Profiling Strategies for Social Care Teams
By Edwin Margulies
The careful development of profiles in your social engagement for customer care platform has a profound impact on the efficiency of your social care agents. If your profiles are too broad, you get a lot of spam and unwanted noise. On the other hand, if they are too tight, you may miss the opportunity to help a lot of customers. Here are some tips on fine-tuning your profiles for optimal performance.
What is a Profile?
A profile is defined by key word searches you set up in your social engagement for customer care platform. The key words and phrases that you set up as your "social search" criteria act as a dial on a radio, so you can tune in to social posts dealing with a certain subject or brand.
For example, if you run a supermarket chain called Acme Market, you could set up a profile called: "Produce."
The keywords and boolean operands could be:
("Acme Market" OR "Acme") AND (Produce OR Fruit OR Vegetables OR Veggies)
With this example, the system would fetch all mentions of Acme or Acme Market in which any of the words, produce, fruit, vegetables, or veggies was mentioned.
Similarly, you can set up a second profile using keywords and phrases dealing with "Meat" or "Delicatessen."
Some social care professionals also set up mirror images of these self-profiles dealing with competitors. This is a way to mine the social cloud for sales opportunities or competitive trends.
Do a Dry Run First
It is easy to get carried away when you first set up profiles for a social engagement for customer care platform. Some folks have a tendency to set up too many profiles which can make things harder to manage and also more expensive. To avoid this, it is a best practice to do some dry runs first.
For example you can test certain profiles and key word combinations and then simply eyeball the results coming back to see if you are getting too many posts, or not enough.
A good first test is to simply type in the name of your company or brand and see how many posts come back. If you do not get too many mentions, that means your brand is not garnering a lot of buzz, positive or negative. You may then experiment with mentioning more than one brand or even other companies in order to bring in more data.
If you get too many to possibly read, that is a sign that you need to tighten-up the search criteria on the next try.
Generally, speaking, it is a good idea to start first with a broad search criteria, and then to incrementally tighten it down on subsequent tries.
Once you have established the brand and basic subject (i.e. Acme & Produce), you can start to play with qualifying phrases and words. For example, you could create a profile that is searching for social mentions dealing with customer service or quality in the produce department. To do this you might want to add to the search words or phrases like:
… AND (Spoiled OR "not fresh" OR rotten OR wilted OR ruined OR Bad)
By adding this to the original ("Acme Market" OR "Acme") AND (Produce OR Fruit OR Vegetables OR Veggies) you can create a profile that scans for poor quality produce mentions. This will yield posts in which people are observing or complaining about the quality of the merchandise.
All-in-One vs. Specialized Profiles
It is a good idea to have multiple profiles because you will later be able to do reporting by profile and it's nice to understand how each profile is trending from a sentiment standpoint, and also in SLA (Service Level Agreement) and in terms of resolution.
You may want to just do an all-in-one profile that takes in all the key words and phrases you contemplate for your company, but you will be giving up a lot of control in terms of specialized routing, tagging, and reporting if you do this.
Before you opt for an all-in-one profile strategy, consider three or four broad profiles all sharing the same company or brand attributes. For example, one for billing, one for sales, and one for service issues. Alternately, you could profile by product grouping.
Layering on Rules and Triggers
Each profile you establish drags through hundreds even thousands of social posts. Once you've fine-tuned the key word and search criteria, it's a good idea to characterize your posts as well. Here are some post attributes that can be used as the basis for rule triggers:
- Time of publication
- Source (Blog, Twitter, Facebook, etc.)
- Public Influence Score of Author
- Sentiment of Author
- Cluster (Conversation Topic)
- Existing conversation thread in system
Once you have established these attributes, you can apply them to your rules engine. For example, you may want to trigger rules on all authors who have an unhappy sentiment who are talking about a hot topic you are tracking. Some platforms will even allow you to automatically route or disposition posts based on certain criteria.
If you do not have a rules-based platform, try to at least "tag" the post with these attributes so your social care team can take action based on pre-determined business rules that you set up.
Dry run testing of profile key words and search criteria goes a long way to making your social care agents more efficient and effective. Make fine-tuning of your profiles an ongoing process so you can get more accurate and impactful results. It may take a little tinkering but making incremental improvements in your profiles' make-up will ensure a happier, more productive social care team and in turn, a happier customer base.