Joe Shellard
“What can theaters learn from big tech like Netflix? Is it possible to use their advanced technology in our regional theaters? And should we even be trying to?”
With those words, Joe Shellard, Head of Box Office and Ticketing for advertising agency AKA UK, welcomed INTIX Live! 2021 Digital Conference attendees to his SecuTix Inspiration Stage presentation on creating a Netflix for theatre.
“We will be discussing some of the techniques that have helped Netflix grow to almost 200 million subscribers, as well as the technical algorithms we implemented,” Shellard continued as he explained that his presentation would delve into three important topics. They included what regional theaters can learn from tech companies like Netflix, how to create personalized recommendations, and the pitfalls and failures of trying to emulate big tech in a small-town theater. The small-town theater that is the topic of this success story is the Wycombe Swan, a regional receiving venue north of London, normally offering over 300 different events per year across various genres and two performance spaces.
Before joining AKA UK, Shellard spent a year as Marketing and Sales Manager at the Wycombe Swan. When he joined the organization, he was presented with the season brochure. “It would come out three times a year, has 48 pages and lists over 100 events,” he said. “Most shows get a small, 300-word entry with a few like ‘Cilla the Musical’ that get greater coverage. Usually, this is paid for by the producers and limited to a small number of big-name events. The truth is that with this format, we cannot expect audiences to read every page, and certainly not read about every show. So our journey starts with this story.”
A previous marketing manager had gone over and above for the Wycombe Swan’s closest donors and members, using a marker to highlight the shows that were most relevant to each customer. When Shellard heard this, he became excited — it was a personal touch that made donors feel particularly valued. “It also put a spotlight on niche shows and worked to increase demand for smaller shows, which otherwise would have been overlooked. The problem, of course,” he said, “is it is simply not scalable.”
Indeed, it could easily take all year to hand-pick recommendations for each person who receives the brochure.
The team started brainstorming. Could they produce multiple smaller brochures — for example, one for plays, music, dance, opera, comedy, musicals and beyond?
“When we started to segment the data, we realized that there were a lot of people who would need to be sent two or three brochures … and that was when it dawned on us: individual brochures,” Shellard said. “What if we print tens of thousands of brochures, each unique and specific to each individual customer? This should not be anything new. Amazon has been doing this for years.”
He continued, “When we started to learn about how the recommendation algorithm worked, we found out that often, for example, if you buy a shovel, you end up getting a lot of recommendations for other shovels. I am sure this is something we have all experienced, getting endless ads for a product that you have already purchased. In our world, if someone watches ‘Romeo and Juliet,’ there is no use to keep recommending that they see ‘Romeo and Juliet’ each week. That was what led us to look at Netflix, which has built its whole business around bespoke recommendations from a vast catalogue.”
The difference, Shellard explained, is that Netflix hosts approximately 3,000 movies. Even a popular theater like the Wycombe Swan has only 300 different shows each year. And while a viewer generally watches 60 movies annually, the average theatergoer only sees 2.5 shows in the same time period at a typical venue.
“We knew we were dealing with a much smaller pool of data. The challenge ahead of us was to increase awareness of niche productions, which could not easily afford expensive marketing campaigns,” he said. “We wanted to make specific recommendations at greater scale and go beyond just telling people to see more of the same. We did not want to just suggest buying more shovels.”
Recommendations Based on Similar Tastes
Imagine this: two friends bump into each other at a show. One asks the other what they have seen lately, and it turns out they have both been to performances by the same three stand-up comedians. In addition, one of the two friends has seen a fourth show. There is a good chance, Shellard explained, that the other friend would also like that fourth show because of the high level of crossover in their tastes.
The idea then is to scale this way of thinking across an entire schedule of shows.
“We started off by downloading lists of customers from our ticketing system who had seen each of our shows in the last couple of years, and also everything that was currently on sale. To give an example, let’s say we have 900 customers who booked for ‘Cilla the Musical’ and 1,000 customers who booked for ‘Evita.’ It is very easy to compare these lists and see that there are 100 customers who went to both shows, so that means that there is an audience crossover of 100 people. What we need to do as marketeers is to decide if that is a big crossover or not,” explained Shellard, adding that a mathematical calculation called “intersection over union” is then applied.
From there, the audience of ‘Evita’ is compared against every other show. Then, every show is compared against the audience of ‘Cilla the Musical,’ and so on. The result is tens of thousands of combinations to see the crossover between each combination of shows. Admittedly, that is the hardest part of the process. Luckily, you only need to do it once.
“It means,” Shellard explained, “that you can look up any event and examine which other events share an audience with it. For example, if you attended ‘Cilla the Musical,’ you would really like to see ‘Evita,’ followed by ‘Gimme! Gimme! Gimme!’, followed by ‘Chicago.’ The data gets much more accurate once you start to build shows on top of one another, so if people have seen a combination of shows, we can be much more specific in our recommendations.”
Ultimately, combining past attendance reveals much more about a customer’s preferences than just one show, or just one genre. For example, a consumer might like to see one of their favorite comics performing in a play or they might like to attend a dance show with a friend, and puppet shows with their kids.
Future Show Recommendations
“What is interesting is that [the algorithm] starts to make recommendations, which are much more thoughtful than just going on the same genre. For example, if you like stand-up comedy and plays, it might recommend one of your favorite stand-ups performing in a play,” Shellard said.
He continued, “Combining past attendance told us much more about an audience’s preferences than just one show or even just one genre. After doing some tests, we looked to improve the algorithm and realized that the same three or four big musicals were appearing in a lot of people’s recommendations. We realized that we would be better off to feature those season highlights separately and let the recommendation section be more specific.”
Shellard further explained that feedback revealed it was important to suppress certain shows. “People might not want a reminder on their doorstep of that adult entertainment show that they saw last year. So, we discreetly removed some of those more sensitive titles from the system.”
Bringing Data and Marketing Together
As many INTIX Live! 2021 Digital Conference attendees know, customer bookings are generally fairly easy to extract from ticketing platforms. “But we also needed to get up-to-date marketing copy about each of the shows we would recommend,” Shellard said. “I am sure anyone who has worked in the theater marketing department knows that it is often very hard to get your hands on consistently sized artwork, but we realized we had it already for our website. We used a free, off-the-shelf tool, which simply downloaded the contents of our website with all of the images and up-to-date copy in one place.”
Initially, the Wycombe Swan considered putting their recommendations into a letter, which would be enclosed in the main 48-page season brochure. Once they realized most of the hard work had already been done, they instead began the process of bringing individual marketing materials to life. The end result was a customized, four-page brochure with each person’s name on the front cover, season highlights on the left inside page, three customer-specific recommendations on the right inside page, and the customer’s address with generic show recommendations on the back cover. The design was done in-house using Photoshop.
When you see several brochures beside each other, you can really appreciate how different each individual piece is and how this strategy truly recommends different shows to every audience member based on their own attendance.
Results and Feedback
In reviewing outcomes, Shellard shared key insights gained from both customer feedback and lessons learned. “Everyone was very impressed by the results when they came in. We saw a 4.2 times return on investment, and 50% more people bought a show that was specifically for them compared to those that were in our general highlights, so our core hypothesis of being able to sell more tickets by giving more accurate recommendations was proved to be true.”
Other key findings included:
- Sold more tickets overall than with the usual brochure.
- Sold many times more tickets for niche shows.
- Many of the generic recommendations on the back cover got no bookings, so that space can likely be better used.
- Producers understood that a more premium placement to a target audience was more valuable.
- It did not cost more to produce or distribute the individualized brochures; the investment was the same as the conventional version.
- Put the brochure in an envelope with clear branding.
- Work with a designer instead of trying to do it without one.
- Customers want relevant suggestions but they do not want to be reminded about the depth of data that is being stored.
- Be bold in explaining that recommendations are based on past attendance.
- Strategy works best for a large audience with high frequency and a broad program; be sure to make adjustments if you have fewer shows to offer.
- Can find an audience for a cult hit or small show with niche target audience.
- Use digital platforms to be more targeted, offer recommendations with higher frequency and be more environmentally friendly.
“Since the pandemic, the pace of innovation and digital advertising has moved forward at a greater pace. Our approach nowadays at AKA is digital first,” Shellard said. “If we were starting this project today, we would certainly look at digital platforms first and send out these recommendations in the email newsletter and on digital advertising. This would be more cost-effective and allow us to have multiple touch points with our customers. Imagine, for example, having each customer receive their own version of a newsletter by email every week with no need to write new content, but have it populated itself from your website and use the latest sales data to make recommendations. Cookies are being phased out for advertising over the next couple of years, and so the role of CRM data is going to be increasingly important for relevance, and using data like this gives organizations much more ownership of how they are running programmatic advertising.”
Beyond the Initial Trial
After the initial trial, a designer was hired to elevate the work. Highlights were included on the left side of the centerfold and recommendations in the center. The designer also came up with the concept of an individual ticket discount as a call to action.
When conventional season brochures were distributed, fewer were sent. At the same time, open rates were increased by personalizing envelopes with the top two shows for each recipient. The opportunity cost was negligible, said Shellard, because the Wycombe Swan was already printing envelopes with individual addresses.
What’s Next for Recommendations?
“Now we have product selections made for each customer, so I want to think about how we best promote that product to each audience member,” Shellard said. He referenced a quote from a group of engineers at Netflix who said their job was initially to get the right movies in front of the right user; however, now it is to get the best artwork in front of each user and highlight aspects of a title specific for each customer. The series ‘Stranger Things,’ for example, has at least nine different live versions of the artwork and the version shown to different audiences depends on their tastes.
“When Netflix promotes ‘Pulp Fiction,’ they will show a different image to a user who watches Uma Thurman films than what they would show to someone who watches John Travolta. It is the same film, but the advertising differs depending on your tastes. Netflix does this on a massive scale by continuously testing the appeal of specific artwork to different audiences,” Shellard said.
He continued, “As we look to expand these ideas, I think that it is a no-brainer to put recommendations in the email newsletter, and certainly the days of sending the same email to everybody are long past. In fact, we should not be segmenting our audience and sending a ‘plays’ newsletter to [people who book plays] and sending a separate one to [people who book musicals] because the data shows that audience habits are not defined by genres. Instead, we should be giving specific recommendations to each individual person based on their actual bookings. We can send automated emails based on triggers, so if someone has had a great night at our theater, we should be emailing them on the way home with the next three shows that they could be seeing.”
Shellard concluded that online digital adverts are a clear opportunity. Rather than concentrating marketing dollars on blockbuster season highlights, organizations can monetize the opportunity and sell premium spots in the mailer or provide more support for particular shows that need it.
“Two years ago, our small marketing team proved that all of this is possible for regional theater, and the results demonstrate that it delivers at the box office. Now,” he said, “we need to democratize this technology so that any arts organization can do it without specialist expertise.”
You May Also Like
Want news like this delivered to your inbox weekly? Subscribe to the Access Weekly newsletter, your ticket to industry excellence.