BPS-Sberbank, one of the oldest and largest banks in the Republic of Belarus, decided to launch a Conversational AI Virtual Assistant, Alesia, with Omilia.
From the launch of the new Alesia at the beginning of 2020, during full 11 months, Alesia processed almost 400,000 clients requests based on Omilia without any agents help! This converts into 30% automation in calls, and 18%, in chats.
The bank managed to ensure a smooth & comfortable reach to the Contact Center for its clients, having a 30 seconds maximum dialling time. Results? Colossal costs savings on agents hiring and great customer experience.

See below the transcription of the interview between Angela Bulachova, Business Development Director Omilia and Anna Litvinovich, Director of the Customer Support Center BPS-Sberbank.

Angela: Good afternoon, Anna! Thanks that you have found time to join us today regardless busy schedule. And are ready to share some information about Sberbank.
Could you please tell us about BPS-Sberbank, your philosophy, and corporate spirit?

Anna: Good afternoon, again! Well, to answer this question… To tell in a few words, BPS Sberbank is one of the biggest banks in Belarus and is one of the top 5 banks here, considering our assets. Being a part of the Sberbank Group, BPS Sberbank sets this very ambitious goal… to become one of the best financial institutions in the world. That’s why we focus all our drive and ambition to reach it.

Angela: Sounds great! Could you please tell me why did you decide to automate your contact center? Was it the necessity to cut costs or you simply wanted to follow new trends? Tell us a bit more about it, please.

Anna: Well, you know, the decision to automate our contact center was taken in a distant 2016. Essentially, there were 3 reasons to do so.
Firstly, growing incoming traffic, due to new products, services, activation of new clients by the Bank, as well as the increased popularity of online channels among our clients. That’s the first one.
Secondly, expectations of our clients to quickly connect to an agent and rapidly solve their issues.
And lastly, business tasks that were requiring fast and high-quality client service, regardless of growing traffic, as well as cover sales without involving more human resources.
Eventually, our Virtual Agent, or as it’s trendy to say, Virtual Assistant project, has become an answer to all of the above challenges.

Angela: Why did you choose Omilia? Nowadays, there are lots of options in the AI market. Why did you decide to go for Omilia in the end?

Anna: You know, Angela, our way of development was far from being standard. The thing is, am… first… during the initial launch of our Virtual Assistant Alesia… we used a different solution, not Omilia. We were young and inexperienced, we had no one to ask for advice, no one to copy from. There were no such solutions in Belarus at all. A… and after a while, we realized that we had completely different expectations from such technology. I’m talking about model training speed, ability to quickly add new scenarios without the vendor’s help. And that’s why after 2 years, by the end of 2018, we decided to switch to Omilia. That’s how it all happened.

Angela: That’s great. So, it was quite a long process. Tell us a bit about Omilia project itself. How much time did it take from the start to launch? Maybe, there are some other details you’d like to share.

Anna: Well, from making the decision to launching a project, switching to Omilia – ’cause the technology or project, in principle, weren’t new for us – coincided with another event, migration of our Contact Center to new versions. So, we synchronized these two projects and, in a year, launched a new Alesia, based on Omilia.
And in 6 months later we transferred all services and skills from old to new Alesia, adding a list of new, complex, and unique services that didn’t exist before. I mean different dialog elements, with complicated integrations that started to be available for our clients.

Angela: That’s wonderful. And what exactly has changed in your contact center? You described quite a long way to introduce innovations and changes. What changes did Omilia bring in the end?

Anna: Well, I would be better to share some numbers to reply to this question. Talking of changes…You see, from the launch of new Alesia, which happened at the beginning of 2020, during full 11 months, Alesia herself, without agents help, processed almost 400,000 clients requests based on Omilia. Relatively speaking, because contact centers are of a different size, this converts into 30% automation in calls, and 18%, in chats.
What does this percentage of automation mean? Firstly, it gives us better bandwidth capacity, which, if you remember our background, is very important in a situation of increased traffic, which continues to grow up to date.
Secondly, we managed to ensure a comfortable reach to the Contact center for our clients – dialing time during this year has been maximum 30 seconds. And thirdly, the key is the costs reduction, that involves two factors. First of all, we save colossal costs on hiring – with this level of automation, I mentioned, and current traffic flow, we save on 18 employees monthly. This can be easily transferred into money equivalent. On the other hand, we get direct financial benefit from cutting our staff back that happened last year, current year, and is planned for the next year. So, besides… in an increased traffic situation, we are not only saving funds but also optimizing our agents.

Angela: Am I correct in understanding that some percentage of you agents were granted an opportunity to do more complex tasks, let’s say, more beneficial for the Contact Center?

Anna: Yes, that’s absolutely right. During the first stage, when we were learning to work with similar solutions, we didn’t set a goal to cut our staff back, we had as a goal… to route new traffic to our Alesia, and more difficult issues, to be solved by the agents.
But now the level of maturity has reached the point, when we can easily automate issues during increased incoming traffic, as well as optimize the agents. And all this contributes to the effectiveness of the contact center, of course.
And what was the attitude of your clients towards Virtual Assistant? Possibly, you measure the level of customer satisfaction or get their feedback?

Anna: We conduct yearly polls, we even do it twice a year, and it is about the level of satisfaction by the Virtual Assistant exactly. You know, we get lots of useful insights. The first thing that makes us happy is that the reports contain information that clients do use Virtual Assistants in everyday life, they look for the info on the internet, use navigation, etc. That’s really good news for us, plus this trend continues to grow.
So, in principle, the level of trust towards Virtual Assistant that lives in our contact center is growing among our clients. They have a good attitude towards it, they’re ready to interact with it, they know what to do and learn to use it. And, if something goes wrong, clients give constructive remarks and tell us what we can do better.
Of course, there is some percentage of clients, unwilling to talk to VA. And we always make available an opportunity to talk to an agent and solve issues with them for such clients.
In general, the market in Belarus, as I’m sure the neighboring ones as well, is ready to communicate with such solutions.

Angela: That’s great that level of trust is growing. It’s still a bit controversial topic and not everyone is prepared for Virtual Assistant implementation. Could you please share with us what self-services are available for your customers?

Anna: You know, I would answer this question a bit broader. Very often in our terminology we use the word “skill”. Talking of which, our Alesia, the new Alisia, has got more than 110 skills.
Starting from the basic, she can direct clients to correct skill-groups that exist in the Contact center, depending on the question or segment of a client.
Then she can also answer simple questions that don’t contain complex dialogs or integrations. She performs numerous automatic services, which already include integration, e.g. card balance or block of client’s card.
This being said, in such automatic services, we have a unique for Belarus case – voice verification of customers, in cases when it’s needed, with the help of Voice Biometrics, that we have integrated together with our partners, which has caused quite a “wow” effect for the customers, as well as protected from fraud.
Except for automatic services, Alesia, in incoming traffic, can react to malfunctioning situations, that can happen inside the company, and after that informs our clients about the recovery of the services, products, etc.
She can conduct dialog scenario. That’s how we call cases with suspended requests or smart scripts. For example, she can register complaint about malfunctioning equipment or withdrawn card through the dialog. These are the main skills, groups that our Alesia possesses today in the incoming traffic.

Angela: Sounds amazing. And what percentage of calls is still being serviced by agents and what are the reasons for it?

Anna: Well, if we start from the opposite, as I mentioned before, Alesia serves 30% in calls and the rest is being performed by agents. Excluding some lost calls. You know, there are several reasons for it. Firstly, we automated a maximum amount of simple routine cases to save our agents from them. Now, there are more complex scenarios involved, and, if this year we devoted to migrating old skills to the new system, now we can see that, thanks to Omilia, we have an opportunity to build more dialogs. We’ve completed several cases this year already but will do more next year.
And I’d also like to tell that we had some, I wouldn’t say disappointments, but some mismatching expectations when we were to start the project. We’ve got 30% in voice and 18% in chats – one would have thought that it’s easier to understand client in chat, faster to reply. In practice it went a bit differently, a… in text, for example in the banking app, where the VA can solve lots of issues without agents, to check the balance, for example, we have lower automation.
In general, we continue working on it. We are not setting unachievable goals, and by next year we’d like to get 40% in calls and 30% in chats.

Angela: I hope that you’ll manage to do that next year, and you’ll reach your targets. Having come this long way in introducing innovations, automating your contact center, what could you recommend to other banks that are starting their digital journey?

Anna: Yes, you know, we’ve been working with speech technologies for a while, and virtual assistants in particular, and I can guarantee that for any contact center this kind of solutions bring massive results.
Firstly, they save rather expensive agents from massive workload and automate it. Secondly, they are very agile in adjusting to quickly changing situations that happen inside the company, for example some malfunctions, as well as on a national level – like currency fluctuations, or even the COVID-19 story. The tuning happens in a blink of an eye.
Besides these solutions, our Alesia is based on, like Omilia, they have broad specter and can be used not only in Voice in incoming traffic, but also in chat, or in outbound interactive campaigns – what we are using on a daily basis.
Lastly, the technology is friendly to other speech solutions, like voice biometrics or speech analytics. And when they are all implemented in the contact center, we get a completely different effect and new level of customer experience.
That’s why I can surely say that once you try it, like we did in 2017, you can’t imagine the contact center’s life and activities without such technologies. Especially, if one wants to improve customer service.

Angela: I’m glad to hear such feedback from you. Thank you for your advice for other banks. And I’d like to thank you for your time and readiness to answer our questions. Have a great day and good-bye!

Anna: Thank you, you too!