Better measurement of your ROI and your customer satisfaction with an improved analytics dashboard to understand all of it.
- Generally much faster management and training of the bot.
We're also going to involve our customers more at every step of the product development cycle and we're going to communicate more frequently about what we’re building and why. If you'd like to be one of the teams we test new features on, please email me.
With that said, I think 2018 will be the year automation-first live chat becomes ubiquitous and we’re thrilled to be leading the charge.
I’d like to end this with a sincere thanks to all of our customers. Because of your trust in our team, our business has grown by a factor of six this year. This growth has allowed for a much larger team which of course allows us to further improve the platform while growing it in the directions above much more effectively.
Head of Product
¹ This feature is hard to get right. We’re still developing and testing it. We’ll properly release this to everyone by Q2 2018. If you’d like to help test it with your own APIs, please email me.
² This may not seem like a lot, but these are “NASCAR” numbers. Every percent of confidence increase is exponentially harder-won than the last.
³ For example, if one of our bots can’t help, we should make the process of talking to a person better than it had been before us.