Sentiment Analysis Saves the Day for Airlines

Has your flight ever been delayed? Or cancelled? Or your luggage lost? Chances are that, like most air travelers, you have experienced these issues more than once. And chances are that they have caused you no small amount of frustration. As much as air travel is convenient and a given in today’s interconnected world, it is not without bumps and those headaches are not always due of inclement weather. Some issues are beyond the control of airlines and customers are generally understanding when it’s not the airlines fault. For other issues, not so much.

In any competitive industry, customer satisfaction is a must. But with today’s travelers often turning to social media, especially Twitter and forums, to vent their frustrations, there is now additional time pressure on airlines to detect and mitigate any issues as soon as possible. That is, before they lose customers or worse, before an incident turns into a PR nightmare. As seen in recent past, mishandled incidents like a passenger being removed from a flight or lost or even killed family pets can go viral in no time. But how can issues be detected in real-time amid the vast volumes of social media content?

There is a wide variety of issues to watch out for. Here are some examples of complaints (airline names redacted):

• Price (“I hate flying <AIRLINE1>. $79 for a bloody window seat. Get me back to <AIRLINE2>”, “$1700 is not reasonable to charge for a flight change…”, “What is wrong with these prices? One seat shouldn’t be $400″)
• Lost luggage (“@<AIRLINE> lost my luggage last night”, “How is it even possible for <AIRLINE> to lose a bag on a direct flight?!”)
• Delays (“flight is delayed again”, “@<AIRLINE> flight now delayed for 4th time at sfo. Traveling with kids = never again on <AIRLINE>.”)
• Leg room (“Less leg room on @<AIRLINE> planes.”)
• Customer service (“Awful service”)
• Amenities (“On a <AIRLINE> flight now. No more free movies, $10 for Wifi”)
• Unassigned seats (“First time flying <AIRLINE> and its giving me major anxiety. Watching the clock like a hawk. Much rather pick my seat in advanced.Thank you”, “Not sure I like <AIRLINE> format of boarding.”)
• Never flying again (“@<AIRLINE> never again am I flying <AIRLINE> airlines ….”, “I will never use them again for int’l.”, “never flyin @<AIRLINE> again”)

Luckily it’s not all complaints. There are praises too. For instance:

• Lounge (“Thanks for the pre-flight warm up @<AIRLINE> #<AIRLINE> #vegasbaby”)
• Customer service (“@<AIRLINE> Just wanted to say thank you to the two stewardess’ on flight 749 from MIA to ATL earlier today. Above and beyond 4 a sick passenger”; “My flight w/ <AIRLINE2> was cancelled &amp; the agent from <AIRLINE> saved me &amp; rebooked on their flight”, “Service so good I thought I was on an international flight!”)
• Leg room (“Holy leg room!! Thank you @<AIRLINE> #economycomfort really makes a difference!”, “Lots of legroom this morning on <AIRLINE>”)
• Upgrades (“Finally made a flight home. First class too! <AIRLINE> treats me quite well.”)

While customer care and brand protection are primary reasons for airlines to monitor social media, there is certainly more value that can be gleaned. Social media is also rich in insights regarding competition and market preferences to better understand what your customer base wants. Social media can help airlines understand what their strengths and weaknesses are, what differentiates them from others, and what new industry trends are shaping up. For instance:

• pricing (“It’s hard to remain loyal to @<AIRLINE> when fares are $100 or more higher than competitors for the same flight. Guess I’ll be flying <AIRLINE>..”).
• flight booking (“Why is it so hard to book a flight with anyone other than @<AIRLINE>?””)
• preference for certain flights (“I’m never flying any airline other than <AIRLINE>for a regional flight.”)

How does one go about finding all these insights in social media?

Sentiment Analysis: a Must in the Age of Social Media

It’s not just about finding negative and positive language. Figuring out what exactly travelers like or dislike requires deeper analysis. With so many comments being generated in real time, a manual analysis is just not feasible or cost-effective. There is no simple automated solution either because language is rich and creative and there are myriad ways of expressing opinions, likes, and dislikes.  Here is where an AI-powered solution comes in handy.

Sentiment Analysis, also known as opinion mining, is about detecting likes, dislikes, emotions, and opinions and it’s often applied to social media, reviews, blogs, forum posts, chats, and similar sources.  Sentiment Analysis has many applications across many industries, but it’s especially well-suited for gathering insights and trends for airlines from social media.

Here is how:

1. Likes and Dislikes. Sentiment Analysis detects positive and negative sentiment. At a very basic level, it identifies positive and negative language and may associate those with co-occurring or high-frequency terms.  At an advanced level, through Entity- and Aspect-based Sentiment Analysis, it pinpoints the specific aspects that those positive and negative sentiments are about. For instance, travelers may love an airline’s leg room but complain about high seat prices or frequent flight cancellations. Being able to pinpoint what exactly people like and dislike is critical insight.
2. Intent. Advanced Sentiment Analysis detects language suggesting intent and sentiment-based actions like positive and negative recommendations, or worse, intent to never fly again or boycott.  Detecting those intended actions can allow you to address the root causes of any perceived issues both for a specific customer and potentially for other travelers in similar situations.
3. Large Data Sample. Advanced Sentiment Analysis can scale to Big Data size, making it possible to process massive amounts of data in real time. Naturally, the larger the amount of data and the faster it can be processed, the more reliable, useful, and timely the insights will be.
4. Analytics. Advanced Sentiment Analysis normalizes the extracted information thus enabling data aggregation and quantification to provide the overall view from a large collection of content. A dashboard presents multiple views of the sentiment information through various types of interactive graphs and charts. Sentiment data can be sliced and diced as desired, for instance by positive and negative aspects (e.g., customer service, price, on time departures), and, if desired, the user can drill down to the source text for inspection and further analysis. Other useful charts show sentiment evolution over time through a sentiment timeline.
5. Competitive Analysis. The same type of analysis (e.g., likes, dislikes, intent) can be performed on competitors providing critical insight into an airlines’ strengths and weaknesses in the marketplace and helping understand where your airline ranks compared to peers.

To summarize, NetOwl’s advanced Sentiment Analysis can analyze social media quantitatively and qualitatively to provide critical insight to thrive in the competitive airline industry.