Don't miss
  • 1,972
  • 5,500
  • 5,703
  • 115

[Gamesbriefers] How do you define whales?

By on December 3, 2012

This week’s Gamesbriefers post combines our usual suspects, who respond to our question by email, comments from a recent post where Nicholas asked you how you think whales should be defined in a free-to-play game.


Question:

Nicholas Lovell

I’m toying with the idea of defining whales as “those players who represent 50% of your revenue”, meaning that you can quickly get a sense of how whale-dependent your business is.

Is this a good idea and/or benchmark, and if not, why not?


Answers:

Tadhg Kelly Creative Director at Jawfish Games

Is it maybe a bit too relative? I think whales has caught on because its usually tied to a tangible number (> $100) that most managers, accountants, financiers and so on can emotionally understand as well as intellectually.

 


Patrick O’Luanaigh CEO of nDreams

That definition doesn’t work outside of F2P games – if you’re trying to define ‘whales’, it would be best to have a definition that makes sense to boxed product guys as well (obviously 50% of the revenue for most boxed game comes from 50% of the players). For me, whales are “people who are willing to spend a substantial amount of money on a game that they love”, and the key is setting a value for what a ‘substantial amount’ is - that value being different for different games.


Oscar Clark Evangelist at Applifier

I agree with Tadgh that its too relative and that as an approach is not really statistically useful.

From a User-Centric view (marketing) we want to understand the motivations of the player by category or segment and to define a segment by percentage of revenue doesn’t allow us to understand what is going on. For example if you are poorly monetising a Whale community you may not be getting any players generating (for example) $100 per month or more, and the 50% of profits proposal is of little value as a data point.

Better to segment your audience and identify consistent bands of behaviour to which you can apply meaningful data – e.g. 35-40 female spending $100+ pcm. Only if we can apply this same definition across multiple games does the segmentation stand up for itself and become useful.


Teut Weidemann Online specialist at Ubisoft

Its easier to define once you get rid of the 50% of whatever:

Whales are people who spend 10x ARPU or more.

I guess that also works for Facebook (low arpu) or other f2p games with high arpu. As arpu also includes non payers this also gets rid of the conversion rate as a parameter.


Andy Payne Founder of Mastertronic

In two of our [ little ] worlds  namely flight and train simulation,  which used to be completely boxed goods about 10 years ago at which point we went digital, we define Whales as those who spend 10 x ARPU which sort of equates to a piece of add on content for their virtual world (flight or train) each and every month. We have hundreds of those. I may even run a report to get the up to date data. These are the same customers who come and visit us at Aviation and Railway shows.


Eric Benjamin Seufert Head of Marketing at Grey Area Labs (via comments)

In my opinion, a top-down approach doesn’t work for “whale spotting” — whales are primarily defined by behavior (revenue spent is the artifact left behind by that behavior), and I feel that grouping them as the “50% (or whatever arbitrary number) of revenue” group doesn’t actually add anything to the organization’s understanding of its players but another graph on the dashboard. If you know, bottom up, who whales are because you’ve identified the behaviors that most whales tend to exhibit, then you can actually use that information to inform the evolution of your game.

As a very high-level benchmark, 50% of revenue is pretty good. One way to proxy bottom-up while still using broad revenue numbers could be to define a whale as someone who has spent at least 1% of the total product catalogue (i.e. if the total product catalogue is €5k then a user who has spent 50€ is a whale)

About Gamesbriefers

Every week, we all ask our august panel of luminaries a burning question in the world of free-to-play and paymium game design. Or we ask a broader question on the future of the industry. We’re not going to announce who is a GAMESbriefer. You’ll just have to read the posts to see who is saying what to whom. We have CEOs and consultants, men and women, Brits, Germans, Americans, indies, company people and much more besides.
  • Jim

    I’m not sure the 10 x ARPU benchmark works for casual f2p games. Some of those titles have such low conversion rates that effectively anyone that monetized would fall into the “whale” bucket.

    Looking at the ARPPU can provide more insight. If you consider anyone who spends twice the monthly ARPPU to be a whale then you will get a better picture of the distribution.

  • Aaron Bannin

    - I don’t like the binomial notion (you are or you aren’t a whale). It’s best to break out the monetization curve into smaller segments; I like to have spending behavior tiered into 4-5 segments.
    - You should always have a time dimension when talking about spending. Lifetime numbers are vanity metrics, daily behavior is what keeps the business alive. I like to have those tiers be 30-day rolling windows.
    - Tadhg’s definition is absolutely the best (if you apply the rolling window); an absolute threshold. The challenge with making it relative (50% of your revenue, top 2% of spenders, etc) is that you may have a game that is full of whales. While that type of player may represent 0.5% of the general population, your player base could be entirely whales.
    - To put $100 per month in perspective, that’s $3.33 per day. Way too low. $400 is better, $13.33 per day. I tend to like $1000 per month, $33 per day for the top tier.
    - 10x ARPDAU is a less than desirable definition for the reasons stated above. In my mind, that would be a lower/middle tier, depending on how strong your ARPDAU is.
    - The ARPPU idea from Jim is better, but gets really wanky for whale heavy games. Taking an average of a long tail distribution will tell you nothing.