#STATS Sample suggests that the general effect of this change on Klout scores is approx. -20%,

@peteralderliestPeter Alderliesten

New Klout score of regular accounts minus 20%| bit.ly/ufITcf #klout #tweetbewust #in

1 minute ago via TweetDeckFavorite Retweet Reply

On Wednesday, October 26, Klout started using new algorithms on Klout scores. A limited sample suggests that the general effect of this change on Klout scores is approx. -20%, with scores of ‘spam’ accounts decreasing more and those of IRL celebrities seeing little or no effect. The data below…

Regular accounts

Below an overview of the Klout scores last week of regular parties in social media – again, with my account added in for the fun of it - compared to those of Wednesday, October 26, the first day of the new algorithms:

 

.	Klout score
Publishing accounts	Oct 19	Oct 26
Frankwatching	71	63
Dutchcowboys	75	63
MarketingFacts	72	57
.	Klout score
Personal accounts	Oct 19	Oct 26
RoosvanVugt	75	62
PaulusVeltman	65	51
Carmenvriesema	57	47
BasW(estland)	63	49
TonyBosma	66	53
peteralderliest	26	15
.	570	460	19,3%

 

Spam in Twitter

The same numbers, now for the ‘followback’-accounts identified earlier:

 

.	Klout score
Followbacks	Oct 19	Oct 26
Teamfollowwacky	83	77
teamfollowwack	78	69
Headlessgang	75	57
loopinloops	69	48
glauciacalasans	67	48
folloowbacks	66	48
Unfollow_tool	62	40
operationRT	55	36
afollowbackstar	50	32
allfollowmax	50	44
Followback68	48	21
Follow500perday	45	20
Folowmeback	45	35
.	793	575	27,5%

 

 

Both ‘Shout Out-accounts showed an average decrease of almost 35%, identical to the effect on my own social experiment, lifeevents2:

 

.	Klout score
Only Shout outs	Oct 19	Oct 26
alfianstracci	62	38
bigrichmufucka	67	47
lifeevents2	39	25
.	168	110	34,5%

 

Extra control groups

As an extra control measure, I compared the Klout scores of a couple of celebrities, where it became clear that these were virtually not affected.

 

.	Klout score
Celebrities	Oct 23	Oct 26
charliesheen	79	81
Justinbieber	100	100
BritneySpears	86	85
amywinehouse (official twitter account)	66	62
.	331	328	0,9%

 

Now, how about the 10 most influential accounts on the topic ‘Klout’ as measured by Klout on October 23? The average decrease in score here is approximately 16%

 

.	Klout score
Most influential	Oct 23	Oct 26
Klout	85	85
Cindy Ratzlaff	71	56
Social News	53	42
Moo Print	72	62
Mark Schaefer	81	64
Pam Moore	79	60
Aaron Lee	85	65
Calvin Lee	79	69
TNW social Media	64	60
jaybear	82	67
.	751	630	16,1%

 

 

The parties that had provided the best content om Klout in the last 90 days showed an average decrease of – again – 20%, where ThemelisCuiper had a decrease in score that was an outlier and 2 accounts I could not trace back anymore and were excluded from the calculation (sorry if I have offended anybody here). If ThemelisCuiper were excluded from these numbers, the decrease would be 11%.

 

 

.	Klout score
Best Content	Oct 23	Oct 26
ThemelisCuiper	65	15
Calvin Lee	79	69
Social Today	70	?
Megan Berry	71	59
Joe Fernandez	67	66
Garin Kilpatrick	70	52
Klout	85	85
Michael Todd	81	?
.	372	331	11,0%

 

 

Conclusion

Although there is no claim to objectivity or comprehensiveness here, the above data suggests the following impact on Klout score per group of users. I added an interpretation of these numbers in the last column:

 

IRL Celebrities	0
Best content on Klout	-11%	-1/9
Influential on Klout	-16%	-1/6
Normal accounts	-20%	-1/5
Followback	-27,5%	-1/4
Shout outs	-35%	-1/3

 

So the change may not have been to everybody’s liking, but this data suggests that it is at least in the right direction.

I am still unsure how to interpret the score for my personal account, which dropped more than 40%, apparently in a category of it’s own. (Despite a trebling of the website visits in the last week resulting from a significant increase in mentions and RT’s and so on.)

However, definitive proof that this new algorithm is still far from perfect is given in the data from twopcharts below: @pamgreeble is scoring 48 on the new Klout algorithm, it is 4 days old, has 2,343 followers, 105,037 tweets(!), and (recent) tweets are limited to RT’s of tweets with #followback. You have your work cut out for you, Klout!