Abhay Singh

Yes. I can.

Hello.

Not really. Everything was fine. Thank you.

Yes. I am.

We can't proceed.

Thank you.

No, it's not required. We can proceed now. Thank you.

I am an ops leader and customer success, uh, customer experience background strategic leader over the years. Uh, I am put up in, uh, Hyderabad with family. I have spent around seventeen to eighteen years in the customer experience and customer client facing vertical, uh, leading multiple geographies and markets. Uh, that includes eleven years of strategic leadership working in multiple startups, uh, involving fintech, travel tech, edtech, Internet, and much more.

White excited to be here.

Thank you.

Definitely. So, uh, this was way back in one of the startups, uh, when I first, uh, did a lot of market research in controlling the, uh, hold times for the customers calling us from across the globe. Uh, the call abandoned rate was around, uh, fifteen to twenty percent, which was quite high. I went ahead and led this project of, uh, controlling the call abandonment rate, uh, and that was also impacting the customer experience metrics such as CSAT metrics by fifteen to twenty percent down that we were targeting. Uh, so what I did was, uh, I explored what is available in the market in terms of chatbot systems, chatbot features, and, uh, I single handedly let, uh, this project with the tech teams and product teams to implement the chatbot system in the in the flow. Uh, this really helped us to curb around, uh, sixty to seventy and even more, uh, manually handed tickets by the system going forward, adding more, uh, NLPs in the system day on day, uh, in three to four weeks. We achieved around eighty percent of, uh, such queries being handled by the chatbot system. Eventually, uh, when this was in full flow, uh, my customer experience metrics such as CSAT improved considerably by around twenty to twenty five percent. Uh, we were around seventy, seventy five percent overall CSAT. Call abandoned abandoned percentage also was controlled with just a three to four percent after the botched implementation. And, uh, this also impacted the customer retention rate Churn rate by around twenty percent. So, uh, this was all due to the RPA automation process. Uh, one of the experiments and one of the projects that I had led, uh, which was really impactful, uh, process related and also making us, you know, uh, evolve well and, uh, a fast paced setup that we were scaling. So this is a really good experience I had to share today.

Uh, we had lots of, uh, reporting dashboards, customized dashboard, uh, such as mixed panel, database, uh, that clearly showed us the numbers, the, uh, trend lines, uh, even, like, uh, Power BI reports were clubbed from the CRM... or mapped from the CRM that helped me to track daily trend lines, footfalls of the customers, uh, how much of the dissatisfaction rates, how much cost... how many customers were not happy, or how many customers were happy. These metrics were very easily tracked from, uh reporting dashboards that I was continuously monitoring on. We also had the CRM called Freshdesk, which had this reporting dashboard feature available that really helped us to track daily... the process metrics in terms of low or average or above average, and then take the required action planning accordingly. So external reporting dashboards, some internal customized dashboards, clubbed and mapped into the CRM that we were using. When I say CRM, it's the ticketing CRM I'm talking about. This really helped us to and me specifically leading the process to take corrective actions for anything that we found as a loophole for the previous day or for the previous week numbers.

So any feature that goes live, uh, it's very important that you, uh, check the success metric of that product or tool. When it was a chatbot system, I made sure that every day I was tracking, uh, the performance of the chatbot, uh, linked with the performance dashboards and reporting dashboards. I used to call the, uh, end users who were using those chatbot features. Uh, the end user, meaning the customers and the clients. And I was taking their feedback in, uh, what went good, what went wrong, any improvements required in the chatbot flow. This was really helpful and very much impacting because every day I got some feedback that this is something not working well. Let's improve this. So I used to go back to the product teams and tech teams, development teams to make sure that change

His work corrected and the changes were made accordingly. So the refinement is very much important day on day when you have any product launch, which includes the chatbot and AI automation, which I was doing as a best practice every day. That... the numbers was giving me the, you know, the nuances or anything that is working well or no... not working well. And then customer feedback, which is a part of the VOC process, this also helped me to track anything that was working well or not working well for some reasons. Accordingly, the action steps and the corrective action was taken by me and the team.

So, see, I mean, when you have a dashboard system in place, you...

In startups, the the best part I see is you can customize your own dashboards. Right? So whatever numbers you want to see, you can customize it either by yourself, running an SQL query, or sitting closely with the BA or the MIS person to see what numbers you want to track. I know that I'm working for a particular goal in terms of a success metric outcome that I want to achieve in the next four four to five months. Success metrics includes AHT, FCR, the SLAs, FRTs, CSAT percentage, NPS retention, churn rate, So these are my benchmarks which, uh, you know, you always have to set as a target that I will be achieving these numbers by the end of three months or six months.

Once you track and customize the dashboards and have a daily monitoring in place, you will be able to get to know that your product implementation of the chatbot system is working well or not. So this is something I've been doing over the years in any process automation or anything that I have been involved in terms of robotic automation in place. The same was done here as well.

which really

give me good insights of the success metrics.

Yes. So, uh, I I can I can tell you that the cost per, uh, cost per agent, uh, was reduced by around, uh, twenty five to thirty percent because we had less number of people, uh, working for manual queries. Most of the process was automated. So, uh, we were saving cost on hiring. We were saving cost on training because, uh, most of the times, we were able to achieve the numbers and the metrics with limited number of people, and the... it doesn't mean that we are firing people or we were, uh, like, removing people from the process. We were using them for other process related, uh, you know, initiatives and process related projects in some other departments, uh, internally, uh, which, uh, was a part of, uh, you know, putting down cost for a particular team. Say, I'm a process, uh, owner for operations. Uh, I need to make sure that x x number of, uh, agents or x number of associates are only available, uh, which used to be y number earlier, which is more... which is, like, less than the previous numb... so automatically, my cost per agent, uh, in terms of operational, like, burn will reduce. So when I see, uh, the full full effect of the chatbot implementation, uh, I Was able to reduce it by twenty to twenty five percent operational cost when it comes to cost per agent revenues, uh, in terms of, you know, the revenues versus the burn we were doing or the expenses we were doing in the operations. Because the people's strength was reduced considerably, uh, which were, uh, in use in other processes and other departments. Uh, hence, the ROI was really impacted in the positive side.

Around six months?

it took because, as I mentioned, this was a very pilot project for me when it comes to this new chatbot feature, which was very new that time in the market. I personally had to deep dive well, take some time off from the normal processes. I I was leading there. And whatever time I got, I used to invest more time in doing the r and d and checking the nitty gritty's, errors, anything that the chatbot could not, you know, like, fetch from the the queries that it used to handle. In earlier days, it was around ten to twenty percent you know, twenty percent around error rate, which gradually reduced to around five percent only in terms of the error rate by around three to four, five months. So I can say the breakeven was achieved in five to six months easily when we were able to get the most of the chatbot flow, which used to handle around six to seventy percent of the queries that were landing into us. So that was quite an achievement for us as a startup.

Yes, I have. I have. Because, uh, the implementation cost, uh, and the time given by the product teams and tech teams in terms of man hours, everything is in consideration. The cost reduction in terms of salary, uh, any expenses, or anything in terms of the dashboard that we had to remove, uh, any extra expenses that was used to, uh, make sure that same numbers are achieved, all expenses are considered here in terms of the ROI calculation. Uh, so which is around twenty five percent than before.

Like less than before, yeah.

So, uh, we had a routing in place. So every day, the number of queries handled by the chatbot flow, uh, we... I was able to get all the queries, uh, which... so there was a deflection rate, you call it. Right? So deflection rate was easily measurable and quantifiable metric, which, uh, was, uh, easily available and fetched from the dashboard that I was monitoring. So whatever the queries, uh, routed through the chatbot and answered by the chatbot, it gave me an, uh, calculation also and the other... ROI effectiveness of the chatbot. So there are ways how you can do that. So how many of the queries are handled by the manual, uh, agent or humans, and how many of the flow, uh, gets routed to the chatbot when it gets to answer those queries. And then you calculate accordingly that, uh, how much of the man hours or time that is being saved or the AST that is being saved, uh, the occupancy level, uh, the utilization rate has been saved. So you calculate accordingly that how much savings have you done, uh, with this chatbot implementation. So, yes, all these factors were included.

It comes from, like like, real time process requirements of, you know, the the targets that you have, the OKRs that you're driving, the KPIs that you're

driving. So prioritization of SLAs, a culmination of every process, OKRs that you're driving as a business also. You know that, you know, certain care... KRS and certain KPIs are on top priority because that is

At a business level, okay. That is a part of the business level. Okay. So the priority level comes on that. If you notice that, say, the HD is a constraint and the HD is a blocker in the process, then that becomes your SLA to control. Right? So how will... you will control what efforts you will make in the training, audits, operational efficiency. So that is a part of the process requirement in terms of how it is synced to your process and then to the business OKR as an organization, which helps me to prioritize the SLAs accordingly. and deprioritize the other lab, you know, like, metrics and SLAs accordingly.

it's all about the process requirement, uh, and, uh, something that we want to achieve as a process as a business, which this... which decides the priority level.

I can say I had deprioritized outbound calling in one of the metrics. So we were doing some up... upselling of certain, you know, add on features in one of the processes as of... like, I was handling. So the same service agent person who is supposed to take an inbound query, which is post onboarding stage, had to sell something also which was a part of their KPIs at one point of time. But I noticed it is very much impacting their productivity and... because they were not... you know, it's all about the skills that a person has. A person, I noticed wherein that same person who is very efficient in handling an inbound query for a support function might not be a good seller or he or she might not be able to pitch well in terms of sales. So I had to deprioritize this SLA when it comes to their outbound selling, like, metrics that was earlier being tracked and how much of the revenue they are coming in or bringing into the table in terms of outbound selling or upselling. This really helped well because I segregated two teams completely when comes to cold calling or upselling process and then separated the support function, which was just focusing on the after sales support queries. So this deprioritization of, uh, the upsell, uh, you know, KPIs or the role for a support function, I had to deprioritize, and this really worked well in enhancing the efficiency and productivity of both the teams after separation. So, uh, the benchmark and the targets were set accordingly. They were more focused on their own

KPI's and PRAs in terms of sales or upsell or support. So this was one example I I have to share when it comes to deprioritizing one of the SLAs that were existing earlier.

Yeah. So earlier, the, uh, the SLA of, uh, one of the, uh, the existing process wherein the same agent had to upsell.

The HD was always, uh, high because, uh, they used to spend most of their time and they used to struggle when it comes to the selling part. So on the same call. And that reduced, uh, I mean, uh, the, uh, motivation level efficiency because that was also mapped to their own performance and performance metrics and productivity also. So I, uh, noticed that

when we separated this, the AHT, which is one of the key metrics of any contact center, that considerably reduced back to around four to five minutes, which earlier used to be around eight to nine minutes, something as an average call handling time. So this was one of the most important metrics that we were able to achieve. Plus, on the other side, when the other team who was dedicatedly taking the upselling and the selling part in terms of their own PRs and KPIs, there, like, we noticed a considerable increase in the conversion rate. So conversion rate of the same customer when we were doing an outbound call and selling something, because we we wanted to have that lead first on this portfolio, and then the call was diverted to a person who was considerably trained on sales pitch. So we noticed that the... when a dedicated person is calling the same client or customer for the particular sales query or anything that we noticed that he or she is interested, that revenue increased much better than the the same earlier person handling on the same support call. So HD impact was noticed, and the revenue was noticed as well for... and the conversion rate. So these three metrics was really helpful for us to see up on a positive side.

Earlier, the conversion rate was around twenty to twenty five percent when the same support staff was taking the task... the responsibility of selling. But once we had a separate dedicated team with their own KPIs and KRS assigned and and mapped to their performance metrics, the conversion rate improved by around fifteen to twenty percent. So it was around forty, forty five percent conversion rate than the earlier twenty, twenty five percent conversion rate. So that was really a good achievement for us when it comes to this Conversion rate metrics take shifting to a higher side.

In a span of around one to two months, because the separation of the team also required a bit of training, a refresher of, you know, what what to do, what not to do on the calls. So this took some time. We had to identify people who are good in sales and convincing skills, separate the team accordingly. And so the entire revamp of the process took around two months. The third month, we could see the results.

See, I mean, uh, it's all about the intent of... and and the the the goal for our team, uh, be it the support of these sales. It's it's it's very specific to the performance metrics, which is, uh, again, same to the KPIs. So, uh, it's not because of the seasonality. It's about, uh, the intent, uh, from the customer side when we are supporting them, the end user who is reaching out to you.

So it does not actually depend on the seasonality. It was most... mostly, I can say, statistical, like, difference we could notice when it comes to a throughout conversion rate... consistent conversion rate throughout the season, not for a specific season. So because every month, the people... the the staff or the team had to achieve certain targets and goals because of the split between the two teams. So every team member was aware of their individual goals in sync with the process goals and targets. So this really helped well, uh, and not depending just on the seasonality here. Uh, but, yes, such as... statistically, I can say the, uh, the difference was clearly visible in conversion rate throughout the year.

I will mostly say the intent and approach, the mindset of the candidate, how much... firstly, when you... you're taking an an interview for the customer experience manager, he or she has to be, you know, in the footsteps or in the shoes of the customer himself or herself. So I will have mostly the scenario based questions in terms of gauging how that person will handle in customer query or an escalation or a complaint coming in, how polite and how humble and how calm and composed that person is. So it will come out in the thirty minute discussion round with the way they intense the the post the post share, the body language, the calmness of that person, which is very, very important skill set when it comes to a customer experience manager. The tone is very much important. So this this attribute should be there in a customer experience manager. In thirty minute call is sufficient to gauge apart from the the experience that person carries in different roles and different companies in terms of titles or designations. That is obviously very important. But the approach, mindset, body language, calmness, smile, on the face, uh, really, uh, you know, uh, like, going with the flow. These attributes can only be gauged when you speak to somebody for thirty minutes, uh, which is sufficient for, uh, the interviewer to gauge that this person will be a good customer experience, uh, professional when it comes to handling clients and customers. So, um, this is what I will do apart from the CV and the experience that person carries in the same role. Uh, how does he portray himself or herself during the conversation is very much important to gauge, uh, that if he or she is in the real time scenario, How will he behave? Right? So you'll come to know the... like, about that.

Yeah. So I will... so it depends again on the industry and the process that that person will be leading the customer experience team, but I will more specifically ask about any customers' refund pending, which shows that it's already refunded from the company, but it's still not received to the customer's account. Because this involves lot of investigation between the aggregator, the payment gateway, the bank, And I have seen this is most, uh, I did situation when, uh, the customers call you or reach out to you. They are very, very frustrated, angry because money is involved. Your money is stuck with, uh, you know, with you. So, uh, this is something I would ask as a scenario when it comes to how will you, uh, make that customer become and compose and wait for the, you know, the particular transaction to, uh, I mean, to make him or her understand about the process, how does it work in terms of a a payment coming to you, and going back to their account via certain medium, uh, because they... I have seen that customers, uh, in my overall experience, they don't really wait and understand when it comes to money stuck somewhere. In any process, uh, wherein credit card or transactional, uh, you know, cases are involved. So this is one scenario I always ask that, say, I'm a customer, uh, my certain x y z, like USD stuck with you, I haven't received the refund within five, six days is over, where is my money, uh, how will you take care of that situation. So this will be, like, giving me a perfect... because it involves a product technical knowledge as apart from soft skills.

I will see the SLA of the query being received, like, how much of the time has been over. So this is first thing, which is the SLA, the turnaround time, you call it.

Second metric will be the FCR, I can say. Like, the first call resolution and... which which includes, like, nonrepetitive answers.