Top Seven Social Care Metrics
By Edwin Margulies
If you are a practitioner of social engagement for customer care, you are probably already dealing with the issue of metrics. New to social care? One of the biggest challenges will be to measure the effectiveness of your work and that of your team. Here are the seven top social care metrics you'll need to embrace for a successful operation.
Who Cares about Social Care Metrics?
First, the folks who are writing your paycheck care about performance and metrics are a way to show off your performance. Insights into customer satisfaction, agent productivity, and group efficacy all figure in to budgeting for your business unit. If you do not perform well, it is hard to justify your position. On the other hand, having the right set of metrics will not only justify your position, but also be the basis for bonuses and raises.
Second, customers care. Whether it's verbatim quotes, NPS scores, or continued loyalty, customer feedback has real value. The ability to measure your work's impact on customers is invaluable. Sentiment is one metric that helps to establish what's going on with customers - but there are many more.
Third, individual practitioners care about metrics. That is to say people who have pride in their own work care about how their work can be measured. By paying attention to the essential metrics, you can establish milestones that help you to constantly hone your craft.
What are the Essential Metrics?
There are so many things to count, it is sometimes hard to know where to start. So I've picked seven essential metrics that should help you to get started and perhaps prioritize for your own situation.
An SLA (Service Level Agreement) is a metric that attempts to establish thresholds of performance based on a covenant you have with customers. Keep in mind that SLAs do not always have to link to a hard contract with customers. You can establish SLAs based on shared best practices, group goals, or even corporate milestones.
There are myriad metrics you can use for a Group SLA. For example, the average length of time it takes to respond to social posts is a good metric. This becomes a "group" metric when it is applied across a profile or cluster. If you have eight people working on posts dealing with a specific brand - even a specific issue related to that brand (i.e. support, billing, complaints, etc.) - you can average-out the response time for all of the agents in that group to get a Group SLA. Ditto progress time, queue time and other indicators you may wish to track.
Individual Agent KPIs
A KPI (Key Performance Indicator) is a metric representing contributions by individuals. KPIs are a great way to "compare yourself" to colleagues and of course aid supervisors in giving guidance. In the context of social engagement for customer care, some of the more common metrics for agent KPIs include: 1) The amount of outreach you do each day (replies to posts); 2) The average progress time in researching and preparing for outreach; 3) Outreach volume by profile.
Agent KPIs are manifest in the form of agent home pages or dashboards. These are useful because agents can easily view the stats representing how well they are doing compared to their peers. In more advanced systems, supervisors are also able to see KPI data for each agent. (In fact if you are looking for a social engagement platform, you should consider one that has built-in agent KPI dashboards).
Volume by Sentiment
Volume metrics are arguably one of the most important for a social care initiative. But unless your social engagement for customer care platform is equipped with Natural Language Processing (NLP/NLU), you cannot automatically account for the sentiment of each authors' posts. Without NLP technology, you'd be stuck manually tagging each post for sentiment before you could run a "volume by sentiment" report.
In advanced platforms, you can call-up reports using different intervals. For example, you can take an hourly view, a daily view, a weekly view or a monthly view. On top of this, you can generate reports that break down the volume into separate lines or bars for each sentiment type you are tracking. A red line may signify "Not Happy," a yellow line may signify "Neutral," and a green one for "Happy."
A nice twist on volume by sentiment is to rank brands or profiles side-by-side. This is useful if you want to compare how "hot" certain brands are running. If you are interested in doing competitive comparisons this side-by-side comparison is especially useful.
Volume by Cluster
Anyone who has experience in traditional customer service centers can relate to the idea of Automatic Call Distributor (ACD) Queue reports or DNIS (Dialed Number Identification Service) reports. Here, you can easily arrange volume metrics by business issue or product issue. For example, one ACD-type report could show the volume of calls (or emails or chats) for a certain product. Likewise, there are separate ACD queues for billing, for sales, for complaints, and for technical support. Often, discrete inbound numbers are used to route certain types of calls to specific ACD queues.
In the context of social engagement for customer care, you can still use the same model traditionally used for ACDs with the use of "clusters." These are persistent business issues that can be automatically identified by NLP technology. For example, consider this tweet: "I went over on my data plan this month and my bill is ridiculous. Gotta get it adjusted." The NLP engine can be trained to automatically tag this post as being in the "Billing" cluster. Here's another: "I just hate XYZ Cellular. I am going to drop them and move over to ABC." In this case, the NLP engine may automatically characterize this post as being in the "At Risk" cluster.
Now you can imagine that having the ability to view volume by cluster is very powerful. You can see which "queues" are demanding the most resources and make staffing adjustments accordingly. By tracking traffic by queue you are also able to make forecasting decisions on certain products and services.
The way in which social agents disposition posts and tag them for follow-up can be very useful in establishing workflow milestones. In fact, you can use a rules engine to trigger specific actions based on dispositions. For example, you might automatically route or filter based on posts that have been tagged as: "Transferred to Retention Desk."
Dispositions are simple tags that agents can use to characterize the actions taken on a specific post. These disposition tags can be created and arranged in such a way that they represent workflow milestones. These milestones make for essential metrics because you are able to see the state that all of the work is in. For example, you can see how many posts have already been responded to, which ones have not been worked on yet, and the ones that are in progress. The most modern social engagement platforms allow you to create custom dispositions so you can be as granular as you want in collecting these metrics.
While there is always a benefit in swapping best practices and methods amongst peers in each vertical industry, every enterprise is different in some way. The rules, policies and workflow habits of each customer service center is different. As you can imagine, with the diversity of how each company is set up to work and tackle interactions, customization is a must.
It is imperative that you be able to collect metrics on any kind of custom attribute you can think of. You must be able to change and expand your metrics depending on the needs of your company. A few examples of custom attributes include: 1) Regional Data; 2) Case Management Tiers and Escalation; 3) Reason Codes; and 4) Outcomes.
With custom regional metrics, your agents are able to tag posts for the regions customers are served by. Sometimes, this can be automated based on geographical coordinates embedded in the source network feed. But most of the time, this is determined with agent intervention and dialog with the customer. Once a regional code has been tagged on a post, it is then possible to run reports in which the regional code is used as a filter. For example, you can run a workflow milestone report on the "East Coast" region and compare it to the "Midwest" region. This gives you a handle on bottlenecks and possible regionalized service issues.
Likewise, modern systems allow you to create custom attributes for escalation and case management. This makes it easy to run reports on how many customer issues have been escalated or require the intervention of the retention desk.
Reason Codes and Outcomes are stock-in-trade metrics used by most traditional customer service centers. A reason code is used to identify the reason a customer is contacting the company. A reason might be a product problem, a warranty issue, a personnel complaint, or a product recall just to name a few. Similarly, outcomes are used to establish the "final result" of a communication or series of communications. For example, an outcome might be "extended the warranty" or "waived service fee" or "made the sale."
As you can imagine, based on the particular needs of your enterprise, the number of custom attributes you can create is limitless. But the idea from an essential metrics point of view is to be able to report on and use these metrics. This is why you should look for out-of-the-box reports that allow you to drill down on volume statistics, sentiment, or other filters by custom attribute. For example, you should be able to run a report called "Waived Service Fee" for each brand over time. You could then compare which brands required a service fee waiver over time so trends alert you to specific product problems. This can be particularly useful for product launches or promotions.
Top Voices History
Lastly, I've chosen "top voices" as an essential metric. A top voice is a person who is both important and has influential and meaningful things to say about your brand. There are several ways to characterize a top voice. For example, the number of posts is a good measure to use. Keep in mind that in a social engagement for customer care platform, the number of posts would not include random posts about what kitty is doing or chatting about a beautiful sunset. In a care context, we would be counting only the posts that matched specific keyword or phrase criteria. Bottom line is the number of posts we are talking about are the ones most likely to be relevant.
Another way to characterize a top voice is by their public influence score and also their corporate influence (loyalty or importance rating). There are both proprietary and common instruments for calculating a public influence score. The most well known and popular is a Klout score. Klout takes into account the number of followers, posts, likes, retweets, favorites, and now even Microsoft Bing hits to name a few.
Your top voices metric can also be a combination of several factors. For example, you might give influence a weight of 70% and volume a weight of 30% to establish your own top voice metric. The idea here is to be able to quickly identify what the most important people are saying about your brand. These are the people that other people listen to so tuning-in to them is essential if you are serious about social care.
Whether you're just getting started with social engagement for customer care, or an old pro - it is important to establish and maintain essential metrics to measure your success. I chose these seven based on the priorities of ROI justification, efficiency, and most of all customer loyalty. If you incorporate these metrics into your reporting and analytics suite, you will be well on your way to a more satisfying social engagement practice.