The year is 2022 and the continued rise in popularity and use of social media means your brand can easily and quickly go viral, sometimes for the right reasons but often due to circumstances you can’t fully control. Today, we’ll explore ways to gain that control and own the narrative with a robust approach to social listening.
The elevator pitch? You must be aware of negative conversations, and arguably, at a higher priority than the positive ones. Read on for the extended explanation!
Why every business should consider social listening
Given the traditional nature of the chemicals industry, social listening was not always top of mind for a company like Solvay. However, in 2018 a single tweet relating to our brand went viral. Despite the damage it did to our reputation, it was our “ah ha” moment that opened our eyes to the need of having a setup in place that allows us to monitor mentions and be able to respond fast to conversations considered a threat.
This moment also spurred our Communications department (PR, Brand monitoring, Benchmarking, and Data teams) to come together. This collaboration continues to this day. Our various sub-teams work in Meltwater, executing their own needs and business cases. As we share the platform, the data, and the analysis - we benefit from each other’s work and ensure business-wide consistency.
Social listening enabled us to gain control but also improve internal workflows to match the requirements of the market.
Where do you start?
At Solvay, I began working with our Reputation Manager to list the issues where our reputation could be at stake. Our choice of topics is based on a risk assessment related to (possible) controversies linked to our activities on the basis of NGO and media scrutiny, regulations updates, and scientific knowledge. Following this sort of conclusion, we focus on several areas including social issues, animal welfare, environment, human exposure, terrorism, finance, and C Suite conversations. Opening up this discussion and encouraging participation from other departments meant we kept our eyes open to all issues potentially affecting the business. Working in silos or a fragmented approach to this process might result in the omission of crucial themes.
Once we had our topics confirmed, we aligned on priority and secondary languages used by our audience. The geography of your business, and the languages in which your brand is typically mentioned, will determine the level of complexity you will need to go. For us, it meant working with English content and moving on to translations once we had the setup nailed.
Now we have our topics + languages, then comes the intimidating part… boolean building!
We started the setup with English keywords, as this is our primary business language.
When creating a boolean, think extensively about the keyword selection to cover all conversations surrounding your topic of interest. Variations of words or phrases describing the same thing are crucial. For example, a manufacturing company wanting to track mentions relating to a plant fire, should consider phrases like "site on fire" OR "chemical fire" OR "plant on fire" OR “burning fire”. I would recommend doing research with the Meltwater Explore feature to skim through Twitter results. Searching for extra words that will reveal the way audiences describe such events (screenshot shows us we should consider adding: oxygen or extinguisher to future searches). Do your best to mimic this to be sure you’re covered.
The next step in our process is to test the query using the words we collected + the brand name or variations of the brand (i.e. old names from acquired companies, and misspellings of the company name).
Selecting only Twitter (to start) makes for an easier scan of the content to check if indeed it is about your brand.
Pro tip: if you find that when doing this search, your results are not only about your brand, but maybe competitors too, and you only want to focus on conversations about your brand, use Operators like NEAR/10 to ensure the mention is only about you.
Personally, we did not go for this. Some reasons are that 1) our overall mention count is relatively small being in a B2B space 2) we wanted our results to be wide so we are fully aware.
Once your test is done, be sure to select as many Sources as you wish to track: i.e. Twitter, News, Blogs, etc.
Next step: Alert setup
This is where some troubleshooting will come into play depending on your business. For example, we started with using the Spike Detection alert for each topic, but according to Meltwater's standards, and as I mentioned earlier, our volume of mentions may be lower than the average client, leading the alert not to trigger when it should. In other words, something that we consider to be a spike might not be caught using this alert type.
We decided to move forward instead with Top Reach as we find it works better for smaller quantities, but high reach. To be extra careful, we also have Twitter Influencers launched. When receiving these alerts, it’s very important to be checking the results, especially in the beginning to determine if there are words triggering inaccurate results. For example, the Solvay brand is also used as the name of a University in Belgium. We had to find a way to differentiate us vs. them, and add that keyword variation to a NOT list which is added to the query using a Custom Category.
Once you’ve tested your alerts for about a week or so, you should be ready to add the translations. We started with a quick automated translation of the English keyword list, then asked local speakers at the company to check the translations and add everyday variations or change things that may not be captured by Google, such as cultural differences or local slang.
Pro tip: use the “Language:fr AND” operator in your Boolean building. It might feel like it could be limiting your results, but because many words are the same in multiple languages, it helps to keep the results clean.
The final step is building dedicated dashboards per topic to analyze how these issues evolve.
For Solvay, this was done using the Analyze feature and building custom dashboards with separation by source 1) Social Media and 2) News. If you feel it’s not necessary to build a custom dashboard for your reporting needs, the Explore feature gives you the top-level information you need to remain aware of the issues surrounding your brand.
The process of being prepared for a crisis may be long and feel non-urgent, but you’ll thank yourself later by prioritizing it. Especially by building it at a time when there is nothing going on, as opposed to a moment when you wished your system was in place.
Are you tracking issues at your organization?
Let’s start building a library of keywords to help each other get started on this social listening journey! Add them below and the Meltwater team will add them to the Boolean Library.