Across 10 products and over 40,000 customers from 176 countries around the world, Freshworks' support teams receive over 3,000 chat conversations every day - ranging from simple L1 queries that can be resolved within a matter of minutes to complex L3 integration queries that often require development effort to resolve.
We take a look at the challenges that have come with scaling chat support for our customers and the underlying processes and support principles that have allowed us to be on top of our game.
The biggest challenge that comes with scaling chat support is being able to provide standardized high-quality support experiences to all customers, irrespective of their subscription status, still without compromising on times to respond and resolve. This becomes particularly challenging with the scope of multiple different products inside the Freshworks ecosystem, where the complexity and types of queries vary.
Users prefer to chat because it is the most easily accessible support channel. Users are able to initiate chats from multiple entry points - using the widget on the website, product-specific support portals, developer documentation pages or from inside each product itself. The first routing condition is triggered here - the source of chat initiation.
When a chat is initiated from the website, it is routed to the Central Chat team. This team handles an equal mix of both support and sales queries. The agent handling the chat ascertains the nature of the support request and hands it off to the relevant team/agent right within the app by changing options on a simple drop-down box.
Unlike, chats initiated on the website, chats initiated from the support portal or inside the product are largely from existing users and tend to be focussed more on product support. These chats are directly routed to the respective product support teams.
As a support philosophy, we do not differentiate chat support for customers based on their subscription status. All users are able to initiate chats and have come to expect the same kind of support. However, when 85% of all customer enquiries for the Freshchat support team are on chat, where response time expectations are much lower, it becomes challenging.
The Priority Inbox feature helps the team render conversations based on the longest wait time for the customers. The support rep is able to focus only on conversations that need their attention immediately and prioritize accordingly. It indicates the time that each conversation has been waiting for a reply, which allows the agent to respond to the most critical conversations without having to think through or manually pick and choose chats. This has helped significantly in maintaining the teams' average first response and resolution times, powering consistent support experiences for all customers alike.
While where the chat is initiated from gives us a decent idea about the nature of the query, Topics in Freshchat’s messenger help further streamline this process. Visitors are able to choose the relevant topic from which they wish to engage depending on the nature of their query. Each Topic is routed to different corresponding teams.
Furthermore, a combination of Assignment Rules and IntelliAssign are in place to ensure the incoming chats are assigned to the right teams and agents to act on. Depending on the time and region from which a chat has been initiated, chats are assigned to agents who are online and available to answer chats.
IntelliAssign ensures no agent is overburdened with volume as this may compromise on the quality of support. It considers the current number of active conversations per agent rather than the total number of conversations assigned and balances the load by distributing accordingly.
The Freshchat support team, in particular, is made up of 20 customer champions. Despite the best of intentions and product understanding, people are unique and bring to the table their own style of communication. However, when communicating as a brand, this posed the challenge of information consistency. To solve for this challenge, a repository of Canned Responses for commonly asked questions were created for agents to use. While this ensured consistency in answers from all agents to similar questions, this also meant that the agents wouldn’t have to waste time typing answers to already answered questions and focus on the ones that really needed their attention.
That said, agents are also able to create their own Canned Responses outside the central repository.
While chat is definitely more convenient, it also invariably takes longer to articulate relatively complex information. The nature of the query determines the number of chat lines it takes to resolve. In some scenarios, it becomes difficult for the agent to articulate the information into a few lines of chat and this is when co-browsing becomes extremely useful.
Agents are able to instantly jump on a screen sharing session and hand-hold the users through their product experience, thus reducing resolution times quite significantly. Also, particularly in the case of trial users, who may not be entirely familiar with the product and need help with minor configurations, Co-Browsing has helped reduce resolution time by 76%.
Jumping on co-browsing calls from right within the conversation has also allowed our support agents to build rapport with the customers while helping them resolve their queries. This had a noticeably positive influence on CSAT ratings.
Freshdesk Integration: While chat accounts for 85% of all support conversations for Freshchat, our support workflows run on Freshdesk. With the unified support and messaging workflow, messages with a higher resolution time on Freshchat can be easily converted to tickets on Freshdesk.
Customers expect quicker responses for L1 queries. And chat as a channel by itself is conducive for fast turnaround times as it provides a way to resolve queries ‘then and there’. And in cases where we are unable to solve, Freshchat allows us to pass chats as tickets into Freshdesk.
Agents are able to add additional information while resolving chats as tickets, such as type of ticket (L1, L2, or L3), ticket status (open, pending, resolved, closed), priority, agent assigned, and other custom fields. This has a positive impact on agent productivity as these fields are auto-populated within Freshdesk and the agent has enough context to work on that particular ticket.
For our support team, Freshchat has helped in realising the dream of being able to deliver timely and adequate support even as we have scaled exponentially in terms of customers and support chat volumes in a short period of time.
We have improved response rates and CSAT ratings. For example, wait times for customers have dropped 42%. We have also identified trends and used this to prioritise knowledge articles and FAQs, which are constantly updated by the team. In conjunction with other support initiatives, we have increased the quality of support we provide - an overall average CSAT of 4.9/5 over the last 12 months, is just the cherry on top.
Sorry, our deep-dive didn’t help. Please try a different search term.