Providing Nuance to Sentiment Analysis
For years companies have been doing sentiment analysis to try to figure out if they are loved, liked or despised. To do this well we would combine algorithms from tools with manual evaluations and adjustments. The truth is, the vast majority of comments and input are neutral.
The emoji roll-out will provide a second line of sentiment measurement. However tempting it may be to start using Facebook emoji metrics as guides immediately, companies should lay the groundwork.
- Ignore emojis for a while. Use the first month or two to simply gather data. This will provide a baseline for your organization upon which to base decisions.
- Don’t write to elicit specific emoji reaction. Stay true to who you are. After you’ve gathered enough data, then evaluate the sentiment from emoji buttons to see if there are any surprises. If your brand is warm and fuzzy, but your facebook emoji metrics lean toward sad or angry, then evaluate the content. As an example, a brand that wants to be known for a positive / feel good attitude and social responsibility may post content related to issues in urban areas. These types of stories and elicit feelings ranging from motivating and uplift to very upsetting, often simply based on how it is presented. The emoji evaluation can help a company understand if the style of writing aligns to their image.
- Use the first month or two of “do nothing” data collection on facebook emoji response and compare this to the sentiment analysis that you’ve been doing for years. One of the challenges on the sentiment analysis is that it lacks nuance, an understanding of the dynamics in the middle where the majority of customers reside. The emoji analysis should provide you with the nuance that traditional sentiment analysis lacks.
I’ve been in digital for a lot of years and a constant reaction to change is to jump in and start messing with it. I admit that doing so is a lot of fun. But it also hinders our true understanding of what the new tools or feature really mean. Take the time to let the data educate you, then start acting.