UK Merlin park attendance over the years

Matt N · Sep 19, 2022

Note: I intend to keep editing this post as time goes on and more years are released.
Hi guys. During the coaster consultations in 2021, Merlin released an attendance graph showing all their parks' attendance figures since the early 1980s. I had some time on my hands this evening and decided to try and extrapolate some slightly more precise guest figures for each park from this graph to try and determine the precise(ish) attendance trajectory of each UK Merlin park from the earliest year listed here (1984 for Alton Towers and Chessington, 1983 for Thorpe Park, 1987 for Windsor Safari Park and 1997 for Legoland Windsor).

For some idea, this is the original graph I was working with: https://www.cwoa-consultation.com/proposals?lightbox=dataItem-komw1163

To make things easier for myself, I divided each million on the graph into 8 rows (thus leaving ~125,000 guests per row, as my rather crude MS Paint annotation indicates):

As such, I then decided to extrapolate a precise(ish) figure from the graph by looking at what row each park's figure fell within. All of these figures are rounded to the nearest 31,250; I know that sounds oddly specific, but it's 1/32 of a million, and a quarter of one of these rows, so it's the most precise figure that remains easy to determine by eye. It also keeps the margin of error to only 1 or 2 percent in most cases.

The precise(ish) trajectories that I managed to extrapolate for each park, including percentage increases and decreases for each year, were as follows:
Alton Towers - opened 1980, first year on graph 1984

1984: 1,843,750 (first year, #1/3 on graph)
1985: 1,812,500 (-1.7%, #1/3 on graph)
1986: 2,250,000 (+24.1%, #1/3 on graph)
1987: 2,312,500 (+2.8%, #1/4 on graph)
1988: 2,875,000 (+24.3%, #1/4 on graph)
1989: 2,437,500 (-15.2%, #1/4 on graph)
1990: 1,937,500 (-20.5%, #1/4 on graph)
1991: 1,843,750 (-3.6%, #1/4 on graph)
1992: 2,625,000 (+42.4%, #1/4 on graph)
1993: 2,843,750 (+8.3%, #1/4 on graph)
1994: 3,312,500 (+16.5%, #1/3 on graph)
1995: 2,843,750 (-14.2%, #1/3 on graph)
1996: 2,875,000 (+1.1%, #1/3 on graph)
1997: 2,875,000 (0.0%, #1/4 on graph)
1998: 2,875,000 (0.0%, #1/4 on graph)
1999: 2,593,750 (-9.8%, #1/4 on graph)
2000: 2,437,500 (-6.0%, #1/4 on graph)
2001: 2,187,500 (-10.3%, #1/4 on graph)
2002: 2,656,250 (+21.4%, #1/4 on graph)
2003: 2,562,500 (-4.7%, #1/4 on graph)
2004: 2,086,750 (-18.6%, #1/4 on graph)
2005: 2,187,500 (+4.8%, #1/4 on graph)
2006: 2,218,750 (+1.4%, #1/4 on graph)
2007: 2,250,000 (+1.4%, #1/4 on graph)
2008: 2,593,750 (+15.3%, #1/4 on graph)
2009: 2,656,250 (+2.4%, #1/4 on graph)
2010: 3,062,500 (+14.0%, #1/4 on graph)
2011: 2,687,500 (-12.2%, #1/4 on graph)
2012: 2,406,250 (-10.5%, #1/4 on graph)
2013: 2,593,750 (+7.8%, #1/4 on graph)
2014: 2,312,500 (-10.8%, joint #2/4 on graph)
2015: 1,968,750 (-14.9%, #2/4 on graph)
2016: 1,750,000 (-9.7%, #2/4 on graph)
2017: 1,875,000 (+7.1%, #2/4 on graph)
2018: 2,187,500 (+16.7%, #1/4 on graph)
2019: 2,500,000 (+14.3%, #1/4 on graph)
2020: 968,750 (-61.3%, #1/4 on graph)
2021: 2,343,750 (+141.9%, #1/4 on graph)

All-Time Peak: 3,312,500 (1994)
All-Time Low (excluding 2020): 1,750,000 (2016)
Peak Within Merlin Era (2008 and later): 3,062,500 (2010)
Low Within Merlin Era (2008 and later, excluding 2020 and 2021): 1,750,000 (2016)

Chessington World of Adventures/Chessington Zoo - opened 1931, first year on graph 1984, first year as CWOA 1987

1984: 625,000 (first year, #3/3 on graph)
1985: 562,500 (-10.0%, #3/3 on graph)
1986: 500,000 (-11.1%, #3/3 on graph)
1987: 875,000 (+75.0%, #3/4 on graph)
1988: 1,187,500 (+35.7%, joint #3/4 on graph)
1989: 1,250,000 (+5.3%, #3/4 on graph)
1990: 1,062,500 (-15.0%, joint #3/4 on graph)
1991: 1,437,500 (+35.3%, #2/4 on graph)
1992: 1,218,750 (-15.2%, #2/4 on graph)
1993: 1,531,250 (+25.6%, #2/4 on graph)
1994: 1,687,500 (+10.2%, #2/3 on graph)
1995: 1,875,000 (+11.1%, #2/3 on graph)
1996: 1,812,500 (-3.3%, #2/3 on graph)
1997: 1,843,750 (+1.7%, #2/4 on graph)
1998: 1,843,750 (0.0%, #2/4 on graph)
1999: 1,656,250 (-10.2%, #2/4 on graph)
2000: 1,562,500 (-5.7%, #3/4 on graph)
2001: 1,531,250 (-2.0%, joint #3/4 on graph)
2002: 1,281,250 (-16.3%, #4/4 on graph)
2003: 1,312,500 (+2.4%, #4/4 on graph)
2004: 1,218,750 (-7.1%, #4/4 on graph)
2005: 1,093,750 (-10.3%, #4/4 on graph)
2006: 1,031,250 (-5.7%, #4/4 on graph)
2007: 968,750 (-6.1%, #4/4 on graph)
2008: 1,250,000 (+29.0%, #4/4 on graph)
2009: 1,343,750 (+7.5%, #4/4 on graph)
2010: 1,437,500 (+7.0%, #4/4 on graph)
2011: 1,500,000 (+4.3%, #4/4 on graph)
2012: 1,406,250 (-6.2%, #4/4 on graph)
2013: 1,531,250 (+8.9%, #4/4 on graph)
2014: 1,562,500 (+2.0%, #4/4 on graph)
2015: 1,437,500 (-8.0%, #4/4 on graph)
2016: 1,437,500 (0.0%, #4/4 on graph)
2017: 1,500,000 (+4.3%, #4/4 on graph)
2018: 1,593,750 (+6.3%, joint #4/4 on graph)
2019: 1,656,250 (+3.9%, #3/4 on graph)
2020: 500,000 (-69.8%, #4/4 on graph)
2021: 1,281,250 (+156.3%, #3/4 on graph)
2022: 1,468,750 (+14.6%, N/A)

All-Time Peak: 1,875,000 (1995)
All-Time Low (excluding 2020 and 2021, including pre-CWOA years): 500,000 (1986)
All-Time Low (excluding 2020 and 2021 and pre-CWOA years): 875,000 (1987)
Peak Within Merlin Era (2008 and later): 1,687,500 (2019)
Low Within Merlin Era (2008 and later, excluding 2020 and 2021): 1,281,250 (2008)

Legoland Windsor/Windsor Safari Park - opened 1970, first year on graph 1987, first year as Legoland Windsor 1996

1987: 812,500 (first year, #4/4 on graph)
1988: 875,000 (+7.7%, #4/4 on graph)
1989: 968,750 (+10.7%, #4/4 on graph)
1990: 1,062,500 (+9.7%, joint #3/4 on graph)
1991: 1,031,250 (-2.9%, #3/4 on graph)
1992: 968,750 (-6.1%, #4/4 on graph)
1993: 937,500 (-3.2%, #4/4 on graph)
1994: N/A (-100.0%, N/A on graph)
1995: N/A (0.0%, N/A on graph)
1996: N/A (0.0%, N/A on graph)
1997: 1,468,750 (first year as LLW, #3/4 on graph)
1998: 1,312,500 (-10.6%, #3/4 on graph)
1999: 1,500,000 (+14.3%, #3/4 on graph)
2000: 1,687,500 (+12.5%, #2/4 on graph)
2001: 1,531,250 (-9.3%, joint #3/4 on graph)
2002: 1,593,750 (+4.1%, #2/4 on graph)
2003: 1,437,500 (-9.8%, #3/4 on graph)
2004: 1,437,500 (0.0%, #3/4 on graph)
2005: 1,500,000 (+4.3%, #3/4 on graph)
2006: 1,625,000 (+8.3%, #3/4 on graph)
2007: 1,500,000 (-7.8%, #3/4 on graph)
2008: 1,875,000 (+25.0%, #2/4 on graph)
2009: 1,906,250 (+1.7%, #3/4 on graph)
2010: 1,906,250 (0.0%, #3/4 on graph)
2011: 1,906,250 (0.0%, #3/4 on graph)
2012: 2,031,250 (+6.6%, #2/4 on graph)
2013: 2,312,500 (+13.8%, #2/4 on graph)
2014: 2,312,500 (0.0%, joint #2/4 on graph)
2015: 2,343,750 (+1.4%, #1/4 on graph)
2016: 2,187,500 (-6.6%, #1/4 on graph)
2017: 2,312,500 (+5.7%, #1/4 on graph)
2018: 2,125,000 (-8.1%, #2/4 on graph)
2019: 2,062,500 (-2.9%, #2/4 on graph)
2020: 687,500 (-66.7%, #2/4 on graph)
2021: 1,562,500 (+122.7%, #2/4 on graph)

All-Time Peak: 2,343,750 (2015)
All-Time Low (excluding 2020 and 2021, including pre-LLW years): 812,500 (1987)
All-Time Low (excluding 2020, 2021 and pre-LLW years): 1,312,500 (1998)
Peak Within Merlin Era (2006 and later): 2,343,750 (2015)
Low Within Merlin Era (2006 and later, excluding 2020 and 2021): 1,500,000 (2007)

Thorpe Park - opened 1979, first year on graph 1983

1983: 843,750 (first year, #1/1 on graph)
1984: 1,031,250 (+22.2%, #2/3 on graph)
1985: 1,093,750 (+6.1%, #2/3 on graph)
1986: 1,093,750 (0.0%, #2/3 on graph)
1987: 1,093,750 (0.0%, #2/4 on graph)
1988: 1,187,500 (+8.6%, joint #3/4 on graph)
1989: 1,343,750 (+13.2%, #2/4 on graph)
1990: 1,000,000 (-25.6%, #4/4 on graph)
1991: 968,750 (-3.1%, #4/4 on graph)
1992: 1,093,750 (+12.9%, #3/4 on graph)
1993: 1,281,250 (+17.1%, #3/4 on graph)
1994: 1,218,750 (-4.9%, #3/3 on graph)
1995: 1,125,000 (-7.7%, #3/3 on graph)
1996: 1,187,500 (+5.6%, #3/3 on graph)
1997: 968,750 (-18.4%, #4/4 on graph)
1998: 875,000 (-9.7%, #4/4 on graph)
1999: 906,250 (+3.6%, #4/4 on graph)
2000: 937,500 (+3.4%, #4/4 on graph)
2001: 1,187,500 (+26.7%, #4/4 on graph)
2002: 1,437,500 (+21.1%, #3/4 on graph)
2003: 1,531,250 (+6.5%, #2/4 on graph)
2004: 1,468,750 (-4.1%, #2/4 on graph)
2005: 1,562,500 (+6.4%, #2/4 on graph)
2006: 1,812,500 (+16.0%, #2/4 on graph)
2007: 1,843,750 (+1.7%, #2/4 on graph)
2008: 1,843,750 (0.0%, #3/4 on graph)
2009: 2,125,000 (+15.3%, #2/4 on graph)
2010: 2,187,500 (+2.9%, #2/4 on graph)
2011: 2,125,000 (-2.9%, #2/4 on graph)
2012: 1,843,750 (-13.2%, #3/4 on graph)
2013: 1,786,250 (-3.1%, #3/4 on graph)
2014: 1,843,750 (+3.2%, #3/4 on graph)
2015: 1,531,250 (-17.0%, #3/4 on graph)
2016: 1,625,000 (+6.1%, #3/4 on graph)
2017: 1,562,500 (-3.9%, #3/4 on graph)
2018: 1,593,750 (+2.0%, joint #4/4 on graph)
2019: 1,500,000 (-5.9%, #4/4 on graph)
2020: 562,500 (-62.5%, #3/4 on graph)
2021: 1,218,750 (+116.7%, #4/4 on graph)
2023: 1,500,000 (Unknown, N/A)

All-Time Peak: 2,187,500 (2010)
All-Time Low (excluding 2020 and 2021): 843,750 (1983)
Peak Within Merlin Era (2008 and later): 2,187,500 (2010)
Low Within Merlin Era (2008 and later, excluding 2020 and 2021): 1,500,000 (2019)

To sum up each park's trajectory:

Alton Towers may have been top dog for the bulk of the years since 1984, but it has also had the most volatile guest figures. It has had peaks as high as 3.3 million in 1994, but also troughs of only slightly above 2 million in the mid-2000s or even slightly below in the early 1990s and mid-2010s, with a nadir of 1.75 million being reached in 2016. At that point, it was well away from #1 and almost rubbing shoulder to shoulder with #3 park Thorpe. Interestingly, its peak was early, in 1994, and only 2010 has ever come close to that since. Merlin have attained fair growth at Alton Towers; between 2007 and 2019, attendance grew by 11.1%.
Chessington World of Adventures started off fairly well, attaining steady growth from 1987 up until 1994, where it stayed at its peak until about 1997. However, attendance dropped through the floor from 1998 onwards, hitting a low of under 1 million in 2007, so it's fair to say that Chessington's trajectory has been far from uniform, although things improved notably under Merlin. Interestingly, Chessington is the park that has thrived most under Merlin, with attendance having grown by 74.1% between 2007 and 2019. Nonetheless, the high water mark was hit quite early on at Chessington, with that near 2 million peak guest figure being all the way back in 1995, and no year post-1997 has yet come close to it.
Legoland Windsor has had the most consistent growth trajectory of all the parks. With its low back near opening in 1998, its peak in 2015 and no particularly catastrophic attendance drops (COVID aside), it's grown fairly consistently over the years. It's also a park that has thrived pretty well under Merlin; between 2005 and 2019, attendance grew by 37.5%.
Thorpe Park has had a bit of a roller coaster of a growth trajectory. The 80s and 90s were a little bit choppy at Thorpe Park, with peaks of close to 1.5 million and lows of under 1 million. The park really hit its stride from 2001 onwards, maintaining a near perfect growth trajectory right up to the park's 2.2 million peak in 2010. However, things have been a bit of a struggle since then, with guest figures having almost consistently declined since 2011 right back to a low of 1.5 million in 2019. The park has comparatively struggled under Merlin, with attendance having fallen by 18.6% between 2007 and 2019.

I hope you find this interesting! If you don't agree with something I've done or notice any errors, however, don't be afraid to flag them to me.

Heth · Sep 19, 2022

This is fascinating, well done for digesting all of this. I do hope you're going into a career in statistics or numbers in some form!

Matt N · Sep 19, 2022

Heth said:
This is fascinating, well done for digesting all of this. I do hope you're going into a career in statistics or numbers in some form!

Thank you very much; I’m glad you like it!

Believe me, the possibility of a career in data analysis has not passed me by! I’m currently studying a Computer Science degree, and I’ve picked the optional module on Data Analytics, so that is certainly a path I’ve considered!

MackMan · Sep 20, 2022

This is awesome!

I do believe that some of this data explains some of the investments- like how Chessington has been steadily growing primarily over the last 10 years or so, and also to why Merlin seem interested in it (hence the new B&M Wing). It certainly is delivering for being such a small lowkey park

Thorpe’s constant climb downwards also makes sense with the non existent investments and Merlin kind of sidelining it.

Alton and Legoland are both very profitable due to their resorts/merchandise so the gate numbers probably are only half the story, both generally have the highest numbers in the chain anyway, hence why both tend to get favourable investments.

Sent from my iPhone using Tapatalk

Nicky Borrill · Sep 20, 2022

Excellent work Matt, but the 2020 data is pointless as it’s incomplete, being only a partial year up to summer, it seems. So I’m pleased you didn’t use it in any analysis.

Would be interesting to see the data for ten years or so before Merlin. As I’d love to know if Thorpe was on a trajectory to match their investment.

Matt N · Sep 20, 2022

Nicky Borrill said:
Excellent work Matt, but the 2020 data is pointless as it’s incomplete, being only a partial year up to summer, it seems. So I’m pleased you didn’t use it in any analysis.

Would be interesting to see the data for ten years or so before Merlin. As I’d love to know if Thorpe was on a trajectory to match their investment.

I purposely didn’t use the 2020 data in any of the analysis because let’s face it, 2020 had a pretty artificial attendance cap caused by lockdowns not letting the parks open until July and numbers having to be capped compared to previous years (although I’m admittedly not sure how much effect the COVID cap actually had seeing as it was based upon full capacity, which the parks very rarely hit anyway). On the basis of it only being a partial season alone, I didn’t think it was fair to include 2020 in my analysis, as you would have done very well to have less than a 60% attendance decrease in 2020! I’m assuming Merlin did log the whole of 2020’s figures in this graph, seeing as its first outing was in May 2021 and also seeing as the decreases I logged roughly lined up with those reported by other parks, but I’m not sure why they didn’t just have the line running all the way to the end of 2020 as opposed to stopping it in the middle…

I can definitely tell you if Thorpe was on a trajectory to match their investment in under Tussauds, as the graph figures for Thorpe, as well as the figures I extrapolated, actually go all the way back to 1983!

Between 1998 (the year Tussauds bought the park) and 2007 (the year it was sold to Merlin), Tussauds actually gained incredible growth from Thorpe Park, with figures growing by 110.7%! The park’s 1998 figure was a mere 875,000, while the 2007 figure was 1.84 million, so attendance more than doubled due to Tussauds’ huge investment into thrill at the park! Interestingly, this spike didn’t really start until 2001 when the 3 thrilling flats were built, but once it started, it practically didn’t stop!

The park actually rode this wave of success for a good few years into the Merlin era. The park’s attendance shot up by another 300,000, or 15%, with the addition of Saw in 2009, and then stayed at a peak of around 2.1-2.2 million right through until 2011. If I look at the total growth between Tussauds’ initial purchase in 1998 and peak attendance in 2010 (2.19 million), then it provides an even better picture, with attendance having grown by exactly 150.0% between 1998 and 2010!

Nicky Borrill · Sep 21, 2022

Matt N said:
I purposely didn’t use the 2020 data in any of the analysis because let’s face it, 2020 had a pretty artificial attendance cap caused by lockdowns not letting the parks open until July and numbers having to be capped compared to previous years (although I’m admittedly not sure how much effect the COVID cap actually had seeing as it was based upon full capacity, which the parks very rarely hit anyway). On the basis of it only being a partial season alone, I didn’t think it was fair to include 2020 in my analysis, as you would have done very well to have less than a 60% attendance decrease in 2020! I’m assuming Merlin did log the whole of 2020’s figures in this graph, seeing as its first outing was in May 2021 and also seeing as the decreases I logged roughly lined up with those reported by other parks, but I’m not sure why they didn’t just have the line running all the way to the end of 2020 as opposed to stopping it in the middle…

I can definitely tell you if Thorpe was on a trajectory to match their investment in under Tussauds, as the graph figures for Thorpe, as well as the figures I extrapolated, actually go all the way back to 1983!

Between 1998 (the year Tussauds bought the park) and 2007 (the year it was sold to Merlin), Tussauds actually gained incredible growth from Thorpe Park, with figures growing by 110.7%! The park’s 1998 figure was a mere 875,000, while the 2007 figure was 1.84 million, so attendance more than doubled due to Tussauds’ huge investment into thrill at the park! Interestingly, this spike didn’t really start until 2001 when the 3 thrilling flats were built, but once it started, it practically didn’t stop!

The park actually rode this wave of success for a good few years into the Merlin era. The park’s attendance shot up by another 300,000, or 15%, with the addition of Saw in 2009, and then stayed at a peak of around 2.1-2.2 million right through until 2011. If I look at the total growth between Tussauds’ initial purchase in 1998 and peak attendance in 2010 (2.19 million), then it provides an even better picture, with attendance having grown by exactly 150.0% between 1998 and 2010!

Fairly certain it was stated at the consultation that the 2020 figures did not cover up until the end of 2020, charts also seem to indicate as much. Not that it matters, crap year anyway, like you said.

Wowzer, 110% wonder what made them change tactic.

Matt N · Sep 22, 2022

Nicky Borrill said:
Wowzer, 110% wonder what made them change tactic.

I think guest figures decreasing in 2012 when Swarm was added gave them cold feet about that strategy. Even though I would have argued that many extenuating circumstances of the year contributed to that (extremely rainy summer, Olympics), something clearly gave them cold feet when Swarm didn’t raise guest figures.

JoshC. · Sep 23, 2022

Nicky Borrill said:
Excellent work Matt, but the 2020 data is pointless as it’s incomplete, being only a partial year up to summer, it seems. So I’m pleased you didn’t use it in any analysis.

Nicky Borrill said:
Fairly certain it was stated at the consultation that the 2020 figures did not cover up until the end of 2020, charts also seem to indicate as much. Not that it matters, crap year anyway, like you said.

As I understand it, those numbers are the full 2020 numbers. They certainly wouldn't be up until summer; the parks basically opened in July (bar maybe a couple of days in March), and certainly wouldn't have hit those numbers in a month or two.

Matt N said:
I think guest figures decreasing in 2012 when Swarm was added gave them cold feet about that strategy. Even though I would have argued that many extenuating circumstances of the year contributed to that (extremely rainy summer, Olympics), something clearly gave them cold feet when Swarm didn’t raise guest figures.

Thorpe Park 2012, with its attendance figures, the changes it bought to the philosophy of investments and everything else is an interesting story which could fill a book to be honest.

The attendance the park received 2009-2011 was simply not sustainable. The park doesn't have a wide enough appeal, a wide enough target market and a big enough revisit appeal to have kept those numbers up for so long. 2009 and 2010 worked so well because of the success of Saw as a brand. 2011 continued to ride on that, and was buoyed by an increase in domestic travel. I'd argue that 2011's successes were unexpected.

Thorpe/Merlin have expectations with new investments and, crudely speaking, these will revolve around raw numbers like attendance and profit. With these unsustainable visitor numbers, the park was bound to hit a rough patch soon, and unfortunately 2012 was that year. The Olympics and poor weather certainly contributed too, as people were less interested in theme parks generally.

So Swarm wasn't the instant success it had to be. It seems weird that a company as big as Merlin didn't take it on the chin that external factors caused issues (I'm sure they're aware of it), but they did also look to see what they could have done differently. A marketing campaign that overhyped the ride (one of the key complaints from guests was it wasn't as intense as expected) and a lack of recognisable brand meant they felt the urge in future to go bigger and intense, and also lean on IPs. These were, of course, two of the big contributing factors to Saw's success too.

Matt N · Oct 9, 2022

Disclaimer: This post is quite long; I apologise in advance!
If any of you are interested, I decided to play with the attendance data a little more this evening and answer a couple of questions!

For starters, I'd like to apologise sincerely, because I only just realised that I messed up a couple of the numbers so that they were off by about 20,000... sorry about that. I've corrected the incorrect readings now.

The first thing I decided to ask was...
How would the parks be ranked in terms of average all time attendance?
To answer this question, I applied the statistical measures of mean, median and mode to each theme park's attendance.

The first statistical measure I used was the mean. The mean is a calculated average, which I worked out by adding all of the attendance figures together and dividing them by the number of years each park had listed in the graph. When ranked by mean attendance, the parks stacked up as follows:

Alton Towers - 2,429,688 (excluding 2020), 2,389,358 (including 2020)
Legoland Windsor - 1,820,652 (excluding 2020), 1,772,135 (including 2020)
Chessington World of Adventures - 1,425,189 (excluding 2020), 1,397,978 (including 2020)
Thorpe Park - 1,396,250 (excluding 2020), 1,372,664 (including 2020)

I must admit I was surprised at this... I wasn't surprised at the placements of AT and LLW, but for some reason, I was expecting Thorpe to be above Chessington... I guess its #4 placement makes sense from looking at the graph, but I must admit that it still surprised me.

The placements were much the same when I used median (the middle value):

Alton Towers - 2,437,500 (excluding and including 2020)
Legoland Windsor - 1,875,000 (excluding 2020), 1,781,250 (including 2020)
Chessington World of Adventures - 1,437,500 (excluding and including 2020)
Thorpe Park - 1,343,750 (excluding 2020), 1,312,500 (including 2020)

The mode (most common value) for each park was pretty different from the mean and median (which suggests clusters of higher and lower values rather than a fully normal distribution), but the rankings remained roughly the same:

Alton Towers - 2,875,000 (excluding and including 2020)
Legoland Windsor - 1,500,000 (excluding and including 2020)
Chessington World of Adventures - 1,437,500 (excluding and including 2020)
Thorpe Park - 1,093,750 (excluding and including 2020)

So all in all, I think we can conclude that interestingly, the parks' average attendances of all time align pretty nicely with the 2019 attendance ranking, with the parks being ranked as follows under all of the statistical measures:

Alton Towers
Legoland Windsor
Chessington World of Adventures
Thorpe Park

The second question I asked was...
Have any of the parks had anomalous years, or outliers, in their history?
To answer this question, I decided to calculate the standard deviation, which is a statistical measure that tells you how much variation there is in your dataset.

I hear you asking "well, how is the standard deviation relevant to outliers?". Well, standard deviation is relevant because the definition of an outlier is any reading that is more than 2 standard deviations away from the mean on either side. With that in mind, I calculated the standard deviation of each park's attendance, and used that to work out the "normal" range of each park's attendance (the range of figures that falls within 2SD of the mean; if the attendance data followed a perfect bell curve/normal distribution, approximately 97.5% of the data would fall within 2SD of the mean). Any figures that fell outside this range were deemed to be outliers, or anomalous years.

Here's how each park fared in this regard...
Alton Towers
Excluding 2020
Mean All Time Attendance: 2,429,688
Standard Deviation: 392,762
Upper Bound of "Normal" Attendance Range: 3,215,212
Lower Bound of "Normal" Attendance Range: 1,644,163
Outliers: 1 (3,312,500 attained in 1994)

Including 2020
Mean All Time Attendance: 2,389,358
Standard Deviation: 458,428
Upper Bound of "Normal" Attendance Range: 3,306,214
Lower Bound of "Normal" Attendance Range: 1,472,503
Outliers: 2 (3,312,500 attained in 1994 and 937,500 attained in 2020)

Chessington World of Adventures
Excluding 2020
Mean All Time Attendance: 1,425,189
Standard Deviation: 259,927
Upper Bound of "Normal" Attendance Range: 1,945,044
Lower Bound of "Normal" Attendance Range: 905,335
Outliers: 1 (875,000 attained in 1987)

Including 2020
Mean All Time Attendance: 1,397,978
Standard Deviation: 301,149
Upper Bound of "Normal" Attendance Range: 2,000,276
Lower Bound of "Normal" Attendance Range: 795,680
Outliers: 1 (500,000 attained in 2020)

Legoland Windsor
Excluding 2020
Mean All Time Attendance: 1,820,652
Standard Deviation: 339,098
Upper Bound of "Normal" Attendance Range: 2,498,898
Lower Bound of "Normal" Attendance Range: 1,142,457
Outliers: 0

Including 2020
Mean All Time Attendance: 1,772,135
Standard Deviation: 408,021
Upper Bound of "Normal" Attendance Range: 2,588,157
Lower Bound of "Normal" Attendance Range: 956,094
Outliers: 1 (656,250 attained in 2020)

Thorpe Park
Excluding 2020
Mean All Time Attendance: 1,396,250
Standard Deviation: 384,292
Upper Bound of "Normal" Attendance Range: 2,164,833
Lower Bound of "Normal" Attendance Range: 627,667
Outliers: 1 (2,187,500 attained in 2010)

Including 2020
Mean All Time Attendance: 1,372,664
Standard Deviation: 405,989
Upper Bound of "Normal" Attendance Range: 2,184,643
Lower Bound of "Normal" Attendance Range: 560,686
Outliers: 2 (2,187,500 attained in 2010 and 500,000 attained in 2020)

It's really interesting to me to see that every single park has had at least one outlier at some point in its history! I knew that every park would have had 2020 as a giant outlier (that one needs no explanation, really...), but surprisingly, every park apart from Legoland had an anomalous year even prior to COVID. In Chessington's case, it was the park's all time low in 1987, while in the case of Alton and Thorpe, it was actually their all time peaks, in 1994 and 2010 respectively, that were outliers compared to the parks' other years attendance-wise.

You're probably wondering how outliers are actually relevant to theme park attendance. In fairness, they're not as relevant to theme park attendance as they would be in, say, a science experiment, but I think they're still important to consider, even though we are talking about theme park attendance rather than the growth of germ cultures in a petri dish or whatever.

In statistical terms, outliers are usually excluded from calculations such as averages and predictive analytics models. They are readings that are said to have a low chance of repeatability under normal circumstances, so my data suggests that the parks' anomalous years (both peaks and troughs) have produced attendance figures that there's a fair chance of the parks never hitting again, under normal circumstances at least. And when you look at the circumstances surrounding the parks' respective anomalous years, that would make some sense; the widespread anomalous low of 2020 was likely caused by COVID-19 and the associated lockdowns and social distancing (which certainly couldn't be referred to as "normal circumstances"), the anomalous Chessington low of 1987 was likely caused by it being the park's opening year and it only having half a season (which is unlikely to be repeated... the park will never have an opening year again, and a half-season is unlikely to be repeated under normal circumstances), and the anomalous Alton Towers and Thorpe Park highs were both likely caused by the opening years of freakishly successful major ride installations (I know Thorpe's was 2010 rather than 2009, but 2009 and 2011's figures were on the cusp of being outliers themselves, and 2009's massive increase was almost definitely stimulated by Saw), which likely wouldn't be considered "normal" circumstances.

I know that theme park attendance is a touch too nuanced for outliers to be set in stone, but I still think it's an interesting point to consider.

Another question I asked was...
Which parks have the most volatile attendance?
In order to answer this question, I once again employed our good friend standard deviation to measure the variability of each park's attendance.

In terms of standard deviation, the parks were ranked as follows:
Excluding 2020

Alton Towers - 392,762
Thorpe Park - 384,292
Legoland Windsor - 339,098
Chessington World of Adventures - 259,927

Including 2020

Alton Towers - 458,428
Legoland Windsor - 408,021
Thorpe Park - 405,989
Chessington World of Adventures - 301,149

So in terms of pure numbers, Alton Towers has the most variable attendance and Chessington has the least variable attendance. However, one key piece of the puzzle that pure numbers on their own don't take into account is the mean that the SD varies from in the first place; if the mean is larger, the SD will often be larger by default. To counter this, let's instead rank the parks by their standard deviation as a percentage of their mean attendance...
Excluding 2020

Thorpe Park - 27.5%
Legoland Windsor - 18.6%
Chessington World of Adventures - 18.2%
Alton Towers - 16.2%

Including 2020

Thorpe Park - 29.6%
Legoland Windsor - 23.0%
Chessington World of Adventures - 21.5%
Alton Towers - 19.2%

Interestingly, Thorpe Park actually has the most volatile attendance of the Merlin parks when SD is measured as a percentage of mean attendance; its standard deviation is surprisingly high given that its mean attendance is comfortably the smallest of the 4 parks.

This really puts into perspective how much higher the late 2000s and early 2010s were for attendance at Thorpe Park than any other point in its history, as well as how different Thorpe's growth trajectory has been to that of the other 3 parks. In spite of Thorpe having the most years on the graph, it only had one proper peak that it attained fairly late in its life. The park mostly ticked along at a fairly low attendance in its early years, peaking not much above the 1 million mark (quite often a bit below it, in fact; the park had numerous years of 900k or so, and its low is a little below 850k) for 20 years. The 1.5 million threshold was not breached until 2003, 20 years into Thorpe Park's time on the graph, and the peak of 2.1-2.2 million was only ever hit during the 2009-2011 period, with no year before or since coming even remotely close to it.

By comparison, Alton and Chessington both appear to have managed roughly 2 big growth cycles (or perhaps 1.5 in Chessington's case, as a trough never began in the 2nd one before COVID), and they've maintained a fairly high relative attendance throughout their lives on the graph (since 1984 in Alton's case and since 1987 in Chessington's case). Legoland never really had a trough, with that park having managed a fairly consistent, steady upward trajectory since appearing on the graph in 1997.

On a side note, I guess this also proves why you can't really rely on pure numbers on their own; Alton Towers had the highest actual number as its SD, but this was likely caused by its mean attendance being a solid peg higher than that of the other 3 parks, as its SD actually equated to the lowest percentage of mean attendance out of the 4.

The final question I asked was...
Which park fills to capacity the most?
Now I know that this one is based upon slightly more spurious reasoning, but I did think that this could be one question where the 2020 data might come in handy. I thought that the 2020 attendances might provide an interesting insight into this because in 2020, the parks were all open for the same amount of time, and all of them had heavily capped attendance, so hit capacity most of the time.

The way I thought I'd answer this question was by ranking the parks by how much lower their 2020 attendance was than their 2019 attendance (the percentage decrease). The park with the highest percentage decrease should, in theory, be the one that fills to capacity the most.

In terms of percentage decrease, the parks' 2020 attendances were ranked as follows:

Chessington World of Adventures - -70.4% (500,000 vs 1,687,500)
Legoland Windsor - -68.2% (656,250 vs 2,062,500)
Thorpe Park - -66.7% (500,000 vs 1,500,000)
Alton Towers - -62.5% (937,500 vs 2,500,000)

So from this, it can be ascertained that Chessington World of Adventures hits capacity the most and pushes the limits of its capacity the most on a typical day, while Alton Towers hits capacity the least by quite some margin. This would make sense given Chessington's well documented capacity struggles, and the fact that I can count on one hand the number of times I've ever known Alton Towers hit capacity.

The 2020 figures would also imply that Alton Towers has the highest capacity of any of the UK Merlin parks by quite some margin; its 2020 figure was close to 1 million, while the others were more like 500,000 (Chessington and Thorpe were joint lowest at bang on 500k, while Legoland was a little higher and edged closer to 600-700k).

One final thing I should reveal is that I have condensed all of this data into a Google Sheet, which has each park's data and statistics listed alongside a labelled line graph showing all of them together for visual perspective:

UK Merlin park attendance through the years

docs.google.com

If you're wondering why every park in the spreadsheet starts from 1983, with the ones that start later than 1983 having 0 listed as their attendance for the early years, this was because the graph didn't seem to like it when I tried to start each park from the year it actually started on Merlin's graph... it got very mixed up, so I just decided to have 1983 as a uniform starting point for all of them and put 0 for any years before the park actually appears on the graph. That's why every park apart from Thorpe has a massive surge from the bottom towards the left of the graph... that's simply that park "entering" the graph, if you like.

I hope you find this interesting! If you find any errors in my calculations, do tell me (I apologise once again for the earlier mess-up of a few of the figures...), and also tell me if you'd like me to ask any more questions about this dataset!

rob666 · Oct 9, 2022

Sorry Matt, I didn't quite get that, could you run it past me again, but with a bit more detail?
And where is my "too long didn't read" conclusion in bold eh?

I know you don't always get my humour, so I can only say...Just kidding...again!
One big point...parks don't simply generate their own attendance, outside factors like the weather, other competing events and simply "fashion" trends have as much influence as the new big ride.
Entertaining read all the same, statistical geekery of the finest order.

Matt N · Nov 16, 2022

I'm back again! A new Merlin consultation can only mean one thing... a new version of the attendance graph updated with 2021 attendance! Here is the new graph for your viewing pleasure:

I've had a play about with the new graph, and I've managed to extrapolate some figures!

I should firstly note that I have restated a few of the previous figures following closer examination of this new graph. I apologise for my earlier error; the mistakes have been rectified in my opening post.

The most notable thing I discovered are the 2021 figures for each Merlin park, as well as some seemingly altered 2020 figures and the 2022 figure for Chessington.
Final 2020 Figures

Alton Towers - 968,750 (-61.3%)
Legoland Windsor - 687,500 (-66.7%)
Thorpe Park - 562,500 (-62.5%)
Chessington World of Adventures - 500,000 (-69.8%)

2021 Figures

Alton Towers - 2,343,750 (+141.9%)
Legoland Windsor - 1,562,500 (+127.3%)
Chessington World of Adventures - 1,281,250 (+156.3%)
Thorpe Park - 1,218,750 (+116.7%)

Interestingly, Chessington also had a 2022 figure given. In 2022, the park received 1,468,750 guests, a change of +14.6% on 2021.

Now we have this information, we can work out... which park recovered the best from COVID in 2021?

There are multiple metrics by which we can establish this.

The first potential metric to use is the percentage increases in attendance that each park registered in 2021...
Percentage Increases

Chessington World of Adventures - +156.3% (1,281,250 in 2021 vs 500,000 in 2020)
Alton Towers - +141.9% (2,343,750 in 2021 vs 968,750 in 2020)
Legoland Windsor - +127.3% (1,562,500 in 2021 vs 687,500 in 2020)
Thorpe Park - +116.7% (1,218,750 in 2021 vs 562,500 in 2020)

This metric would suggest that Chessington World of Adventures had the strongest recovery while Thorpe Park had the weakest recovery. However, this is not a perfect metric. This is because the parks' attendance decreased by marginally different amounts in 2020; while all parks were under the same COVID restrictions, some parks were affected more than others. For instance, Chessington's percentage decrease in 2020 was nearly 70%, while Alton Towers' was barely above 60%. This is important because if a park is coming from a more diminished base compared to 2020 in the first place, then its percentage increase is bound to be higher.

A more reliable metric to use is the net percentage decrease from 2019 to 2021...
Net Percentage Decreases Between 2019 and 2021

Alton Towers - -6.3% (2,343,750 in 2021 vs 2,500,000 in 2019)
Thorpe Park - -18.8% (1,218,750 in 2021 vs 1,500,000 in 2019)
Chessington World of Adventures - -22.6% (1,281,250 in 2021 vs 1,656,250 in 2019)
Legoland Windsor - -24.2% (1,562,500 in 2021 vs 2,062,500 in 2019)

This metric would suggest that Alton Towers had by far the strongest recovery in 2021, while Legoland Windsor had the weakest. Alton Towers' recovery in 2021 was miles stronger than that of all the other parks; Alton's 2021 attendance was only 6% lower than its 2019 attendance, while Thorpe Park's was close to 20% lower, Chessington's was over 20% lower, and Legoland's was nearly 25% lower. Why this could be is anyone's guess... I'm certainly stumped by it, as every park was hit by the same COVID restrictions in 2021.

It should be noted that 2021 was still not an entirely COVID-free season, however, as the parks opened nearly a month late and attendance was still somewhat capped until July. With this in mind, 2022 attendance, when it is available for all parks, may be a better indicator... for what it's worth, Chessington's 2022 attendance remains approximately 11.3% lower than its 2019 attendance, so even in 2022, it has not recovered as strongly as Alton Towers managed to in 2021.

Finally, I should remind you that all of this data is viewable in a Google Sheet:

UK Merlin park attendance through the years

docs.google.com

I hope you find this interesting! I know I certainly found it interesting to compile... I'll hopefully see you again some time soon when the 2022 attendance is made available for all parks!

Nicky Borrill · Nov 17, 2022

I feckin knew they'd smashed it in 2021 at Alton Towers despite the shorter season!!! I've been saying this for ages... Finally some vindication..

And considering they lost more than half of their operating season on the start and their busiest few days on the end of 2020... a million that year is insane. (although at the Thorpe consultation it was mentioned that those figures weren't final, so to discover they were is a little disappointing. I guess they meant weren't 'up to date?')

Great work Matt.

#GoTeamAltonTowers

Edited to add: Alton Towers 2021 figures for a partial season matched Legoland's all time record... Take that you crappy little bricks of pain.

Matt N · Sep 5, 2023

Disclaimer: This is a very long post with a fair amount of statistical talk. I'll try my best to simplify some of it a bit and explain a few things, but if you don't like statistics, this post may not be for you! There is a TL;DR at the bottom condensing the overall findings into a more concise format.
Sorry to bump this particular thread, but following the really interesting topic posted by @Chris Brown earlier about 2012 vs 2023 in terms of UK theme park attendance and what effect various factors have on attendance, I was inspired to revisit this dataset and explore the relationship between UK theme park attendance and various extrinsic factors, as these have previously raised some interesting discussion points surrounding the topic of UK theme park attendance.

Before I explore various different factors and their effect upon UK theme park attendance, I should firstly set out that the attendance I use is the combined attendance of all four theme parks from 1997 (the first year where all four are listed under their current guise) through until 2021. I tried all of my tests for the dataset including both 2020 and 2021, the dataset excluding 2020 only and the dataset excluding both 2020 and 2021, as I felt that the circumstances of 2020 in particular were too anomalous not to consider and I was unsure whether to even place 2021 among "normal" years, as the parks were still restricted to some degree for part of or all of the season. As such, I tested the data both including and excluding the COVID years, so that we could see the relationships exhibited pre, during and post COVID.

To test out whether a significant causal relationship exists between two variables, I used a Pearson correlation coefficient test, and the two metrics I used to determine this were the correlation coefficient itself and the p-value. To explain what each is:

The correlation coefficient is a number between 1 and -1 that denotes how strong the causal relationship between two variables is. A correlation coefficient of 1 indicates a perfect positive correlation (i.e. "as x increases, y also increases"), a correlation coefficient of 0 indicates no correlation (i.e. "x has no significant effect on y"), and a correlation coefficient of -1 indicates a perfect negative correlation (i.e. "as x increases, y decreases"). As it is staggeringly rare to have a perfect correlation, I will denote the strength of the correlation by using the absolute value of the correlation coefficient as follows; an absolute coefficient of 0-0.25 indicates no significant correlation, an absolute coefficient of 0.25-0.5 indicates a weak correlation, an absolute coefficient of 0.5-0.75 indicates a moderate correlation, and an absolute coefficient of 0.75 or higher indicates a strong correlation.
The p-value is the probability that a relationship does not exist, and it is a decimal falling between 0 and 1. It can be represented as a percentage; for instance, a p-value of 0.55 indicates that there is a 55% chance of a relationship not existing. In hypothesis testing, you want the p-value to be low if you are wanting to prove your hypothesis (in this case, that a causal relationship exists) and disprove the null hypothesis (in this case, that no causal relationship exists). I will denote how strong the evidence for a causal relationship is by using the p-value as follows; a p-value of 0.1 or higher indicates insufficient evidence in favour of a relationship, a p-value of 0.05-0.1 indicates marginally significant evidence in favour of a relationship, a p-value of 0.01-0.05 indicates significant evidence in favour of a relationship, and a p-value of less than 0.01 indicates extremely significant evidence in favour of a relationship.

Now I've explained some of my processes, let's move onto the analysis! The first external factor I tested out is one that has been particularly relevant this summer... it's that good old chestnut known as the weather!
The Weather
The weather is often referenced as a factor that could potentially be affecting UK theme park attendance, so I thought; why not test that theory out? Now, I hear you asking "Matt, there are so many different metrics of weather; which one did you test out?". That would be a fair question, and in answer, I tested out three different weather metrics; average rainfall in millimetres, average maximum temperature in degrees Celsius, and average number of hours of bright sunshine. To gain the relevant weather data, I took the months between April and October (the 7 months in which the parks are operating for the full month) for each metric for every year since 1997 from the Met Office weather data archive (https://www.metoffice.gov.uk/research/climate/maps-and-data/uk-and-regional-series). I averaged out the values for the months from April-October of a given year and used that as that year's value for a given metric. I set the region as "England"; as all four Merlin parks are in England, I figured that the weather in Wales, Scotland and Northern Ireland was irrelevant for this particular investigation.

Average Rainfall (in millimetres)
So for our first metric, average rainfall in millimetres, the distribution of the data including 2020 and 2021 was as follows: