Support organizations are often asked to maintain quality and efficiency while keeping costs low.
Paying an hourly rate for internal and/or outsourced Agents can add up over time, especially when you need to add more to reduce a sudden backlog of Issues. Running a successful and cost-efficient support operation requires you to reduce the number of new Issues, while at the same time maximizing each Agent’s output. Helpshift is designed to support you with achieving those initiatives but are you confused as to where to start?
In this multi-part series we are going to show you how to identify your inefficiencies with Power BI, followed by subsequent content that will outline key Dashboard features that you can use to improve them. Below are shortcuts to help you easily navigate different topic areas:
If you don’t know what the problem is, how can you solve it? We have designed our Power BI Analytics reports so that you can measure key metrics, identify Agent inefficiencies, and easily track your SLAs. Power BI is a business intelligence tool, so the vast amount of data may seem overwhelming. That’s why we are focusing on a few important ones to get you started:
- Reduce number of new Issues
- Increase quantity of Issues that Agents can handle during their shifts
- Automate repetitive tasks of tagging and following up
Good FAQs Can Lead to Fewer Issues Reported
Helpshift is designed to deflect up to 90 percent of reported Issues. This can be achieved through self-service if your users read at least one FAQ before reporting an Issue. To encourage self-service, you should have a good repository of FAQs that are relevant, current, and easy to read. The support manager should also be reviewing FAQ analytics on a weekly basis to identify areas of improvement.
If your customers have to write to you, there’s a good chance they are already frustrated with their most recent experience with your brand. Customers file a new Issue because they couldn’t find the answer themselves in your app or in the FAQ section. While you may already have a large number of FAQs, when was the last time you viewed feedback on the articles?
With the new FAQ Analytics report in Power BI, you can measure the effect your FAQs have on reducing overall Issue volume.
You’ll also want to go one step further and identify the FAQs that your customers are flagging as “not helpful.” This means that you may want to review your articles to see if they need to be updated.
Helpshift’s most successful customers achieve at least a 94 percent in Issue deflection by having an expansive and frequently-updated FAQ library. Other customers have found higher FAQ engagement rates by incorporating videos and images into each article.
Another way to identify which FAQs to focus on is by seeing which FAQs your Agents are inserting. If they are frequently inserting a particular unpublished FAQ, you may want to turn it into a published FAQ so your customers can find answers without having to waste your Agents’ time. If a published FAQ is being inserted often, then that means your customers aren’t finding it in your FAQ library or they aren’t comprehending it well enough. Take a look at the content as it may need to be renamed, refreshed, and reorganized.
Now that you’ve learned how to improve your FAQs and encourage self-service, you can move on to looking at how your Agents can resolve more Issues per hour, at a quicker rate. If you optimize your workflow, then you should be able to maintain or reduce your Agent headcount and time spent on each Issue. This translates to immediate cost-savings.
Before we dive into the metrics that measure Agent efficiency, let’s start with a basic understanding of the actions taken by an Agent and how they correlate to your operating costs. If an Issue is new or waiting for the Agent to reply, this is a cost to your operations. If you are waiting on the customer to reply back, then an Agent doesn’t need to work on that Issue until the customer responds. Every time an Agent touches or works on an Issue, this adds to your cost, and this cost continues to increase the longer an Issue is waiting idly.
Therefore, your goal should be to have your Agents handle more Issues, with the fewest touch points, in the shortest amount of time. This might seem difficult to achieve, but once you master Power BI reports, you will be able to identify exactly what you need to do and how to do it.
To improve Agent efficiency, we’ll focus on the following metrics and make them our KPIs:
- Time to First Response—how long does it take your Agents to first reply to the customer
- Holding Time—how long are your Agents assigned Issues before responding
- Outbound Responses—how many messages are your Agents sending per Issue
- CSAT—how your customers rate the level of service by your Agents
- Reopen Rate—how often customers reject the resolution provided by Agents
Power BI will enable you to see how your Agents’ performance rank against each other. This way you can promote top performing Agents, and address areas of improvement with Agents who need them.
How do our KPIs above relate to each other? Are higher numbers better? The values above are important to consider every step of the way from Issue creation to resolution. All businesses are different, but this is a great starting point if you haven’t thought about which metrics to prioritize.
Time to First Response: The lower the number, the better. If this KPI is on the lower end, that means that new Issues are being quickly distributed and there isn’t a large backlog that would prevent Agents from responding to new Issues.
Holding Time: When analyzing how long it takes Agents to reply back to the customer throughout a conversation, a lower number is better because it means that your Agents quickly send replies back to the customer. If this number if high, then this could be caused by Agents taking too long to investigate the Issue or come up with a response to be sent to customer.
Outbound Responses: This is the number of messages Agents send to your customers, and this can have an impact on Holding Time. Since Outbound Responses are a top time-consuming task for an Agent (time-consuming = high cost), then you’ll want to keep this number fairly low. If Agents take five replies to resolve an Issue instead of three, then that means additional minutes or hours are being spent on one Issue when they could be spent on another one during their shifts. If this number is very large for certain Agents, then you will want to review their replies and check if they lack substance or a clear solution that customers can comprehend.
Remember that our goal is for Agents to use their time as efficiently as possible during their working hours. So diverting two outbound responses to other open Issues allows your Agents to help more customers in the same time period.
Customer Satisfaction (CSAT): A higher score is better as this is a good indicator of whether your customers think Agents were helpful or not. While Agents can save time by sending three replies instead of five, it’s worthless if your customers don’t get their questions answered properly. For CSAT, you’ll want to aim for higher numbers, and anything above 4.3 is decent. If your Agents’ scores starts dropping below 4, then you should start spot-checking their Issues with low scores. You may also want to work with them on improvement plans.
Reopen Rate: This data point shows how often customers reopen an Agent’s resolution because they don’t accept the solution. Agents should strive for a low Reopen Rate, which shows that quality responses are being accepted often by customers. If Agents have high Reopen Rates they may also have a low CSAT score, as both KPIs may indicate poor quality of work and customer experience.
We’ve now covered how to do #1 (reducing number of new Issues) and #2 (increasing quantity of Issues that Agents can handle) that should help lower your operating costs. Let's move on to our new Workflow Features (New Issue Automations, Time-based Automations, and Queues), and how they can help you improve Agent performance KPIs.
We just covered how to identify inefficiencies in your Support Team. Using Power BI Analytics reports, we showed you how we improve FAQ content as well as identify which metrics are important to make our team’s KPIs. We want you to be successful, and help you learn how to use our new features so you can improve your KPIs too.
Now we’ll be going over Helpshift’s new advanced workflow features. This new toolset includes Automations, Queues, and Teams. Using them correctly will allow you to improve your KPIs and reduce inefficiencies in your team. This is where your true cost-savings will happen!
One of our customers using Queues has seen great improvement in top spending users. This customer saw a 70 percent improvement in Holding Time for those users. Thanks to Queues, users were responded to faster, and Agents took less time replying back to them.
Another customer on Queues has seen an 18 percent improvement in Time to First Response for VIP mobile users from month to month. This translates into a 1.55 hour difference in this KPI, which means that the highest spending customers are being responded to faster. This customer even had an additional 500 Issues to handle in a recent month, but still had better KPIs because of Queues.
Before we dive deeper, it’s important that we understand when and why we should use these features. These advanced features are composed of algorithms, logic building, routing rules, and load balancing. Your goal should be to have the Helpshift system, not the Agent, complete as many repetitive tasks as possible. Pay your agents to do actually solve problems with your product/service, not to do repetitive tasks.
Your Agents’ repetitive tasks can be done by an Automation. These tasks include:
- Classifying Issues with Tags
- Assigning Issues to Agents who are online
- Sending a reply to customers to ask for more information
- Following up with customers who haven’t responded in a period of time
Use these features to replace an Agent when the task is objective and has a standard rule set. For example, you can have an automatic follow-up be triggered whenever it’s been three hours and an Agent hasn’t heard back from the customer.
There are three components to automating the classification and routing of Issues:
- New Issue & Time-based Automations
- Teams & Groups
You can use these new features in a basic or advanced way, which is great if you have a simple or complex workflow. At the most basic level a customer submits a new Issue and a New Issue Automation identifies/classifies it to be assigned to a specific Queue.
Queues have a built-in backlog management, load balancing, and an option to auto-assign Issues to available Agents. This ensures your customers who have been waiting the longest aren’t forgotten and your Agents don’t get overwhelmed by a large number of Issues assigned to them. You will have SLAs for certain users and issue types. By using Time-based Automations, you’ll be able to make sure that those SLAs are met or that managers are notified if they are missed.
Your Agents are assigned to different Teams and Groups. Depending upon their skills and responsibilities, Agents are given access to select Queues so they can reply to Issues. You can have multiple Queues that are ranked by priority, which allows certain Issues to be assigned to Agents before others. This way you are able to always prioritize responding to your paying customers over your free ones.
Let’s set up a simple automated workflow!
New Issue Automation: This will determine what will happen to New Issues from certain channels (mobile, email), during certain times (business hours), and various users or problems (paying users, billing). If an Issue matches your defined conditions, then you will decide what happens to that Issue. In most cases, you’ll have it assigned to a specific Queue. In this New Issue Automation, you’ll also be able to add an auto-reply to the customer (acknowledgement message) as well as add additional Tags for tracking purposes.
Queues: This is the destination for Issues that have been assigned by your previously created New Issue Automation. We recommend you setup your Queues based on common grouping (language, user type) as well as rank them on a priority system (paid vs free users). You have the option to enable Auto-assignment in a Queue. When enabled, each Issue in a Queue is distributed via a round-robin method to whichever Agent is added to that Queue. Our system also checks for Agent availability, so if Agents aren’t online or if they don’t have a stable network connection, then Issues won’t be assigned to them.
Auto-assignment: This Queues specific feature prioritizes any backlog of Issues as well as prevents assignment if Agents have reached the set maximum number of Issues assigned to them. We recommend enabling Auto-assignment for your Queues where you want short SLAs. You can have Queues without Auto-assignment for Issues that have longer SLAs. The Auto-assignment feature also reassigns Issues back to a different Queue if Agents log off and still have Issues assigned to them. This way your customers aren’t impacted by shift changes for Agents.
Multiple Queues allow you to dictate a priority of service and assignment for your business. You no longer have to train your Agents to work out of Smart Views and reply to paying customers first! Our priority Queues system automatically assigns any paying customer Issue to available Agents before Issues from lower priority Queues, like free customers. So if you are struggling with high Holding Time KPI, then you should see an immediate gain by implementing Priority Queues.
So how do you decide which Agents should have access to each Queue? What’s the best way to group and structure your Agents?
Teams: This is how you organize your Agents based upon skill and qualifications. In the Helpshift Dashboard, you can set a maximum number of Issues for each Team. Higher skilled Agents on one Team may have a higher maximum number of Issues than those on a lower skilled Team. Additionally your maximum number of Issues might correlate to live chat vs longer SLA communication.
Each Team can include subgroups for language (French, Japanese) as well as training on investigating complex issues (bugs, billing). If you are managing a large number of Agents, then you may even have an outsourced team through a third-party provider. You will want to group those Agents together into one Team.
Even with all of these New Issue Automations, Queues, and Teams, you may still be relying on Agents to spend time on following up or other repetitive tasks that can be done automatically. That’s where Time-based Automations come in.
Time-based Automations: This feature allows you to automate many tasks that Supervisors and Agents would normally do. Do you currently tell your Supervisors or Managers to start answering Issues when there is a backlog? You can set up a Time-based Automation to automatically notify Supervisors or other Admins if there is a backlog in certain Queues. Are your Agents forgetting to follow-up with customers if they haven’t heard back from them within a few hours? A Time-based Automation can send an auto-reply if the customer hasn’t replied back to an open Issue in 30 seconds, five minutes, four hours, or whichever time you specify.
Time-based Automations will also improve your Holding Time and Time to Resolve KPIs as you’ll be able to follow-up and resolve Issues automatically. By having an Automation handle these outbound messages, your Agents can spend time on Issues that need their attention.
You can also use these types of Automations to reassign Agents’ Issues back to designated Queues when they log off for their shift. This way your KPIs won’t be negatively impacted by forgetful Agents.
In today’s mobile-connected world, your customers use smartphones to communicate with each other via messaging apps like Whatsapp, iMessage, WeChat, and more. If two people are active in a conversation, then they reply back and forth in the form of a live chat rather than asynchronous messaging (like email). Wouldn’t it be great if you could choose whether you want that experience to power customer support for your business?
With our latest SDK releases (iOS, Android), you now have the ability to offer a live chat experience to some or all of your customers. We’ve introduced a new option, Typing Indicator (TI), that allows your customers to see when an Agent is typing a response This is similar to what you see when you are waiting for your friend to respond on iMessage or Whatsapp.
Expectations for Customers
In order to offer a live chat experience, you’ll need to configure SDK flags so your customers have less friction when it comes to starting a conversation. For this example, we are going to assume that you are offering this live chat option for all of your customers.
After you've set up your SDK and app experience, you’ll need to decide what happens when that first Issue gets created. Should it get routed and assigned to a particular group or Agent? What is the maximum number of chat Issues that your Agents should handle at a time?
New Issue Automations: You’ll want to set up one for Business Hours and one for non-Business Hours. Each of those Automations will assign the new Issues to a live chat Queue. You should have a separate Team for live chat where you set the capacity for those Agents at no more than 3 Issues.
Queues: Setup a Queue called “Live Chat” that is your #1 in priority. You can add the Agents on your Live Chat Team as well as any backup Agents to that Queue. This way you have backup coverage if any live chat Agents are busy or on a break. Remember to add a fallback message of “All Agents are busy” if a new Issue comes in and all Agents are at capacity or offline.
In the Auto-Assignment section, you’ll want to set a maximum of 3-4 Issues for your Team and make sure that the checkbox of “Exclude Agent Reply” is not checked. This will allow open chat Issues to remain assigned to Agents if they are waiting for your customers to reply. If you also have a Team of Agents that doesn’t handle live chat Issues, then you would check this box for an asynchronous messaging experience. In that instance, your customers and Agents know there will be a delay of hours or days for replying back to the Issue. Usually this will be for lower priority Issues with a longer SLA.
Offering a live chat experience means you have to have short SLAs so your customers aren’t upset when Agents don’t reply to their messages in a matter of seconds. Live chat is instant, and this is the expectation you are setting for your customers when they submit their first message. By using Time-Based Automations, you can flag SLA violations, send auto-replies, and alert Supervisors when there is a backlog.
So if your Agents have been assigned a new Issue, but haven’t replied to the customers in 30 seconds, then those Issues will be tagged with “SLA violation.” When you view your daily and weekly analytics in Power BI, you can see how many of your Agents didn’t hit the SLAs you set.
If your customers haven’t sent a reply in 45 seconds during an in-progress conversation, then this auto-reply will be triggered. You can run the auto-reply Automation first and then follow it up with the auto-reply and close Automation after. Just make sure that you send the second one to be at 60 seconds so there is a 15 second gap between them running.
If your Agents are at capacity and it’s been 45 seconds since the Issue was created, then a Private Note will be added to the Issue notifying your Support Supervisor that there is a Backlog. For analytics purposes you’ll be able to track when backlogs occur and how many Issues made it to the backlog.
KPIs to Focus On
If you are offering a live chat experience, these are the essential KPIs and recommended data points you want to strive for:
- Time to First Response—30 seconds
- Maximum Chats per Agent—3 Issues
- Holding Time—20 seconds. (We recommend your Agents update your customer consistently if it’s taking longer for the Agent to investigate the problem)
Want to Learn More?
If you want to offer a live chat experience, please reach out to your Account Manager to get started.