I was recently working on a task for my internship that involved converting raw guest feedback into intelligent, visualized information that could then be used as a benchmark for continuous improvement. Most of the data included responses on a scale of 1-5 to questions like "how satisfied were you with the service/pool/bar etc." While these are straightforward, the questionnaire also asked hotel guests for comments that are more difficult to quantify. A quick search led me to this post which highlights Google's Cloud Natural Language that can perform syntax analysis, sentiment analysis and entity analysis. The example in the post analyzed over 400 Airbnb reviews for a listing in Boston and produced an analysis with entity, sentiment score and number of mentions. On the positive side, we can infer that guests like the location, value, stay and Sean (presumably the host), while on the negative side, guests didn't enjoy the noise, mattress, parking etc.
I wanted to apply the same to Google reviews for EIA, but this involves coding that is beyond my capabilities. Looking for help within my social circle didn't lead to much either, so I decided to go about it the long way - sourcing the data, analyzing each review for sentiment, categorizing key elements and counting the number of instances of each. My intent was to distill over 500 reviews of EIA into positive and negative sentiments so I could understand what travellers thought of Edmonton International, my favourite airport.
Unfortunately, there there aren't any open data sets for Google reviews. I tried using a web scrapping tool like Import.io to easily pull in the information but this didn't work either. I realized I had to individually copy and paste the information into Excel which wasn't as tedious as it sounds. I only had to scroll down far enough until all the reviews were loaded, copy and paste into Excel and then clean up the data. I imagine Google provides businesses with all of this data through My Business, but unfortunately I did not have this luxury.
Of the 517 reviews, only 295 included any text with the rest just being star ratings. At first I figured I could break each sentence down to its words, count frequency and be done, but realized that a word by itself doesn't offer much meaning. This approach also fails to take into account the importance of context; the word "good" in a sentence can signal a positive experience, but one that contains "not good" would mean the opposite. Context is key.
With no other tools at my disposal, my best approach was to read each review, understand the sentiment being conveyed and determine its frequency. I came up with a broad bucket of categories as defined below.
- Overall Experience
- Baggage Services
- Retail and Amenities
Further notes about the process:
- Words that conveyed the same meaning - what I call sentiment redundancy - were grouped together. For example, friendly staff/helpful staff more or less mean the same and I didn't make a distinction between these
- Some reviews may be outdated and may no longer apply, but I have included them regardless. For example, one comment was on the lack of Uber service, but the airport has since allowed Uber
- Reviews current as of June 4, 2018
The data shown above is interactive. Hovering over the bar chart will give you an indication of the number of instances for each comment. The filters also allow you to display entries based on category, frequency and whether the comment was in a positive or negative context. Not surprisingly, the highest frequency at 52 is a catchall entry that captures all those comments that simply state the airport is nice or excellent or great or best. You can make the argument that "good" at 20 times is similar, but I chose to distinguish between them to provide some variety.
Filtering out the results to only display those under overall experience we see that there are:
- 44 instances of reviews where the airport is described as being convenient, easy to navigate and understand with no long lines and 4 instances describing the airport as easy to get to and well connected. The flip side sees comments stating the airport is small (4) with an under-connected terminal (1), not enough moving sidewalks (1) and poor connectivity to the city's downtown (7)
- 25 instances describe the airport as being clean and another 4 stating the washrooms are clean. Compare this to 1 instance of "dirty washroom", another of "driving area messy" and a third of poorly cleaned walks in the winter
- 23 comments praise the staff as friendly/helpful and 8 appreciate the airport's level of service. Contrast this with comments of poor service (2), poor management (2), rude or unhelpful staff (1) and staff not being easy to communicate to (1)
- 10 reviews appreciate the fast and free WiFi compared to 7 that have WiFi issues (slow, poor coverage, connectivity) and 2 that indicate the WiFi isn't secure (go figure - it's a public hotspot, after all)
- 11 that think the airport is good sized and spacious, 10 that like that "it's not that busy" and 5 that describe it as utilitarian, efficient and functional. On the negative, 6 instances urge the development of more routes and 3 describe it as boring.
For retail and amenities we note:
- 36 instances of reviewers who are pleased with the variety and quality of food available, another 16 who enjoy the many store options and 3 who specifically mention the Belgium Beer Cafe. On the negative, travellers complain about the poor food choices at the US terminal (2), the lack of services before passing security (2) and the limited and horrible food options (2). Other reviews ask for healthier food options (1), better coffee (1) and better Canadian concessionaires in the US terminal (1)
- 2 instances love the new Premium Outlet Mall and 1 unfortunate instance of a traveller wanting to check out the mall on a layover but being unable to store their luggage
For these services we see 20 reviews that describe going through security, customs and immigration as fast, 2 reviews call personnel friendly and 1 specifically states immigration as helpful. On the flip side, unhappy reviewers describe busy and slow lines (11), always being selected for pat downs (3) and issues with security personnel or procedure (3).
Rather uneventful on the parking side - 3 comments indicate the airport has good parking options and 1 that appreciates the Uber service. On the negative, comments range from not having enough parking at arrivals (2), to parking being expensive (2) and having too many spots reserved for Lexus-only cars (2).
Check-In & Baggage
Surprisingly, there weren't a lot of comments, especially on the negative side. I anticipated more complaints but could only find 2 on damaged baggage, 1 on slow turnaround of bags after deplaning and 3 on problems with check-in.
I undertook this exercise to get an idea of what travellers thought of EIA, both positive and negative. I figured most of the gripes would be about the usual things people complain about when flying - the check-in process, long line ups, missing baggage. Indeed, the majority of the negative comments were about busy/slow lines. Next up, surprisingly, were issues with WiFi (I've never had problems with the WiFi and EIA does offer faster WiFi if you sign up for their free rewards program.) People also complain about poor connectivity to the airport due to limited transportation and overpriced cab services. Hopefully these are somewhat alleviated with news on Uber being allowed at the airport and funding for ETS' 747 bus route. Unfortunately, LRT service to the airport is still decades away.
On the plus side, passengers love travelling through EIA thanks to its friendly staff, navigability and the many retail and food options. Assuming most of the passengers are local Edmontonians, I found it amusing we describe our airport as utilitarian, functional, efficient and "not that busy"; perhaps it's that Canadian trait of not being too bashful or perhaps it's because Canada's next busiest airport and our rival to the south, Calgary, handles around twice as many passengers. Yes, Edmonton lacks the number of direct routes that Calgary offers, but Calgary has geography, Westjet and is a hub for Air Canada.
I also compared Google reviews for Canada's 8 busiest airports to see how they're doing relative to each other. What's interesting is the "Delta" column which is just the difference between an airport's share of reviews and its share of traffic. I've assumed, of course, that these 8 airports represent 100% of the reviews and traffic. While most of the airports are in line, Calgary lags by 2.77% and Edmonton by 1.08%. Montreal, on the other hand, leads by 3.3%; YUL is 14.33% of passenger traffic but accounts for 17.62% of reviews. The top 4 airports all score below 4 with the exception of Vancouver. Of course, the busier an airport is the more once can expect things to go wrong and the harder it is to please everyone. It would be interesting to determine an "adjusted" score after correcting for number of reviews or annual traffic. I'm not sure how the two are related, if at all.
Improving passenger experience is a difficult yet crucial task for airports. Difficult because so much is affected by external stakeholders (airlines, retailers, CBSA etc). Important because according to an estimate by the ACI, an increase in passenger satisfaction generates an average growth of 1.5% in non-aeronautical revenues at airports. "Gate anxiety" is very real and EIA has done plenty to address this concern. The airport has implemented Primary Inspection Kiosks (PIK), allowed Uber at the airport and a installed new 4K HDR LED display to show travellers key information. The reviews seem to echo the efforts made by EIA as travellers enjoy the airport for its ease of use, fast-moving lines, friendly staff, free WiFi and many retail and food options. Kudos to the EIA team.