We conducted a semantic word analysis using over 18k reviews to uncover the world’s most disappointing artistic masterpiece.
We could have never predicted the impact this campaign would have.
We set out to investigate the gap between expectation and reality in art tourism by scraping 18,176 reviews of the world’s top 100 most famous artworks.
These reviews, which mention iconic pieces by Da Vinci, Van Gogh, Rembrandt, Warhol, etc. helped us uncover which art experiences left visitors disappointed and which exceeded expectations.
Our research delves into whether seeing these masterpieces in person truly lives up to the hype or if it’s better to admire them from a postcard.
Our study gained widespread attention, securing coverage in prestigious outlets such as The Guardian, Daily Mail, Smithsonian Magazine, Artforum, Artnet, NPR, The Huffington Post, Xataka, Matador Network, The National Endowment for the Arts, and The Art Newspaper.
Beyond that, The President of France announced that the Mona Lisa housed in the Louvre and a crown jewel in the museum’s collection, was to be moved following the data released in our study.
Check this out:
Campaign Summary
Our client, CouponBirds, offers consumers a wide range of the latest discount codes, deals, and vouchers across various industries, helping people save money on their online purchases.
Our goal was to secure valuable backlinks and brand mentions in authoritative publications, and so we combined this target with a concept focused on travel and experiences – an industry where CouponBirds have a variety of affiliate partners.
For this campaign, we took a deep dive into the fascinating topic of art tourism and the disappointment some visitors experience when they encounter famous masterpieces.
As we’d seen with some previously successful campaigns, like our Paris Syndrome campaign, this focus on being able to name ‘the worst’ in a category gives journalists something new – and something that engages readers – to work with.
People have long shared their disappointment over seeing iconic works like the Mona Lisa, often mentioning how the real thing didn’t live up to their expectations. This inspired our idea to explore this disappointment – but on a larger scale.
Methodology
Given our objective was to uncover the world’s most disappointing masterpiece, we considered the best methodology for doing so was to use a semantic word analysis across as many reviews as we could.
Therefore, the first part of our methodology involved scraping review sites to give us a primary data-set to work with.
The Scraping Process: From the thousands of reviews, we identified 18,176 reviews specifically referencing our target artworks, evident from their mention of the piece’s name.
Analysis Approach: Utilizing these 18,176 reviews, we evaluated their ratings, ranging from one to five stars and categorized them on a scale. Anything below three stars was considered a low-rated piece, while artworks rated four or five stars were deemed high-rated. This allowed us to understand how the Mona Lisa and other artworks were perceived by visitors.
After gathering all the reviews, we used Natural Language Processing libraries to understand how people felt about the artworks. We focused on finding out which words showed up most commonly and whether they were positive or negative. By quantifying the occurrences of these words for each artwork, we delineated the most and least favored pieces within the list.
Ranking Approach: After gathering all our data, our aim was to rank the artworks to determine the best and worst ones. To achieve this, we considered various factors: the frequency of negative and positive words in reviews, the number of low ratings (1-3 stars) versus high ratings (4 and 5 stars), and also factored in the entrance fee to each museum as additional supporting data for each piece (where required).
This formula we employed is a linear normalization equation. It serves to standardize data within a predefined range. The process involves multiple steps: first, it calculates the range of values in the dataset. Then, it computes the difference between a specific value and the dataset’s minimum value. Following this, it scales this difference to fit within a normalized range, typically between -1 and 1. Finally, it adjusts the scaled value to fall within a range between 0 and 2. This method proves valuable in making data comparable and facilitating analyses across different scales or datasets.
Results
This campaign absolutely blew up, and eventually made a bigger splash than we ever could have imagined – but it wasn’t an overnight success.
In fact, this is a perfect case study for how patience and persistence with a great concept and an outreach team who can adapt where needed can yield some remarkable results.
Exactly two months after the campaign page had been published, we’d landed 7 linking root domains. This one didn’t fly out of the gate.
It was finally on the 7th week of outreaching the campaign to journalists that we finally saw coverage land…in The Guardian:
The coverage came thick and fast from there with the campaign gaining additional attention as the impact continued to spiral.
It was the perfect snowball effect.
The Louvre’s Director was the first to publicly address the data, confirming that the experience of viewing the Mona Lisa was under review, which in turn led to move coverage. All of which was 8 months before President Macron officially announced the (disappointing) masterpiece was to be re-housed.
To date, our campaign has built 510 links according to Ahrefs. We’ve also tracked at least 133 unlinked brand mentions from publications like Sky Italia, Capital FR, 20 Minutes, Elle, Yahoo, MSN, Italy 24, and so on.
We’re incredibly proud of how well the campaign performed – not only because of the number of links that the campaign generated – a key goal for our client – but because the impact transpired Digital PR and created a conversation impeaching political discourse.