Last week I gave you 5 reasons you need to add a listening post to your marketing strategy. Even if you DON’T have social media as part of your marketing strategy, you need a listening post to “hear” what others are saying about you online and in social networks. I hope you’re now convinced that its important to listen to what customers and prospects are saying. NOW, your problem is you have a TON of unstructured (non-numeric) data and you can’t make heads or tails out of it.
Having a listening post is great, but only if you can analyze data QUICKLY, get it to the RIGHT person for follow-up, and FIX problems identified through online conversations. Gone are the days when you could simply send an email acknowledging the issue and promising someone would resolve it SOON.
This requires a system that captures data (blog posts, Tweets, Updates, YouTube and Flickr, etc) in REAL TIME based on keywords critical for your business, automatically forwards the data to someone empowered to fix it, then tracks the data to ensure a fix was actually implemented (or reassign the data to someone who WILL fix it immediately).
This is a critical first step in effective listening. Keywords should reflect issues critical for the organization. For instance, “bad service”, “unhappy”, “dissatisfied”, and similar terms certainly should be “listened” for. Likewise, the firm should “listen” to compliments from customers as these can improve marketing strategy. For instance, compliments can be used in other marketing communications, to reward employees who provide superior service, even monitor customer satisfaction levels.
Selecting the right keywords ensures the company has the most important information to assess the success of their current marketing strategy, identify failures in that marketing strategy, and suggest changes to the marketing strategy likely to improve performance.
Sure, you can read through the data generated by your listening post and, if you’re a small business, this might be effective. But, if you get lots of results from your listening post, you’ll not be able to evaluate the data, assign it, and track handling of problems identified. You’ll need a hand.
Here are the steps you’ll need to accomplish:
1. Categorize the data
This can be done based on keywords, such that complaints go together and can further be categorized based on product line, store location, etc. Compliments get categorized together and additional keywords generate additional categories.
To be effective, you may have to work extensively with the data to identify relevant categories in the beginning, but later this might be done automatically (using some of the software solutions listed below). Periodically, these categories might need to be re-evaluated to ensure data is categorized properly.
You probably need to set up a system for handling data that is miscategorized in the first pass to assign it to the right category.
Data are then assigned to employees based on categories (this is why categorizing the data is so critical). Thus, complaints might be handled by customer service employees; while compliments go to marketing. Different categories might be assigned to different employees based on product line or store region.
3. Develop a marketing strategy for complaint handling
Policies and procedures need to be established for how employees handle complaints assigned to them. For instance, at Marriott, assigned employees receive an email as soon as a customer complaint is detected and have 24 hours to satisfy the customer. ANY complaints that exceed the 24 hour cut-off are automatically forwarded to the assigned employees manager, who now has 24 hours to resolve the issue. The procedure repeats at the next level if this time limit is exceeded. Employees are empowered to fix problems and their success in reaching a resolution within the allotted time are tracked to use for employee feedback and to ensure the system is working efficiently.
4. Track success
Daily reports must be generated to track success. Reports should contain a count of post entered into the listening post from trolling the internet, a count by category, a count of the number of posts assigned to each employee, and disposition of each post. You might include information on how long elapsed between the employee notification of the problem and its resolution.
Preferably, this information should be contained in a marketing dashboard to aid in improving marketing strategy. A marketing dashboard would contain data over time (preferably graphically) and have relevant divisions such as product line or region. A dashboard makes it easy to see whats going on (for instance, if several posts have NOT been handled yet), especially whether things are improving or getting worse.
5. Make changes to your marketing strategy
Your marketing dashboard will guide you in making changes to improve your marketing strategy. For instance, if the same type of complaint arises with some frequency, a change in marketing strategy is needed to reduce the incidence. A complaint about long lines that occurs frequently may mean the company needs to hire new employees, reduce the amount of time employees spend with each customer, or should consider adding self service technologies to speed customers through the lines.
Complaints may also identify poor performing employees that should be retrained, reassigned, or removed. Complaints might also reflect bad processes (Deming argues that 90% of all failures are related to having the wrong process). However, without the listening post and related analysis of posted data, the company would have no clue as to the existence of a problem or guidance towards its solution.
As mentioned, this process is likely too complicated to accomplish by hand, and while I’m normally a strong advocate for not using automation in social media, this is one place where it is appropriate.
Here are some software solutions I’ve found effective. They vary dramatically in terms of ability and cost, so the benefits of each one should be evaluated relative to the needs of the organization.***** I am NOT compensated for these opinions *****
HyperRESEARCH is a product of ResearchWare. I’ve used this extensively for qualitative research and find it very intuitive, easy to use and powerful. While I’ve never used this for keyword based categorization, it should be able to handle this just fine. One advantage of the software is it can handle audio and visual data (ie. YouTube and Flickr), however this data must be categorized manually (at least in the version I own).
nVIVO is another excellent qualitative analysis software program. I’ve used an earlier version of this software with excellent results. Again, its intuitive and easy to use. Its not as powerful as HyperRESEARCH, but still a good option.
SPSS text analysis software is a new addition to the SPSS family. Since it shares an interface with familiar SPSS software its easy to use and subsequent data analysis can be accomplished using SPSS statistics after converting data to a numeric representation. A drawback is that this software doesn’t handle visual or auditory data.
A common problem with these software solutions is they are NOT built with a listening post in mind, they are more commonly employed in analysis of research data such as surveys. Hence, I’m not sure how successfully or automatically they would work in this context. Certainly, their power for analysis and integration into a dashboard would be excellent, especially for SPSS, which could run regression or cluster or other statistics on data. (The other software packages are limited to frequencies and means, at least in the version I have). How effectively they will integrate with your listening post (which does the data collection and MAY do assignment) is unclear.
If you know of specialized software built to handle a listening post, please let me know.