Marquee

Movie trends over time

When we hear news reports or commentary about the box office, the conversation usually turns to the blockbusters. Which film was the biggest? How large was its opening weekend? Did it beat the film that opened the week before? What is it expected to earn over its entire run? Where will it sit on the all-time list?

We also hear about the “flops”: films that arrive with A-list casts, enormous production and marketing budgets, then record disappointing ticket sales and quietly fade from circulation.

These stories are exciting. They drive conversation and may even sell more tickets, but they usually describe a snapshot at one point in time. We wanted to provide a wider view of cinema attendance: showing movies arriving and departing, the sequence in which they appeared, which ones competed for attention and which ones dominated for a long stretch.

Notable examples

Before we started designing this visualisation we wanted to first look at some existing techniques that successfully displayed data that changes over time (known commonly as time series data).

The best-known is probably Gapminder’s animated health-and-wealth bubble chart. The Trendalyzer software behind it began development in 1999, but the visualisation reached a global audience when Hans Rosling demonstrated it in The best stats you’ve ever seen. The talk was recorded in February 2006 and released online on 27 June 2006—twenty years ago.

Figure 1Gapminder’s World Health Chart

Gapminder is not a leaderboard. It shows several measures at once, allowing countries to follow different paths through the same space. Pressing play turns a sequence of historical snapshots into trajectories: long-term progress, reversals and periods of sudden change become visible in the movement.

A bar-chart race takes a different approach. Mike Bostock’s D3 example on Observable animates the value of leading global brands between 2000 and 2019. Bar length provides a familiar measure of value, vertical position records rank and interpolation makes every overtake visible.

Figure 2Mike Bostock’s Bar Chart Race

The race is more precise than a bubble chart and exceptionally clear about changing order. Its focus is also narrower. It usually follows one measure and a limited number of leaders; the history of an item becomes harder to see once it leaves the frame. Gapminder and the bar-chart race both animate time, but they answer different questions. One foregrounds multivariate trajectories. The other foregrounds value, rank and overtaking.

Designing Marquee

Marquee needed something from both examples. Like a bar-chart race, it needed to show films changing rank and overtaking one another. Like Gapminder, it needed each object to carry several values at once and to make its path through time meaningful. It also needed to show a lifecycle that neither example was designed for: films arriving, accumulating an audience and eventually leaving cinemas.

Before choosing a form, we considered what the visualisation needed to communicate:

• daily domestic gross;
• cumulative domestic gross;
• the beginning and end of a theatrical run;
• daily rank among the films currently screening;
• the relative size of films still in circulation; and
• movie genre and identity.

A table would provide precision, but every date would remain a separate snapshot. A bar-chart race would show rank changing clearly, but cumulative success and the release lifecycle would compete for the same visual channel. A bubble chart offered less precision, but considerably more room to encode those ideas together.

We eventually chose a force-directed bubble chart. It treats every bubble as a physical object, with forces drawing films towards the centre and collisions preventing them from occupying the same space. This allowed us to encode several values at once while giving the market an organic quality.

After a degree of experimentation, we settled on the following visual language:

• Bubble size represents a movie’s cumulative domestic gross to the selected date, allowing its success to build visibly over the course of its run.
• Daily gross determines rank. Highly ranked films receive a much stronger attraction towards the centre, allowing a new release to jostle its way past older films whose takings have slowed.
• Every date is scaled to the largest film still screening, so periods with smaller releases do not disappear into an empty canvas.
• Films leaving theatrical release lose their attraction to the centre and drift away from the active market.
• Colour identifies genre, while the poster identifies the film.

Fine-tuning the visualisation

Choosing the form was only the beginning. Bubbles are engaging, but people judge length much more accurately than area. For example, a circle with twice as much area does not obviously look twice as large. After several rounds of experimentation we settled on a power scale which essentially exaggerates the gaps to more clearly show the differences to the reader. This treatment creates more separation and leaves room for a successful film to keep growing after opening.

We also chose to normalise the size of the bubbles to the largest film showing at any point in time so that we could keep the overall visual legible and ensure that the comparison at any point remained clear. Normalising every date allows a major release in 1998 appear just as commanding as one from 2019, despite decades of ticket-price inflation and market growth.

We started simply with film labels for each film, but there were issues with text legibility, text overflowing the bubbles and wrapping. The idea to switch to using movie posters was a simple one because it creates instant recognition, particularly for the bigger releases and the mosaic of posters also creates immediate visual interest.

Figure 4Gross determines size; rank determines position

Poster viewAccumulated gross

Rank viewDaily position

What almost thirty years reveal

Start with the centre, then watch the edges. The centre shows the films winning the selected day. Bubble growth reveals cumulative success. Simultaneous arrivals reconstruct crowded release weekends, while the edge records the long tail: specialist films that persist, former hits whose runs are ending and titles that never gather enough momentum to enter the cultural memory.

Over time, the release calendar develops a rhythm. Summer and Christmas bring waves of large arrivals. The annual Harry Potter releases, three successive Lord of the Rings Christmases, the return of Star Wars and the accelerating cadence of the Marvel Cinematic Universe become repeated attempts to take possession of the same centre. Seen at speed, the franchise era is no longer an abstract industry trend. It becomes a visible pattern.

Shared dates create their own stories. Titanic and Tomorrow Never Dies opened together in December 1997. The Dark Knight and Mamma Mia! did the same in July 2008. Barbie and Oppenheimer turned that coexistence into the premise of an entire weekend in 2023. Marquee restores the neighbours that an all-time chart removes.

Then the rhythm breaks. In 2020, as cinemas closed during the COVID-19 pandemic, the field contracts dramatically. The slow return of releases becomes visible over the months that follow. Marquee does not explain why the market changed, but it makes the shape and timing of that change much easier to see.

A few places to begin

There is no single correct route through almost thirty years of cinema. The following moments are useful starting points rather than a definitive ranking. Grosses are cumulative domestic takings in nominal US dollars. Run length refers to continuous presence in the daily source, with short screening gaps bridged.

The films we almost forgot

The flops view is deliberately an estimate of domestic shortfall, not a profitability verdict. It compares cumulative domestic gross with reported production budget for films where a budget estimate is available. International revenue, marketing, distribution terms and ancillary income all matter, and production budgets are reported estimates rather than audited accounts.

Even so, placing a small bubble beside a large reported budget can recover a surprisingly vivid memory. Mortal Engines, Mars Needs Moms, Dark Phoenix, Tomorrowland and dozens of forgotten would-be franchises return not as a verdict but as a question: what did the industry expect this film to become and how quickly did the audience answer?