How a bank, a travel agency and a sales team built call centers on Twilio
- Summary:
- How ING bank, Momentum Travel and internal sales provider Centerfield built contact centers based on Twilio, but without using its Flex call center platform
We built our own Twilio Flex, before Twilio Flex existed.
Or take Eugene Kovnatsky, CTO at internal sales provider Centerfield, speaking in the same session:
We built Flex, basically.
Both seemed in good spirits about it, though. Waiting for Flex would have meant going live later. And there are some things that you can only do if you build your own solution, as Jeroen Visser, Platform Architect at global banking giant ING, told the same session:
For us, [using Flex] hasn't been an option. We have a common front end that we use for ING for both web and internal applications. That's a reason to stick with our own platform.
Two ways to build a call center on Twilio
Flex has been a new departure for Twilio, which is best known for providing text messaging and notification services used by the likes of Uber and Airbnb to communicate with customers. Developers like building APIs for its communications services into their own custom applications. It has gradually expanded its offerings over the years to include voice, SMS, email, chat, video, Facebook Messenger and other channels, along with a series of pre-built components and a drag-and-drop visual editor.
It's no surprise therefore that many customers have built Twilio services into their own home-grown solutions. And even if using Flex would have given them more of a head start with elements such as the user experience, workflows, and add-ins such as contextual intelligence, building a call center app is quite a project either way. Unless the packaged solution met their needs out of the box — and how likely is that? — they would have had to adapt it to their specific requirements anyhow.
You also have to figure with the typical preferences of the kind of people that put in call center solutions. For Momentum, it was really important "to have full control" of the solution, says Syms. At Centerfield, "the company culture is more build vs buy," says Kovnatsky. And we already mentioned why ING followed its own path. These are factors that Twilio acknowledges by emphasizing how customizable Flex is — but however much flexibility it offers, there will be many more companies like these three who will still choose the DIY route.
Why do it yourself?
Centerfield is a good example of company that's very comfortable developing from the ground up. This is a fast-growing, Los Angeles based, internal sales company that works with home services giants such as Comcast, AT&T and others. Its 1,500 strong workforce includes teams in the Caribbean, central and south America, South Africa and India, and it handles more than a billion voice minutes annually.
Digitally astute, the company makes big use of AI to support its sales agents, relying heavily on data analysis to make data-driven decisions. Therefore it's important to be able to tap into different data products and do multi-variate testing, says Kovnatsky. It takes a DevOps approach and typically runs "a couple of experiments a day," such as testing landing pages, or trying out agent assignments based on machine learning recommendations. Its platform includes many AWS components and makes much use of microservices — that makes Twilio's API-first approach a good fit, Kovnatsky says.
Momentum is an online travel agency, based in Canada, which operates the Justfly.com and Flighthub.com brands. It has grown rapidly over the past four years and had tried out other solutions to help support its growth before starting with Twilio in January last year. Within three months, recounts Syms, it had launched a "minimal viable product" to a first group of agents, rolled out a full "version 1" in June and by October had more than 500 on the platform. Data is stored in AWS and Ytica provides analytics.
So far, Momentum has seen a 17% reduction in average handle time, which Syms says has freed up valuable time for training and quality of service actions. It's also seeing improved analytics and the cloud-based solution brings more location flexibility. The team is looking forward to adding AI and omnichannel capabilities.
Global scale at ING
The project at ING Bank is on a global scale, due to handle 40 million contacts a year once fully rolled out. Staff in the Netherlands, the bank's home country, have already been onboarded. This accounts for 10 million contacts a year, 80% of which will be self-service.
The ING contact platform runs in the company's own data center, using Twilio services via APIs. Individual countries then build their own call center applications on the core ING platform, which also supports other internal and customer-facing applications. The ING team have built their own user interface and chrome extension, plus a configuration test screen for agents. One request that required Twilio's co-operation was to put Dutch at the top of the list of languages in the queue for speech recognition.
For those who believe call centers are only for sales and customer service, it's interesting to note that ING chose the HR department as the first group of 200 users to try out the new platform. The reason is that HR would only be dealing with internal callers and therefore any major issues could be ironed out before the system started handling customer calls, explains Visser.
My take
An interesting spread of different use cases for call center software, and an insight into the different reasons why people build their own.
[Updated - Momentum operates the Flighthub.com brand, not Flydot as originally stated.]