0
1
00:00:00,360 --> 00:00:08,160
Now one thing to remember when you are asking freelance designers for app icons is that never just ask
1

2
00:00:08,160 --> 00:00:08,820
for one
2

3
00:00:08,820 --> 00:00:16,850
icon. Always ask for at least 5 or 6 and have some slight variations to those icons.
3

4
00:00:16,860 --> 00:00:23,130
So for example, you know the one in the left hand corner you can see these are on a similar theme but
4

5
00:00:23,130 --> 00:00:24,270
slightly different.
5

6
00:00:24,420 --> 00:00:29,420
And the reason why you need this is because you will need to split test your app icon.
6

7
00:00:29,430 --> 00:00:30,460
Now what does that mean?
7

8
00:00:30,480 --> 00:00:36,510
That means that you are essentially experimenting to find out which app icon performs the best namely
8

9
00:00:36,510 --> 00:00:40,620
which ones produce the most downloads which ones are most people drawn to.
9

10
00:00:40,770 --> 00:00:46,330
So that means that ideally you want 50% of your visitors to see variation A
10

11
00:00:46,680 --> 00:00:53,430
and you want 50% of them to see variation B and then you look at the two groups and compare which
11

12
00:00:53,430 --> 00:00:58,620
ones had higher conversions namely which app icon produced the most downloads.
12

13
00:00:58,620 --> 00:01:05,250
So then after you've done that, you will find that one out of these five or six icons tends to perform
13

14
00:01:05,250 --> 00:01:12,810
the best and then you go back to your designer and ask them to make micro variations on that app icon
14

15
00:01:12,870 --> 00:01:15,950
and then go back and test again and again and again.
15

16
00:01:16,110 --> 00:01:22,710
And the idea is that by the end of all this testing, you will end up with the perfect app icon or as perfect
16

17
00:01:22,710 --> 00:01:23,790
as can be
17

18
00:01:23,790 --> 00:01:25,670
that produces the most downloads.
18

19
00:01:25,680 --> 00:01:31,260
Now a lot of marketing, you know, no matter if its internet marketing or social media marketing or app marketing,
19

20
00:01:31,320 --> 00:01:37,040
it's all about experimentation. And you will have some hypotheses as to what may or may not work,
20

21
00:01:37,200 --> 00:01:43,680
but at the end of the day all that matters is trying it out on actual users and see which ones produce
21

22
00:01:44,010 --> 00:01:46,320
more revenue or more conversions.
22

23
00:01:46,740 --> 00:01:50,460
So do A/B split testing is actually relatively straightforward
23

24
00:01:50,460 --> 00:01:57,000
on the Google Play store. All you have to do is upload several app icons and Google will automatically
24

25
00:01:57,000 --> 00:02:04,920
display a random icon to new users who are viewing the app download page. And based on the conversion
25

26
00:02:04,920 --> 00:02:12,360
rates i.e. whether if that particular icon led to a download, they will reveal to you the data that shows
26

27
00:02:12,420 --> 00:02:14,280
which icon performs best.
27

28
00:02:14,280 --> 00:02:19,770
And funnily enough in my experience it's usually not the one that you think looks nicest.
28

29
00:02:19,770 --> 00:02:25,190
So this kind of experimentation is really really good for pointing out those unintuitive gains.
29

30
00:02:25,350 --> 00:02:31,980
For example did you ever notice that most of the top strategy games on the app stores all have the same
30

31
00:02:32,010 --> 00:02:37,500
app icon and its basically the main character with their mouth open looking like they're shouting something
31

32
00:02:37,530 --> 00:02:42,830
and looking far into the distance? And I mean there is a little bit of psychology behind it because there's
32

33
00:02:42,840 --> 00:02:47,860
nothing that draws a human in more than another human face or human esque face.
33

34
00:02:47,940 --> 00:02:52,770
And then having the human face looking far into the distance you're wondering oh what are they looking
34

35
00:02:52,770 --> 00:02:53,070
at?
35

36
00:02:53,070 --> 00:02:59,040
And then having the so-called raw face or action face means that you know there's a lot of emotion going
36

37
00:02:59,040 --> 00:03:00,420
on in that image.
37

38
00:03:00,420 --> 00:03:02,760
Now some of this is cause and effect
38

39
00:03:02,930 --> 00:03:07,110
namely the app icons for Clash of Clans and game of war.
39

40
00:03:07,170 --> 00:03:09,730
Both have this kind of app icon so there's a lot of--
40

41
00:03:09,820 --> 00:03:11,160
so there's a lot of copycats.
41

42
00:03:11,190 --> 00:03:17,070
But at the same time, I can bet you that these app icons have been split tests to death and they have
42

43
00:03:17,070 --> 00:03:22,680
found through data that this is what draws people in in those fraction of a second while you're scrolling
43

44
00:03:22,680 --> 00:03:24,020
through the App Store.
44

45
00:03:24,330 --> 00:03:31,680
So my point is not necessarily to find a main character, open their mouth and make your app icon just an action
45

46
00:03:31,680 --> 00:03:31,980
mouse.
46

47
00:03:31,980 --> 00:03:37,320
My point is is that the data can sometimes be really revealing and give you unintuitive conclusions.
47

48
00:03:37,320 --> 00:03:42,990
But seriously though, check out the top grossing strategy game and check out how all the icons are essentially
48

49
00:03:42,990 --> 00:03:43,690
the same.
49

50
00:03:43,800 --> 00:03:48,080
So for Google Play, implementing split testing is very easy.
50

51
00:03:48,210 --> 00:03:51,180
But for the iOS App Store it's not so easy.
51

52
00:03:51,180 --> 00:03:57,780
Despite promising us the same feature for years and years now, iOS app developers still don't have access
52

53
00:03:57,900 --> 00:04:01,550
to this experimentation feature or split testing feature.
53

54
00:04:01,770 --> 00:04:09,400
And instead, what you have to do is manually update your app with a new app icon and compare say month
54

55
00:04:09,400 --> 00:04:14,970
to month how your app is performing and presumably based on the change in the app icon or the change
55

56
00:04:15,300 --> 00:04:17,370
in your screenshot.
56

57
00:04:17,410 --> 00:04:19,020
Now there's two problems with that.
57

58
00:04:19,020 --> 00:04:24,170
One is that if you've done any sort of science you'll know that that is not a valid comparison.
58

59
00:04:24,180 --> 00:04:30,330
It has to be side by side because you know this month you might have been featured in Product hunt whereas
59

60
00:04:30,360 --> 00:04:32,090
you know last month nobody heard about you.
60

61
00:04:32,100 --> 00:04:33,900
So there's a lot of confounding factors
61

62
00:04:33,900 --> 00:04:39,420
basically. On top of that, there's a website called App Review times which tracks on average how long
62

63
00:04:39,420 --> 00:04:42,000
it takes for an app update to go through.
63

64
00:04:42,030 --> 00:04:47,460
So at the moment looks like it takes about 4 days for your iOS app update to actually be reviewed
64

65
00:04:47,490 --> 00:04:52,930
and then released because on the iOS App Store it's done by humans and it can be a little bit-
65

66
00:04:52,980 --> 00:04:55,480
It's sped up in the last year compared to before.
66

67
00:04:55,490 --> 00:04:59,660
Previously like two years ago we were waiting something like 2 weeks to 3 weeks
67

68
00:04:59,850 --> 00:05:05,620
for every update to actually get reviewed and pushed through. But still 4 days is a long time and
68

69
00:05:05,620 --> 00:05:09,520
that means you can only really do two experiments a month for example.
69

70
00:05:09,520 --> 00:05:15,750
So it's not really ideal. An alternative to that is using a service like STOREMAVEN.
70

71
00:05:16,030 --> 00:05:22,990
And essentially the idea here is that you send your traffic i.e.  people who potentially download your app towards
71

72
00:05:22,990 --> 00:05:30,340
a page that looks identical to the app store download page. And based on their conversions i.e. whether
72

73
00:05:30,350 --> 00:05:39,160
if they hit the download button or not, you can compare how different icons perform, how your app screenshots
73

74
00:05:39,160 --> 00:05:40,990
perform, how your description perform.
74

75
00:05:40,990 --> 00:05:46,970
So essentially allowing you to split test your app download page and bypassing the app store itself.
75

76
00:05:47,200 --> 00:05:53,430
Now the downside to that is that it's an extra click for your user to potentially give up on downloading.
76

77
00:05:53,440 --> 00:06:00,480
But on the upside, you do get a lot of useful data that will improve your download rate in the future.
77

78
00:06:00,730 --> 00:06:05,350
And they have a 30 day free trial. So check it out if app listing optimization is something you want
78

79
00:06:05,350 --> 00:06:06,520
to get serious about.