Forrester and Freshworks: Delivering Recession-proof, Retention-focused Customer Service

The pandemic has been hard on businesses, as revenues and customer retention rates have fallen. To survive and thrive in this scenario, they must focus on delivering great experiences to customers. In this conversation, Arun and Ian discuss the best CX practices businesses must adopt to succeed in an economic downturn, with three real-world examples.

Arun Mani

President, Freshworks Europe

Ian Jacobs

Principal Analyst, Forrester

What's in this interview

  • Handling crises in a recession: Surviving vs thriving
  • Overcoming crises: Protect revenue, optimize costs, and create operational flexibility
  • Businesses that beat the pandemic with stellar CX: Mecca, Klarna, and PhonePe
  • Best practices: Building recession-proof contact centers

 

 

Part 1: Protecting contact centers against recessions and crises

Arun Mani: The COVID-19 pandemic left an economic downturn in its wake, and the effects were very similar to that of a recession. What does that mean for a contact center? We’d best go back and look at the past. The last recession was in 2008, led by the subprime mortgage crisis, which started in the US and spread to the rest of the world. 

A Harvard Business Review study found that the leaders – companies which acted quickly and did smart things – actually outperformed the laggards by close to 25%.

 

 

Arun Mani: What is the winning strategy during a downturn? The subprime mortgage crisis was different from the one in 2020 and 2021. The lessons are common nevertheless. For the scenario we are in, there’s a framework that smart businesses follow in 2008 that helped them steer clear of some detrimental effects of the recession. There's a pattern we notice here:

 

1) Protecting existing revenue. It is essentially ensuring that our customers are taken care of and provided with the best possible experience. That's a leading indicator of protecting existing revenue. 

2) Optimizing costs early. There are companies which went straight into cost-reduction mode in 2008, versus other companies, which staggered it. The companies that did it upfront had significantly better performance. We are 20 years past 2008 and technology has improved significantly. How are we leveraging it to improve efficiency and meet the cost imperatives? 

3) Operational flexibility. How much flexibility can you actually build into the contact center that allows you to be more agile as the demands of your business keep shifting over time?

 

 

Arun Mani: We will go through each point in-depth. And I'll be calling upon Ian to share his research and insights. A poll conducted by Edelman Trust and published in Forbes revealed that 65% of customers believe that how a brand responds during a crisis would affect their purchasing decisions. Ian, how does this fact reconcile with customer trends that predated the crisis? 

Ian Jacobs: If you think about some of the famous research that came out of the Harvard Business Review around customer effort, it sounded like the research said: this is not the time for wowing people with amazing experiences, it’s more about just getting work done. If you think about the brands in this crisis who ended up with 12-14 hour queues in their phone system for their contact center, and did not provide any other way for customers to contact them: you can see how the notion of high customer effort leads to a lack of customer loyalty. The other thing to think about is that customer expectations were changing well before we ever heard of COVID-19. 

We saw that consumers were starting to think about the kinds of experiences that were so effortless that they wouldn't even think of them as experiences. Any information they wanted would be there, wherever they were, whenever they needed it. That is a really high bar for brands to handle, but you can see that many brands already fell during this crisis, just by making things more difficult than they needed to be. Now, consumers are a little more forgiving, because we all understand that there might be a five-year-old who might interrupt a call with a support representative, because that person's working at home. But that expectation for a more professional experience is going to come back. We are starting to come out of this crisis and our expectations are starting to revert to what they were before the COVID crisis.

Part 2: Increasing engagement and influencing revenue with with self-service (Self-service maturity model)

Ian Jacobs: You can see where brands are starting to try and meet that customer need. This is data from 2019, pre-crisis. But we saw that brands were expecting their digital support volumes to increase dramatically enough that they would offset the decrease in telephone volume. 

What you see here is that almost two-thirds of brands think their telephone volumes will drop. But two thirds of those brands also expect that the overall number of interactions they have with customers is going to go up. Why? Because consumers are expecting information to be available anytime and anywhere they need it – digital experiences are the best way to make that happen.

There have been a lot of digital channels, particularly asynchronous messaging platforms, third-party messaging platforms, and also things like SMS and digital self-service, where we've seen a land rush, to very quickly implement some kind of conversational AI or intelligent virtual assistant. Many of those were designed for crisis communication. If you were an airline or train company, or even a cruise ship, maybe you’d get lots of questions about sterilization procedures and so on. And they weren't necessarily broad. But once brands have invested in that technology, they will have the opportunity to expand those experiences.

Let’s talk about why, before the crisis, brands had a maturity model in mind for self-service, and how the crisis has only amplified the need to move up this maturity model, its four stages being:

Stage 1: Bettering Service with Self-Service: The first stage is using self-service to enable better service – this is what everyone normally expects out of self-service. Just answering the top 10-15 FAQ questions with some kind of conversational AI, a knowledge base, and the search function will allow customers to solve many of their problems on their own. 

Stage 2: Influencing Revenue: But if you look at the three different factors that Arun highlighted for how you're going to not only survive, but actually thrive, one of the ways to do that is to drive customer loyalty by protecting revenue. And to do that, you need to think about influencing revenue. I think of customer service as a tool to influence purchasing decisions by answering questions – not after the purchase, but pre-purchase. I also think of it as proactively engaging and reaching out to customers, when you think that they are going to have a problem with purchasing. 

Stage 3: Using Customer Data in Product Improvement: This is the phase that I don't think many brands have gotten to yet. As we start to move up the maturity model, we need to think about customer engagement rather than customer service, wherein we're thinking of the customers as a constituency within our enterprise: ideating around new products or new services, having the data from these customer interactions indicate where there might be friction points in product usage, and then feeding that data into the parts of the organization that can actually fix those problems – it becomes a full loop within the enterprise. 

Step 4: Lasting Customer Success: And finally, you get to the state of lasting customer success, where you’re thinking about onboarding, for example – teaching a customer how to use a product or service properly. So now you've got this idea of self-service in the pre-purchase phase, in the usage phase and in the post-purchase phase. 

As I said, this crisis has only amplified our need to create this loop. Because again, if you think back to Arun’s three factors of how to actually thrive, you've got the loyalty piece, the revenue piece, the optimization piece – which touches on both optimizing the purchase process and optimizing the usage of a product to alleviate the need for support further down the line. And then, when there is need for support, you can bring in self-service. 

The operational flexibility piece is something that we're going to talk about later.

Part 3: Case Study: Mecca brands

Arun Mani: Thanks Ian. I want to share a fun fact. As we all know, COVID has had a significant impact across the globe. But it isn’t uniform, in terms of impact. There are industries which went straight into hibernation, like travel and tourism. However, L'Oreal, the world's largest cosmetic company, reported that first quarter sales in China increased by 6.4% year-on-year. And this is despite the country's shutdown, and the manufacturing and everything else almost going into a freeze. So this trend was interesting.

There is a Freshworks customer I would love to talk about, who observed L'Oreal and did something similar. This is the story of Mecca brands. Mecca is a company that's based in Australia, it's a beauty and makeup brand with over 100 physical stores across the region. Now, as you can imagine, they are very proud about their high-touch business. They take great pride in their in-store experience: people can actually walk in, get a consultation with a professional makeup artist, test the products, and purchase them. Ian talked earlier about influencing revenue, and engaging with customers to influence decisions. Mecca took that to the next level, they took their entire high-touch business online as quickly as possible. Customers actually engage with them and do whatever they do in-store. And this allowed them to engage with their customers even during the pandemic.

 

 

Arun Mani:  They also enabled chat – Apple Business Chat and WhatsApp, which were the predominant channels in that region. It’s asynchronous nature allowed customers to pick up where they left off, and Mecca saw phenomenal results from this.

I'll also show you how Freshdesk, our customer service product, simplifies the agent experience and makes them more productive. Freshdesk presents a 360-degree view of customer interactions in one place. And here's a screenshot example:

 

 

Arun Mani: The interface shows pretty much everything about a given customer. The conversation history across channels can show what kind of virtual classes they have taken from Mecca, what kind of one-on-one sessions they had with a cosmetics professional, and so on. You also have information on things they have purchased.

This is the type of technology that's being deployed which is enabling brands to engage with the customers and influence revenue: proactively engaging with them at every stage, and also helping them with onboarding, product usage, and support. It's like completing the full circle that Ian was talking about.

 Now, let’s move on to the second pillar of thriving during a pandemic: Efficiency.

Part 4: Operational efficiency – a byproduct of customer service automation

Ian Jacobs:  One of the things that we saw fairly regularly, even before this crisis, was a shift towards conversational AI. It's something that I mentioned before: the ability to mimic human conversation. But before the pandemic, most brands claimed that they were investing in conversational AI to improve the customer experience, and only about half were looking to reduce cost. So the same technology may have different outcomes for different businesses. 

What we're seeing is that, as the desire to reduce cost goes up, the desire to invest in conversational AI will go up as well, so you're just going to see more bots in customer service. Hopefully, brands will keep the customer goal in mind, so they can achieve both: they can reduce cost and improve customer experience at the same time. 

I want to talk about the notion of customer service automation. We love to think about chatbots and conversational AI because they're the new, sexy piece of automation. But if you're looking for efficiency, there are lots of different ways that you can invest in automation to drive efficiencies in your customer service processes. 

 

Think of automation as something that goes from low effort to high effort. On the low-effort side, think of automation that helps your customers navigate their way through websites, to the guided balloons that guide you through repeated tasks. Mindless stuff that makes contact center work seem like drudge work can be removed from the agents’ plates, and this allows them to focus on the experiential part of the job that is more enjoyable for them, and more valuable for your customers – through agent augmentation, or agent-assist mode for conversational AI."

 

Automation also applies to the high-effort tasks, like automating a complex process end-to-end by combining all your customer-facing technologies and creating a conversational front-end for the customer. For example, to help a customer in identifying the right insurance, figuring out whether they’re eligible, completing the application, and processing it, all in an agentless manner. All of that is a complex, long-running, and structured process that might use all of the automation technologies that are out there. So it's not just the conversational AI piece that you should be thinking about for these efficiency gains. 

And with that, I'm going to turn it over to Arun who's got a real world story that illustrates how to use these automation technologies to drive new efficiencies while not ignoring the customer experience.

Part 5: Case Study: PhonePe

Arun Mani: I want to emphasize upon fast-growing contact volumes. I'm going to talk about one such company. PhonePe is a payment company based in India, founded in 2005. In five years, they have amassed 225 million users. Just put that in perspective, that's about 35-40 million users per month. And now, due to the crisis, they not only had to deal with scale, but also minimize costs. It was a huge challenge.

 

-

Arun Mani: They've automated almost 75% of the requests that come in. As you can imagine, in the payment business, ticket size goes from anywhere between sub-dollar transactions to thousands of dollars for luxury purchases. They wanted to deliver consistent service. And they were able to go through service requests case-by-case and automate up to 1400 types of different kinds of requests. The bot is able to handle all of this and deliver on 75% of the cases in very fast fashion. So they don't have to actually scale up their service team to accommodate high volumes, thereby saving costs.

Ian touched upon how you can make agents more productive and give them the assistance they need – the agent-assist bot is basically to help them with onboarding new customers rapidly, because that way, individual agents don't have to be well versed with all the complex product use-cases to do so. The interesting part here is the CSAT. They've seen that the CSAT for chatbots is higher than the CSAT for live agents. 

 

"They've automated 75% of the requests they get ... the AI-powered bot is able to handle all this in a very fast fashion. They've seen that the CSAT for chatbots is higher than the CSAT for live agents."

Part 6: New agent roles, remote workforces, and flexible customer service workflows

Arun Mani: The next pillar we need to cover is flexibility. 68% of CFOs think we’re going to have a hybrid workforce model post-covid, and they think it’s going to make the company better in the long run, and the companies need to do what it takes to provide a workable ecosystem for their agents.

If you think about flexibility, there are three angles we need to talk about. The first is flexibility in terms of people, staffing, and skills. The second is the location – are we ever going back to physical offices? The third is processes: What kind of processes will enable business agility? Ian, within this framing, what does the research say?

Ian Jacobs: Here’s what I’ve noticed thus far.

  • New agent roles are emerging: If you think about the PhonePe case study that Arun talked about, what happens with the remaining volume of tickets that was not automated? It goes through to an agent – these are the issues that are too complex for automation to handle, or they may be the kinds of issues where human empathy is required. Automation just isn't appropriate. We're seeing that the nature of work we're asking agents to do starts to change as you increase automation: we start to see fewer tier-1 agents. Tier-3 and tier-4 agents are becoming the norm for contact center work. We're seeing agents who can handle multiple channels, who are empowered to do more, because they're being asked to do different work than what contact centers were doing five years ago. The pandemic has only accelerated this because everybody's investing in automation right now. We're even seeing potentially new jobs. We've had debates about whether it should be called a bot wrangler, or a  bots supervisor. But we have some customers who have done an early-phase pilot of having the agents monitor five or six different chatbot conversations with customers, and then only injecting their human expertise when those conversations threaten to go off the rails. So they're sort of managing and teaching the bot in real time, but also handling the customer interactions proactively. So we see new kinds of work, and therefore new kinds of people are required for that work
  • Remote work might be here to stay: Brands are just not going to go back to the old model of contact centers. And I would argue that social distancing will remain a mandate, even when people are allowed back into work, the real estate and overhead costs dictate that we can't go fully back because it's not like you can magically make your buildings twice as large – especially since business have done a lot of remote hiring during the pandemic. There’s place-flexibility as long as we're talking about easy-to-use digital tools where it's simple for an agent to log on, get their work done in a secure interface, and interact with customers. It doesn't matter if they're on a couch in the living room in an apartment, or if they're in a formal contact center environment.
  • Business are getting flexible with KPIs, processes, and expectations: I'm also going to touch on the process piece. We're not going back to the old world of voice-first. As Arun said, in some industries, call volumes exploded. For others, it was an intraday firefight, where volumes changed day-to-day and there were unpredictable spikes coming in. So, we've seen all of these digital transformation initiatives that were always cited as something that brands wanted to do, but never actually got the investment that they needed to do it. Again, that means the work that you need people to do, and the processes of how you manage these people, need to change. For example: If you are measuring an agent's quality, you would do it very differently if the customer called in, navigated through an IVR, got to the agent that they needed, and the agent handled that interaction, versus a customer who went to your website, navigating through web self-service, had a proactive outreach from a chatbot interaction, and the chatbot escalated that interaction to a human agent. From the customer's point of view, that's the third touch point for that same interaction. So thinking about the process of what agent quality means has to change if you want to get the flexibility that will be required to succeed and thrive in the new world, coming out of this pandemic. I love Arun’s framing of people-place-process – if you start to get flexible with one customer, the others start demanding the same flexibility. If you start to hire people to handle much more complex interactions, they're more likely to expect a flexible workspace because they're much more like a knowledge worker, they operate much like you and I operate in our jobs. And so the process for managing those people needs to change as well. So they're inextricably linked. 

 

Part 7: Case Study: Klarna

Arun Mani: I want to preface this by speaking about balancing the workload across channels during the COVID crisis. There is a feature that is built into Freshdesk, which is what we call an omnichannel router. Regardless of what type of queries are coming in, you may see a spike in one channel, let's say, through chat, versus email or voice. With Freshdesk, you can automatically assign it to a particular person who specializes in that channel. So you can also load-balance to see if any particular agent is overloaded. 

But for this conversation, I want to talk about a client that has used this in a stellar way. And this client is Klarna. Now, Klarna is very popular in Europe. It's a Swedish payment company. They have roughly 85 million customers, predominantly in Europe. And it's growing fast: Fun fact, Snoop Dogg is an investor in Klarna. They've seen a 20x growth in transactions, and they have 1200 agents spread across Finland, Norway, Germany, Austria, Sweden, UK, Denmark, and the US. They also have multilingual and round-the-clock-service for over 85 million customers across 17 countries. 

 

Arun Mani: Klarna saw a big uptick in chat compared to voice (which is a very expensive channel). Once they introduced bots, they were able to push a lot more of the conversations to chat – right now the number is around 66%. Notice the exponential reduction in contact volumes they achieved, as you can see in the slide. Phone calls dropped by 63% to 32% of their total contact volumes, while emails dropped from 16% to 2%.

They were able to use a lot of the features that are built into the Freshdesk platform: They used Intelli-assign, which allowed for assigning requests to the right person based on exactly how they've been performing. They were also able to auto-resolve conversations if visitors stopped interacting with them after a preset period of time. Again, this reduces the rote tasks that Ian was talking about. What’s more, the chats are also continuous, so customers don't have to authenticate every single time. 

Overall, this helps them reduce the cost, improve the customer experience, and enable a distributed workforce across the board. 

Part 8: Live Q&A

1) How do you measure trends in CSAT for conversations that are contained within automation versus a conversation that escalated to an agent? 

Ian Jacobs: Two different answers here. There's the obvious explicit way where you're doing a post-interaction survey. But there are proxies for customer satisfaction, like NPS and others. There's also the implicit way of using analytics. For a call, you're looking at the call transcript, for a chat or messaging session, you're analyzing the chat log. And then it's essentially text analytics that's trying to find the customer sentiment and understanding the agent's performance. But to your second question, we do see some tool vendors who have data that actually show higher satisfaction for interactions that contain both the bot and a human. You can actually have the upfront information collection in the bot. Even if it can't resolve the issue, it still has reduced the amount of time that you need to deal with that ‘annoying’ human contact center agent, which tends to drive higher satisfaction. So although it sounds counterintuitive, we do see the combination of human and AI interactions creating higher CSATs and higher recognition of a good experience. 

Arun Mani: I would like to add one point to that. This is from the Klarna example where they have the bot as a first line. There, the bar was set as a simple survey to see whether the visitor was satisfied with the answer – anytime it went below a 4 or 5, it automatically escalated the conversation to a human agent. In that way, we say the bar for CSAT via bot resolution is set pretty high, and anytime you have any kind of issue, that becomes an automatic escalation point for an agent. 

 

2) Where would you suggest companies to start when creating automated email services?

 Ian Jacobs: Well, I would first actually question why you're focusing just on email, if you're looking at automation, you should be looking at tools that allow you to analyze any text based input and create an automated answer, and essentially be reuse the information that customers are looking for across different channels and touchpoints. But it depends on the industry that you're in, really finding the simple use cases that should not go to a human agent. Even if it went to an agent and they answered it, they would add no value. The answer doesn't vary for simple, straightforward queries like password reset assistance. So look for the easy use cases, where the humans don't add value. You then want to think about how you're going to layer new functionality on top of it as you get better at it. But start with the things that you probably should have had automated via other channels or other touch points before it ever got to an email in the first place.

Arun Mani: Email is a long-form conversation. And it is inherently more complex and tricky for a bot to answer in an email fashion. And the second thing I would add is oftentimes you want to make it very clear to the customer that you're actually interacting with a bot and not a human agent, so there is no intent to actually mimic an email composition. There have been some attempts by companies to do that. And they didn’t succeed – the customers didn’t like it. So the better way would be to either deflect to short-form conversation like chat, where it can be resolved by a bot. If that's not an option, software like Freshdesk can suggest help articles to customers as soon as they’ve sent you an email, assuming an agent is not able to get to the email immediately. A good example is a password reset, the bot will be able to quickly pick up the intent and even identify the resolution that is necessary, and suggest a few things right off the bat, just by analyzing the content of the support email. So there is no waiting period for the customer. And if it doesn’t work, it escalates to the agent, and the agent will also have that visibility to see what has been attempted so far in the interim period. So the agent is not actually suggesting the same solution article the person has read.