Monday, 29 November 2010
why brand advocacy doesn't convert to sales? only contextual analysis can answer why! sentiment never will. xfactor case study http://tinyurl.com/3yj73x2
Posted by Brand Aura Social Insights at 11:48
So the double elimination resulted in the loss of Katie and Wagner. Our results from yesterday showed that we successfully predicted two of the bottom three (Katie and Mary) but that Wagner's Twitter analysis did not match the public vote. According to our analysis he looked safe and in fact had trended well all week in context with winning words.
So what went on? This is an interesting question, and gives an insight into the dangers of taking at face value what people post on the internet. If we examine the words that are in context with Wagner based on Saturday's data (i.e. before the elimination), then a picture emerges that concurs with our initial analysis. This can be seen in the figure below, where the closer the word is to the centre, then the closer that word is in context to Wagner.
The analysis shows how there is a campaign on twitter to either encourage people to vote for Wagner, or an expression of the hope that they would like Wagner to win. This does not necessarily mean that people have actually voted, and here we can see a disconnect between what is being said (i.e. Vote for Wagner!) and what is actually happening (i.e. people who tweeted did not vote for Wagner to the same scale). This is important to note from a brand management perspective - just because online content seems to be saying one thing, it does not mean that the users of that brand will actually do it.
When we look at the other words surrounding Wagner in context, we see a different point of view emerging. Words such as: annoying, ruin, going home, shite, too long, shit, kill. These words are all of a negative context and relate to an emerging theme which would show that people have had enough of the joke of keeping Wagner in the competition. In fact there are very few positive words coming through at all.
This highlights again the need to ensure that data sources are identified that cover all demographics so that the data analysis is not too biased on way or another.
If we turn to Katie, we can again analyse what words are in context with her based on Saturday's data (i.e. before the elimination). Again, the closer the word is to the centre, then the closer that word is in context with Katie.
This time we see a different picture. We do not see words like win or vote come up close in context to Katie. Instead we see a number of different contestants such as Mary, Wagner and Cher. This is related to discussions that were being had by the general public on who would be likely to be eliminated, as we see by the words Sunday and eviction also coming through. We also see Olly Murs coming through strongly in context, relating to the interview that he gave where he talked about Katie. We also see a large number of negative sentiment words, such as: murdered and killing, which relate to her performance of Sex on Fire. Interestingly we do not see much comment on her second song 'Everybody hurts', which does not appear to come through as closely in context.
However it can be seen here that there is largely a negative reaction to her performances and so it is clear why she was voted out.
Finally we can look at Mary. Again the analysis is based on Saturday's data (i.e. before the elimination) and the closer the word is to the centre, the closer that word is in context with Mary.
Here we see very little comment on her singing performance. However we see a number of words coming through in context that may initially surprise us as being in context with Mary - bearing in mind she is over 50! We see the word thong very close in context along with other words: lucky, Tesco, value, smells. Although this would at first glance appear to be some form of mistake, it is related to the story that Mary has taken to wearing her 'lucky Tesco value thong' when she goes on stage, but that since she has been wearing it for every show, it now 'smells'. Not a very nice mental image, whichever way you look at it!
This highlights once again how the analysis of the data in a contextual manner brings out unexpected and new information about a brand or person, and gives an insight into the public's view on any particular topic.
We will continue our analysis in the run up to the semi finals next week, where we will focus on the positive sentiment to identify who is most likely to be safe from the dreaded elimination. Will Twitter match the actual votes made? Now that Wagner has been removed from the competition, will we see more of a convergence of twitter trends and actual voting trends? In other words, will brand loyalty as evidenced on twitter convert to real votes in the show? As always, we will complement our scoring analysis with contextual analysis to reveal the reasons behind the voting decisions that the public have made - and highlight again how contextual analysis is the only method that fully explains the diverse views of the public and what they are thinking.
Sunday, 28 November 2010
We are rapidly losing contestants now, and so we can look at examining the scores for them all together. The figure below looks at the analysis for each contestant based on tweets over the past week. Each day is analysed separately, and is shown in terms of how close in context each contestant is to words such as 'win'.
The graph shows the rank of each contestant for each day. So Wagner actually scores the best for almost every day (i.e. he ranks as number 1), whereas Mary and Rebecca are scoring poorly with one of them normally 7th (i.e. worst rank) with the other 6th.
As we can see in the figure, there is a huge amount of support for Wagner, which would suggest that he will be safe again this week. This may surprise some people but shows the depth of feeling and perhaps the unique British sense of humour that is looking to produce an unusual winner for this most popular show. Certainly, there has been a campaign online to keep Wagner in the competition, and judging by the analysis here that is certainly being successful. Whether the twitter public want to wind up Simon Cowell or whether they genuinely enjoy Wagner's performances, the end result appears to be that he will again be safe this week, if the general public reflects the view on twitter.
This analysis may not however reflect the views of the voting public, which as has been shown previously has a different demographic to that one twitter. However, it will be interesting to see if his support continues to the same level should he indeed remain in the show when we start to approach the final.
In terms of who is in trouble, the bottom two according to the analysis on twitter is again Mary and Rebecca. We can see that these two were the bottom two last week as well, which is a worry for both of them since we can see that their support is therefore consistent. Last week however Cher and Paije were in the actual bottom two of the show, suggesting that Mary and Rebecca's support if under-represented on twitter. If we look at Cher's support last week and compare to this week, we see that she is doing better this week than last, suggesting that she may be ok and avoid the bottom two.
Katie as always is towards the bottom of the analysis, and we see an opposite trend for her. Whereas last week she performed better on our twitter analysis (finishing above Cher), after yesterday's show she has dropped into our bottom three, suggesting that she will be in trouble as she has a vocal support on twitter that is over-represented when compared to the general public.
This evenings show is a double elimination, so TWO people will go. Going from our analysis then, two of Mary, Rebecca and Katie look likely to go, however as we know this does not always reflect the views of the voting public. It will be interesting to see how the votes pan out! Will Wagner remain in the show to sing another week?
Posted by Brand Aura Social Insights at 14:20
Monday, 22 November 2010
So was it surprising that Cher reached the bottom two yesterday? Certainly, she was not bottom two according to the twitter comments but as we know, this is not necessarily reflective of the voting public.
As we have seen from previous weeks, it is clear that some contestants garner a large support despite not being tweeted about (see Paije until this week and also Mary), whereas others have a large body of apparent support on twitter but find themselves in the bottom two (see Katie before this week for example of this).
Cher also has a large body of support on twitter, but what does her support look like when we examine what is being said about her in context?
If we analyse the data from Sunday up until the show (i.e. before 8pm) then we see some surprising results. The picture below shows words that are in context with Cher. The closer the word is to the centre, the closer in context to Cher that word is.
We see a large number of negative words such as: bottom, annoying, rubbish, crap, tank. There is some good sentiment type words too but there are fewer of them, such as: good, babe.
This comes from people tweeting on the Sunday, so perhaps reflects more the view of the general public and not her ardent fans who will have enjoyed whatever she did.
We can do the same analysis for Rebecca, who based on Saturday's tweets we had in the bottom two. This analysis can be seen in the figure below, where as before the closer the word is to the centre, then the closer in context that word is to Rebecca.
Here we can see that although there is some negative sentiment such as: murdering, hate, crap. However we see alot of positive sentiment also coming through such as: vote, favorite, want, win, supporting, love. We also see a fair amount of context coming through based on what Rebecca was wearing, such as: wearing, vintage, necklace.
This shows much more positive analysis than Cher over the same period (Sunday up until the 8pm results show). We will extend this analysis over the week to look at how changing context on the Sunday reflects the public's opinion on the voting result and subsequent loss of Paije from the competition.
Posted by Brand Aura Social Insights at 17:59
Sunday, 21 November 2010
Very quick post today, simply to show predicted bottom two based on our analysis of tweets from Saturday.
Our ranking is based on how closely in context the contestants are to positive sentiment words such as 'win', or 'winner'.
The bottom four according to our analysis is (with 1st at the worst, and 4th as the best of the four):
This means our bottom two this week are: Rebecca and Mary, rather surprisingly. However as we saw last week analysis of trend may show more, which we will focus on tomorrow.
Posted by Brand Aura Social Insights at 20:22
Monday, 15 November 2010
So Aiden and Katie reached the bottom two in last night's show, with Aiden receiving the fewest number of votes and was evicted from the competition. It is interesting to note that the bottom 4 as we predicted yesterday, were the last 4 contestants in the show as they were made safe. So although Dermot states that the contestants are told that they are safe in no particular order, it would make sense to increase the tension to not have contestants that are obviously safe in the 'drop zone'.
We have had a similar result before - Paije does not trend well in twitter, and there are very few comments made about him. This is interesting in itself, since he has not yet been voted into the bottom 2, so there is therefore a large demographic voting for him that does not have a presence on twitter. This highlights the need to ensure that data sources that cover all demographics are identified for any marketing or PR campaign.
The same largely goes for Mary too, although it is noticeable that her support on Twitter has diminished over the past few weeks, suggesting that she may be in trouble in the coming shows unless she is able to either connect better with the audience or produce better singing performances to keep her in the competition.
It is also interesting to examine the pattern of the results over a number of days. We said yesterday that Mary and Paije were in the bottom two based on Saturday's results only, however if we average over a number of days, then we end up with the following result, that can be seen in the chart below:
The analysis shows that if we look at the average score over the past 10 days, we see that Aiden has in fact been overall in the worst position. We see that Mary and Paije have similar levels of support, and that Katie out of the four contestants analysed here has a core following on Twitter that means she consistently does better.
This analysis therefore perhaps puts yesterdays result into perspective. Katie has a core support which will vote for her regardless of her performance. From previous experience on X Factor we can see that this is necessary, and explains why she continues in the competition, despite there being so much negative press about her.
Aiden on the other hand, has fluctuated dramatically over the past 10 days, with his support not as consistent. For a contestant in this position, he is always one poor performance away from being voted out.
The analysis also indicates that Paije and Mary may also struggle in the coming weeks, although again good performances may buffer them from dropping into the bottom two.
Posted by Brand Aura Social Insights at 09:46
Sunday, 14 November 2010
Quick post today - simply showing our analysis of the contestants in trouble according to the analysis of twitter data.
The analysis can be seen below, and appears to show that Paije and Mary are in the bottom two. However, previous results have shown that Paije in particular does not result in a large number of tweets and so the results may be bias against him in particular.
The results for the other contestants showed a clear gap between these four and the other contestants. However as noted previously, these results may not reflect the vote from the general public if the demographic on twitter is not similar to the voting public.
This week does appear to not be as clear cut as previous weeks, although as noted above the four shown below do appear to be cut off in terms of popularity from the other contestants.
We will continue this analysis tomorrow after the publication of the bottom two and who ends being evicted from the show.
Posted by Brand Aura Social Insights at 15:27
Monday, 8 November 2010
The results show last night had Trayc and Katie in the bottom two, which tallied with our predictions of yesterday. As a follow up we thought we'd look at what was said in context with Katie to see if there are any clues as to why she is not garnering the votes and what the public are saying about her.
The image below shows the words closest in context to Katie. The closer the word is to the centre of the cirle, then the closer in context the word is.
The analysis above shows the words that are in context with Katie. As can be seen, there are a large number of negative comments, such as 'awful', 'cringing', 'worst', 'pained', 'poorly', 'rubbish', and 'hate'. It is also noticeable that Katie comes in close context to Wagner, which is not necessarily a positive connection since although Wagner has not yet been in the bottom 2, he is seen largely as a figure of fun and not a serious contestant by the majority.
It is not all bad news however, there are some positive words such as 'fit', 'girlfriend' and 'best'. So there is hope for Katie although it would seem from this analysis that the majority of the public do not connect with her performances in the same way as with other contestants.
Posted by Brand Aura Social Insights at 10:32
Sunday, 7 November 2010
So ahead of the publication of the results tonight we thought we would again show analysis of the tweets that have been published around the Xfactor this week.
This time around, we will try to take into account the bias that seemed to show through with twitter from last week. Although we consistently had Belle Amie in the bottom two we also had Paije performing poorly and yet he was not in the bottom two. In addition, Katie was voted into the bottom two and we had thought she would be safe based on online comments and was only saved as Belle Amie had fewer votes than her.
The analysis of the online content for this week can be seen in the graph below (with the dates shown on the bottom, with yesterday's tweets coming through for the 6th Nov), which shows the ranking for each contestant against each other (i.e. 1st means that they scored the highest for that particular day):
We have included Cher and Mary this week as both of their scores were of a similar level to Aiden.
The analysis is interesting for a number of reasons. Firstly we can see a definite trend with respect to Paije, and for the first time we see a large number of positive comments about him coming through on twitter. The comments for Cher started off as the highest ranking but dropped towards the end of the week and on Saturday in particular, which is a worry for her.
The same goes for Aiden, who has a worrying trend for him in that he is consistently near the bottom end of the scores. Perhaps people are struggling to relate to his intense singing style? Mary should also perhaps be concerned since a good trend in tweets during the week is reversed on the Saturday, perhaps reflecting her poor performance on the night.
Trayc yet again does not score highly, and would appear from this analysis at least to be at risk. However this was also the case last week, and the twitter demographic does not necessarily match the voting public.
Yet again perhaps to some people's surprise Wagner does not appear on this list as his scores were far in excess of the ones listed here.
We will pick up this analysis tomorrow, after the bottom two and eviction of one of the contestants has been completed.
Posted by Brand Aura Social Insights at 14:43
Monday, 1 November 2010
So in the evening voting show Belle Amie were shown to have received the lowest number of votes and left the competition. Does this tally with our predictions of yesterday?
The bottom 6 according to our analysis on twitter chatter contained both Katie and Belle Amie, and so this aspect of the analysis was correct. However we had Paije as the worst performer along with Belle Amie and Trayc. Xfactor does not release the exact ordering of the contestants, so we don't know who finished 3rd from bottom. But why does our analysis show a different bottom two than were voted for by the public?
We can see the reasons why when we examine the rankings that we posted yesterday in a little more detail.
The chart below shows normalised positive sentiment analysis for the contestants every day based on twitter comments. We haven't looked at negative sentiment since the Xfactor requires you to vote FOR your favourite contestant - therefore the more positive that people are being about a particular contestant, the more likely that contestant will do well in the public vote. The analysis is normalised so that each day can be more easily compared against each other - there is evidently much more chatter on the Saturday and Sundays about the Xfactor when the program actually airs.
We can see in the figure that Aiden tends to have the most positive comments, but we also see Katie doing well. Looking at the positive sentiment, it is clear that Paije for example consistently barely scores at all - this is due to the fact that there are very few comments being made about Paije at all on twitter. The same is true for Trayc and Belle Amie. On a number of days there is no chatter about Trayc at all, and although there is consistent chatter about Belle Amie it is small in comparison to Aiden, and even Katie.
This highlights a couple of obvious conclusions that all digital marketeers should be very aware of! Firstly, the demographic of twitter does not match that of the voting public. It may be possible to identify trends in the populace - for example the high scoring of Wagner throughout the week indicating that there is a large body of support for him. However particularly at this early stage where the vote is split it can be difficult to predict with any accuracy how the general public will vote from the twitter data alone.
This highlights the second conclusion which means that in order to improve our predictive capability, alternative data sources should be identified that more closely match the voting intentions of the general populace. It is unlikely that a single data source will cover this wide demographic and therefore a combination of data sources should be used.