★ TouchArcade needs your help. Click here to support us on Patreon.

Increased App Rankings Fluctuations

01-29-2012, 09:56 AM
#1
Joined: Dec 2011
Location: Philadelphia, Pa
Posts: 124
Increased App Rankings Fluctuations

Wondering if any other developers interested in the analytics and algorithms involved in the app world have noticed the following:

Ranking Increase Indicator

A ranking increase is normally indicated or forecast by what I will refer to as Tetons.

Typically hourly rankings may look something like this:
1pm, 2pm,3pm,4pm,5pm.....
21,21,22,23,21,22,21,23,24,22.....

However right before a move up in rankings you will see a pattern similar to the following:
1pm, 2pm,3pm,4pm,5pm.....
21,21,22,9,21,9,21,9,9,10,9.........

The first two 9's in the above sequence will form peaks or Tetons on your hourly ranking graph. Typically the Tetons are in sets of 2 or 3 prior to the "permanent" move to a new rank.

This pattern had seemed to be a fairly regular and predictable occurrence, until the past 24 hours.

Over the past 24 hours, i have seen the Tetons coming in sets of 7-12. Rankings literally bouncing back and forth hourly for 12+ hours now in almost every country's itunes store. These wildly fluctuating ranking over such a long period are not something I have seen occur in the past 3 months of observing ranking movements.

I was wondering if more seasoned developers have seen this abnormal amount of ranking fluctuation before and whether it may indicate anything. Being from an SEO/Analytics background the first thing it brings to mind is an algorithm change in Google SERPs and how this is very similar to how natural search ranking behave when Google is mucking about with that algorithm.

Any thoughts?
01-29-2012, 12:05 PM
#2
Joined: Dec 2009
Location: Berlin, Germany
Posts: 1,674
Hehe mr numbers again... Ranking are pretty much unrelevant if you dont weigh them heavily against actual sales numbers.. everything below 3 digit sales gives you wonky rankings due to the huge amount of apps catfighting about the same bottom spots.. So not sure what rankings you are checking but if thoose are sub sub categorie ranks you are wasting your time.
If thoose are top100 rankings in the us then that would be weird.

01-29-2012, 02:19 PM
#3
Joined: Dec 2011
Location: Philadelphia, Pa
Posts: 124
This is for the thread:

An hour after I posted this my app went from (19) top 10 stores places in Ipad Dice to (49) Top 10 stores place in Ipad Dice. Saw similar jumps in Strategy category though only into top 100's as that category is a bit more competitive.

If any other developers are tracking hourly rank changes and saw a similar ranking spike in the past couple hours, please lmk as I am trying to unravel the ranking algorithm.



This is for Mr Ugly the resident Thread Crapper:

Ugly (Daniel) what is your goal on here? Apparently it is to comment on every thread without allowing facts to get in the way.

Is it true you have 2 apps on the App store(neither of which has been updated in over a year), yet 1200 posts on the forums?

And what exactly are you basing your comments on if you have 2 apps, neither of which are being downloaded, thus creating no analytics to look at?

Would the fact that my app currently has 85 Top 100 stores places for iphone, and 89 Top 100 places in the iTunes store be enough for you to allow that I may be capturing enough data to share insights that are far more fact based then what comes from your nonstop trolling?

Must you thread crap on every single thread although you have nothing to offer in the way of expertise, experience, or data?

You might want to stick to your Hero Academy threads and leave the developer threads to people actually developing and sharing thoughts to make all apps better.

Every time you post something you prove you know less than I previously thought humanly possible....
01-29-2012, 03:25 PM
#4
Joined: Dec 2011
Location: Czech Republic
Posts: 43
What tool are you using to track your hourly ranking?

Froggies on the Appstore
Follow us on Facebook and Twitter

http://www.froggies.com
01-29-2012, 05:35 PM
#5
For what it's worth, Foozle just went from 12 stores to 19 in the last 12 hours. Including moves into the top 400 in Canada, UK, Australia, and China. These moves all happened between 7am-8am pacific time and have been flat since first popping into the top 400. As to wether these are sales, I'd hope so, but I can't tell you for sure until tomorrow.
01-29-2012, 06:20 PM
#6
Joined: Dec 2009
Location: Berlin, Germany
Posts: 1,674
You wont unravel the the ranking system from down below.

Its not ipad dice charts

Its ipad overall->games->dice and therefore completly meaningless, if you think thoose data are worth anything it shows that you dont see "the forrest because of all the trees".

You are the one accusing others developers of something not me.

And cool that you found a search function, congrats.

Your problem is you assume too much and know less. Especially about me. If you really want to get an proper inside into the system go here http://www.appannie.com/products/intelligence/ and pay for a subscription to all and every numbers. This can be enlightened. If you cant afford, join with others to pay up.

And yes i enjoy hero academy, a brilliant little strategy game.. should be something for you with your analytics skills.. at least its easy to quickly accuse other fellow hard working devs to be shills. Thats your style.

So whats the point of this thread again? Saying there are weird fluctuations at the bottom of the store, bummer. You know the sun is rising tomorrow? Bummer!

What exactly do you want to unravel? That you need to sell one unit a week in , venezuela to chart?

Maybe i dont see the point. The goal, the bigger picture. Good luck with that anyway cheers

Last edited by mr.Ugly; 01-29-2012 at 06:33 PM.
01-29-2012, 07:28 PM
#7
Joined: Nov 2009
Location: Wellington
Posts: 873
If you games are outside of the Top 200 overall, you can find they naturally bounce around quite a bit.

That being said, I've noticed what appear to be some anomalies e.g. for a universal game on iPhone a "big" dropoff followed by a similar bounce back the next hour, seemingly coinciding with the complete opposite happening on the iPad charts.

Given how closely I'm tracking some of these titles, how high up the charts they are, and the relative stability of their rankings and sales otherwise, I'm just putting it down to a quirk of their ranking algorithm (perhaps for universal games?) and am not worried about it actually affecting sales performance much.
01-30-2012, 02:33 PM
#8
Joined: Dec 2011
Location: Philadelphia, Pa
Posts: 124
While I agree this attempt to unravel the algorithm may be fruitless and folly, I am not going to accept it is based on the word of a German kid with 2 apps.

The goal of the thread is simply to allow others who are numbers guys and are interested in trying to work together to see if we can glean information about how the apple store works when it comes to ranking games.

Our conclusion may be it is hopeless to predict anything. However I think the results will be we will gain somewhere between a vague and a fairly decent idea of how apple works their system.

There is published accounts that someone figured out the ranking system 2-3 years ago and Apple subsequently changed it. For what it is worth this was the finding:

Ranking = 8*(daily sales) + 5*(day-1 sales)+3*(day-2 sales)+1*(day-3 sales)

There was another published report stating the formula may be closer to:

(D1+D1+D1+D1+D2+D2+D2+D3+D3+D4+D5+D6+D7)/13 (D=Day)

What none of these reports mention is when is this calculated and what is it compared to in order to tell what is hot or not.

So here is the hypothesis that I am trying to test:

1) I think Apple tests the ranking twice a day around noon and midnight apple time PST(when I say around I mean the hours around not the minutes). While ranking seem to flutter during the day, they seem to jump during those times more often than at other times.

2) I think that apple compares the recent sale algorithm (something like one of the ones above) and then compares that to the lifetime average of the app and other apps in the genre.

So if you recent ranking algorithm score is significantly higher than your daily avg, and/or the avg or variance of other apps in your genre, your app while be singled out for Whats Hot or What's New.


Would like to know if other developers following daily rankings could see if the hypothesis either fit or do not fit what they see in their analytics.
01-30-2012, 05:35 PM
#9
Joined: Dec 2008
Location: California, USA
Posts: 2,143
I have no clue as to the workings of the App Store ranking system. Seems to correlate to sales...? lol

However, since you mentioned "What's Hot", I have a combination question/gripe.

Daredevil Dave has done great and I'm satisfied with all that. What bums me out is that the game is never featured in What's Hot, even in it's own Racing sub-category. Dave has been as high as #6 or 8 in the app store, and has been in the top 50 in racing for over 1.5 years. He was featured upon release and shown lots of love, but ever since then....nothing.

It wouldn't bother me if I didn't see games that aren't even in the top 100,000 (for a reason IMO) getting featured there routinely. Old ones too, like mine. The game hasn't been updated in forever, but neither have some of these other games. I can't see a reason for never getting featured. Probably wouldn't make much of a difference, but still...more curious than anything.

Back to the main topic. I haven't seen any of what you've mentioned (OP). I watch the rankings every day but I guess I'm not interested enough in the inner workings. I know that when you're selling only a few hundred copies a day that it takes very little to move up and down the rankings. If you're selling less than that, you're going to jump around in rank a lot, as it takes very few sales to make a big difference.
01-30-2012, 06:24 PM
#10
Joined: Dec 2011
Location: Philadelphia, Pa
Posts: 124
Sorry to hear about poor Dave.

Be interesting to see what happened if Dave got an update since his daily average is probably fairly low right now compared to what it would get when it hit as a new release again.

My guess is if you did a new release and supported it with a small campaign for a week, it would have a better chance to show up as what's hot.

I would agree that many games featured are not good games, but quality of game has nothing to do with this. The only thing Apple picks out by "hand" is what we're playing. Everything else is done by algorithm which means it can be figured out and "manipulated" by those with insight into the algorithmic formula.