Will McInnes has another interesting
post today (“Don’t mention the economy”). It got me thinking about how careful many are “not talking the economy into a recession”.
Using
InfluenceMonitor I had a look at the debate surrounding the UK economy.
Figure 1 (below) shows how frequent a number of issues appear in the debate on the UK Economy. (Normally referred to as the ‘Share-of-Buzz’).
Notice the increased focus on the “recession” from October to January, followed by a very small drop.
At the same time, the focus on “slowdown” is mostly constant.
Because the Share-of-Buzz is unadjusted for influence (all voices count the same) the picture closely mirrors the debate among ‘the general public’.
Now, notice Figure 2 (below). It shows the Share-of-Influence of the same issues. Because this metric is adjusted for influence it a closer depiction of the debate among the top influencers including the influential media.
Notice how the decline in the debate on “Recession” is followed by a corresponding increase in the use of “Slowdown”.
It’s almost like some omnipresent voice said “Don’t mention the ‘R-word’ – the safer word is ‘slowdown’”.
I am sceptical to what extent you can avoid a recession by not talking about it. You may be able to postpone it a short while, but if your economy is not sustainable, you will be hit sooner or later. The market forces will (thankfully) always catch up.
It may be telling that the “optimism” element has all but disappeared from the influence adjusted debate on the UK Economy.
(All graphs as of noon GMT January 22nd 2008)
The stock markets took a tumble yesterday (21st of January) but looking at the online buzz it didn’t really come as a surprise.
Figure 1 (below) shows the number of different mentions of
FTSE in the last 30 days as well as the aggregated sentiment.
Notice how the sentiment drops sharply on the 14th – a week prior to the big share price crash yesterday.
Figure 2 (below) shows the relative Share-of-Influence of certain words used in the context of FTSE.
(Share-of-Influence is calculated by calculating the relative frequency of each word, but adjusting for the measured topical influence of the stakeholder ‘speaking’. Example: When the debate is about ‘FTSE’, anything published by Guardian weighs approximately 3 times anything published by Daily Mail.)
Notice how ‘Recession’ and ‘Crash’ have increased their presence in the debate in January.
Figure 3 (below) shows the sentiment of the context that these words appeared in over the last 30 days.
Notice that in general, ‘Recession’ has recently appeared in a more negative context than the other words – even ‘Crash’.
The relationship between online buzz and business outcome (sales, market share, subscriptions, etc.) is a topic of increasing interest to many businesses.
The thought is compelling: What if you could listen to the online debate and precisely predict how much you (or your competitors) are going to sell next month or quarter?
Actually, in many cases you may just be able to do that.
Before I move on to a real life example of how online buzz and sales are related I would like to take a step back and explain the rationale for why that relation exists.
The typical way an organisation drives sales is via some sort of market communication; be it advertising, PR or some other form of activity.
In its simplest form the chain works like this: A company runs an advertising campaign. Those who are exposed to the campaign may be impacted in several ways. Some may accept the positive message conveyed while others may just become more aware of the brand and the offering presented. In all circumstances the brand moves up in our attention. We become a little bit more aware of the brand than we were before.
This increased awareness (as well as the message presented) leads to increased conversations about the brand and/or the message. The increased awareness (and the offering) usually also leads to more sales. This, of course is nothing new. If it wasn’t true, the £100 billion advertising, marketing and PR business would be resting on a sham. But advertising actually works – and more often better than traditional models capture.
A practical example may look something like this: A mobile phone operator (let’s call it ‘Mobuzz’) runs a fairly traditional advertising campaign on TV, billboards and online and in doing so exposes hundreds of thousands (maybe millions) to their message.
Even if the campaign is pretty average the very least it achieves is that it gives Mobuzz a little higher mindshare among those who are exposed to it. However it does more than that. It actually impacts those who are not exposed to the campaign as well. It does that via word-of-mouth.
Imagine two friends having a chat. One person, who hasn’t been exposed to Mobuzz’ campaign, tells the other that he is considering changing his phone provider as his contract is coming to an end. The other person may offer some advice or relay some personal experience, but because this person has been exposed to the Mobuzz campaign this person is slightly more likely to mention Mobuzz than if he or she and not been exposed to the campaign. This now raises the awareness of Mobuzz with the person who hasn’t been directly exposed to the campaign and even though the increase in awareness is miniscule this person is now slightly more likely to choose Mobuzz than before this conversation occurred.
If you could listen in on this kind of conversation between friends, you would be able to detect which brands gets mentioned more (and less) in which contexts and thereby start to form an opinion on which brands are gaining more (and less) mindshare. This in turn would be a strong predictor of future sales.
Aside from the fact that the thought of such monitoring is scary (and probably illegal) it is also not necessary if somewhere else we can capture a representative sample of the conversation. Enter the Internet.
Online, individuals, media and other organisations ask questions about brands, discuss them, rave about them and air their grievances. By capturing these conversations and processing them correctly we have, for many – but not all topics, a large representative sample of the conversations.
Now, turning the observed online debate into actual sales predictions is not as straight forward as you may think.
You may think that if there is a sudden increase in the debate about brand A, compared with the competing brand B, this will lead to more sales for A (compared to that of B’s). However, that would be too simplistic.
Very importantly, you have to adjust for the ‘weight’ or ‘influence’ of the voices that speak. If I write a story about Hyundai cars on my blog it doesn’t carry as much weight as if The Times driving section or a blogger with authority on this brand does it.
Therefore, raw buzz monitoring, where all voices are treated with equal importance, is mostly useless to predict anything.
There is a lot of good research to show that the more influence a brand accumulates, the larger share of the market it will gain.
This is intuitive as well: If a greater proportion of the talk is about a particular brand, this brand is going to be at the front of mind of more people. And if those who talk about the brand are voices of influence it is intuitive that it will have greater impact than if said voices have low influence.
In fact, most research I am aware of more than suggests that, over time, relative market share among brands will equate to relative share-of-influence.
Figure 1 (below) shows the percentage change in monthly influence and percentage change in monthly (UK) sales for Nissan Pathfinder for the period of February to July 2007. We can see that the two graphs follow a similar path and they appear to be strongly correlated. In fact, the
Pearson product-moment correlation coefficient (a metric often used in PR and Advertising research to measure the relationship between observations and outcome) shows a correlation of 0.99 (the scale is -1 to +1); a very, very strong correlation.
For Pathfinder we can see that the typical lag between earned influence and sales seems to be shorter than one month. This is not always the case but in my experience it is typical for cars that have been on the market for some time and thereby well known to potential consumers.
Figure 2 (below) shows the similar data for Nissan Qashqai. The period here is from March to July 2007 (introduced in February).
At first glance the two variables seem to be less correlated, but in fact the correlation (0.98) is almost as strong as for Pathfinder. There is however, a lag between the earned influence and the sales.
Notice how the change in influence increases sharply from April to May while the change in sales increase slows.Change in influence then falls from May to June, only for the change in sales to follow with a one month lag.
The lag may be caused by the fact that this is a new model. If this is the case then it is likely that the lag will become shorter as more become familiar with the model.
The lag may also be explained by delivery shortages, but regardless of cause, the point is to illustrate that measuring influence is a highly effective way of understanding where the sales are heading.
There is much more to be said about how analysis of online buzz can be used to predict future sales and market share, but I hope the above has raised the awareness of some of the possibilities.
Last week, IBM released a report that forecasts “greater disruption for the advertising industry in the next 5 years than occurred in the previous 50”.
The report is really, really interesting – mostly because it is so thorough and comprehensive. I strongly recommend reading it.
Read a summary and download the full report here
I am very interested in hearing your comments/reactions to the report.
NY Times has a fantastic article today about how big brands are sidestepping the traditional media channels and connecting directly with consumers:
“We want to find a way to enhance the experience and services, rather than looking for a way to interrupt people from getting to where they want to go,” said Stefan Olander, global director for brand connections at Nike. “How can we provide a service that the consumer goes, ‘Wow, you really made this easier for me’?”
“We don’t automatically think about television anymore,” said Joaquin Hidalgo, vice president for global brand marketing at Nike. “There was a time when brands like Nike could tell kids through the medium of television what was cool, what was in, what was not in, because that was the only window they had into the world. That has completely changed now.”
The article is totally in line with what our clients are telling us and well worth a read
I will be speaking at the
Sales and Marketing in Travel Europe 2007 in Berlin on the 23rd of October.
I am looking forward to presenting some new interesting findings there and also announce a new partnership.
If you’re attending or in town then send me a mail and we can meet up.
I was looking at the debate on the problems surrounding the subprime lending crisis in our InfluenceMonitor service today.
The figure below shows the share-of-influence(1) of a number of well known investment banking brands has in the online “subprime”-debate.
Notice how the share-of-influence of Bear Stearns and UBS seems to be correlated: Initially UBS was the brand in focus. In June however, Bear Sterns announced problems with some of their hedge funds. For a while it took the heat of UBS.
However from July and onwards the focus on UBS has been steadily growing while the interest in Bear Stearns has dropped almost correspondingly.
Interestingly Citygroup, Merrill Lynch, Morgan Stanley, ABN AMRO, Credit Suisse and Deutsche Bank mostly escape attention.
The focus on Goldman Sachs however seems to be growing.
(1)
Share-of-influence is calculated by measuring the share of earned coverage a brand gets in a particular context and then factoring in the measured, topical influence of each voice that discusses the brand. So, when the topic is "subprime" a mention in NY Times will count for roughly 50% more than a mention in CNN.com’s money section.
We have previous (
1,
2) written about how Nintendo’s Wii has outperformed Sony’s PS3 and indeed all other comparable games consoles.
Until now the focus has mostly been on the battle between Wii and PS3, but in recent months Microsoft’s Xbox has been making waves.
The figure below shows the relative share of buzz about the 6 major games console brands from June 1st to 23rd of September 2007.
Notice how the buzz levels of PS3 and Wii have been relatively steady while that of PS2 and GameCube has been declining.
Microsoft’s Xbox seems to be the only one with steady growth although it seems to be relatively small.
Tracking share-of-buzz is interesting because we know from research that share-of-buzz and share-of-market converges, so if the present trend continues, Xbox is on route to a small but steady increase in market share.
However, we also know that share-of-influence and market share normally converges faster so it is normally our preferred (main) variable when trying to predict future market share.
(The difference between share-of-buzz and share-of-influence is essentially that the latter includes a weighting of each mention according to the mentioning media’s measured topical influence. When calculating share-of-buzz, all mentions essentially counts the same.)
The figure below shows the development in each brands share-of-influence for the same period.
Notice how much more dramatic the picture looks: PS3 and Xbox are losing and gaining (respectively) share-of-influence much faster than the previous graph would indicate.
This would indicate at least two things: First of all, that the coverage of Xbox more often takes place in media with above average influence on the topic of “games consoles” and similar that the coverage of PS3 is usually in media with below average influence on this topic.
Second, we can predict that the increase in Xbox market share and the decrease of PS3’s similar will be more dramatic, both in terms of magnitude and speed, than the first graph would indicate.
A third observation might be that since Xbox’s increase seems to come at the expense of PS3’s, this might indicate that PS3 and Xbox are fighting for the same audience and are considered substitutes. Those who are considering buying a Wii are more likely to be choosing between buying a Wii and not buying a console at all. If this is true then part of Wii’s success is that it is enlarging the market for games consoles with new customer segments.
Now, we also know from research that if a brand has high share-of-positive-influence its market share and share-of-influence tend to converge even faster.
However, this just makes things even worse for PS3.
The graph below shows the development in sentiment or tone-of-voice of the articles/blogs/forums where each brand appeared in the relevant period.
We can see that Wii and Xbox are at a positive sloped angle indicating that the positive mentions (dramatically) outweigh the rest. PS3, on the other hand, is represented by a flatter curve indicating that the posts on this brand are more balanced (or neutral) and thus on average less positive than the two other main brands.
So this all leads to the question of why Xbox sale is performing so well.
The answer is likely to have several reasons, but look at the figure below that plots the change in debate on Halo 3 and the change in the debate on Xbox for the analysed period.
More specifically the figure shows the change in accumulated influence on a month-by-month basis for Halo 3 and Xbox.
Notice how the lines follow the same trend. In August there was an extraordinary large change in influence for Halo 3 but while it does pull up the debate on Xbox it doesn’t do so with the full force of the change.
The figure below gives a more precise picture of exactly how closely correlated the debates on the two brands (Halo 3 and Xbox) are.
The figure plots the % change in accumulated influence for Halo 3 along the x-axis and the similar change for Xbox along the y-axis. The line is a linear trend line showing the correlation coefficient to be 0.788 which indicates a strong correlation.
Whilst we haven’t proven a causal effect here I think it is at least intuitive to assume that it is Halo 3 that mainly causes the debate on Xbox and not the other way around (although some argument can be made for that).
But if we for a second assume that the causal effect is from Halo 3 to Xbox then we can take our analysis one step further.
We can see from the graph that the elasticity of the relationship is about 0.2 (0.171) indicating that if Microsoft is successful in generating a 1% change in the debate on Halo 3 they are likely to increase the accumulated influence of Xbox by 0.2%.
Because of the strong relationship between share-of-influence and share-of-market the above could be translated into monetary value if we had estimated the relationship between share-of-influence and share-of-market for these brands. However, I don’t currently have access to good market share data for games consoles, so it will have to wait for another day.
UPDATE: I have discovered that I was a bit too quick with the results listed below. In the process where I manually (step 2) removed a cluster of non-blogs (NY Times, CNN, Time, Advertising Age and other large media properties) who are also often mentioned in a PR/Blog context I accidently also removed a handful of very influential blogs including (at least) Shell Holtz, Bulldog Reporter, Hyku and Todd Andrlik. I was simply not careful enough when I cut out the big media (or non-blog) cluster.
I have therefore decided to redo the analysis from scratch, so a new version should be posted in 2 weeks or so.
In my
previous post on this topic I promised to measure the influence of blogs discussing “PR” and “blogs” using citation analysis.
The list appears below but first a few comments on how it was done and how to read it.
How it was done:
1. Using a topical crawl of the Internet, blog posts that discussed the topic of “PR” and “Blogs” in the same article were collected along with blog posts that were sufficiently referenced in this context. (Meaning: If you discuss the topic of “PR” and “Blogs” or being discussed in that context, then you are a candidate).
2. Some blogs, that appeared to be very closely related, were consolidated and some blogs/websites were manually removed because they were not deemed relevant to the context.
3. The posts were analysed for references/citations between them. The citations were extracted and turned into a massive system of simultaneous equations that were solved to provide influence.
4. The influence was normalised to a scale between 1 and 100.
(Many more details can be found in the articles referenced in the Part I of this post).
As influence is a relative measure you read the table like this:
Micropersuasion has (roughly) twice the influence of B.L. Ochman when the topic is “PR and Blogs”; or Top Rank Blog has (roughly) half the influence of Constantin Basturea.
The influence is topical so it is only a good measure when the topic is PR and Blogs. If the topic was, say “PR, Blogs and Measurement” the number (and indeed the ranking) could be expected to be different.
A few comments to the list:
Micropersuasion is clearly in a league of its own. No question about that.
However, Edelman have a large network of employees who run good blogs and while I don’t think they have a deliberate strategy of over-referencing each other, no one can blame the network members for being more familiar with the other Edelman blogs and therefore referencing them a bit more than they might otherwise had done.
You may, if you inspect the list, find bloggers whose position you disagree with. You may think this analysis overrates or underrates them.
If you feel a blog is rated higher on this list than you would expect, it is likely because you are assessing their popularity rather than their influence.
A blog you feel is rated to high is often rated higher because it is read (and referenced) by other bloggers who have above average influence. You could say that such “over-influential” blogs “punch above their (popularity) weight”. Blogs like Into PR, The Blog Herald and Marketing Vox are examples of blogs that are somewhat more influential than their popularity should lead everyone to anticipate.
In the recent weeks a number of blog posts dealing with measuring influence have stirred up quite a debate.
Jeff Jarvis and
Steve Rubel aired their thoughts on the issue.
David Brain and Jonny Bentwood from Edelman’s London office
published something termed “Social Media Index” where they propose a brand new method for measuring on-line influence.
Most of my feelings after having read the article were (far better) summed up by
Jennifer Mattern in her
comment.
The article offers no research, no references to prior research, no logical reasoning for its claims and proposes no way of testing the model.
Can you imagine if the CEO and the Head of Research of a large investment bank published a new options pricing model without actual research and without mentioning the
Black–Scholes model? Of course not. So why in PR?
Actually the very notion that influence is not related to a topic is comical. As I understand it, the article proposes that someone has the same influence regardless of the topic discussed. In other words, TechCrunch (for example) is as influential on the topic of fly fishing or jogging as it is on Silicon Valley gossip. I don’t think that’s the case. We know both from intuition and from research that influence is topical.
At best, the Social Media Index is an indication of popularity. But popularity is not influence. Those who are very popular often have good influence but you don’t need to be very popular to have a lot of influence.
When you measure popularity, all “votes” count the same. However, when measuring influence each “vote” counts with the weight of all the votes leading to the voter.
Take a look at the figure below.
Person A is clearly more popular than person B. However, those who listen to person A do not, to any large extent, go on to influence others.
Quite differently so for B. Those who listen to person B they go on to influence others, who go on to influence others, and so on. The aggregated impact (or influence) of B is in this case bigger than A.
The figure (above) indicates why models that only take the first layer into account will come up short when trying to explain the overall outcome.
In pre-Internet times measuring popularity (or “reach” or “circulation”) was a relatively good proxy for influence because those who read a particular news paper or watched a TV programme did mostly not go on to influence anyone outside their close social circle.
But because of the Internet this has all changed and the social
event horizon is now defined by language rather than physical distance. Both direct and indirect influence must therefore be taken into account when trying to measure influence correctly.
Relative influence correlates better to outcome than relative popularity simply because influence takes the indirect effect into account.
So how to measure influence?
Well, it’s been done for a long time. In the academic community “citation analysis” has been used for decades to measure the influence of academic journals, articles, scholars and universities.
The principle is quite simple: You collect all references made between articles about a particular topic. The references are transformed into a large set of simultaneous equations that, when solved, provides the relative influence of each journal.
Thomson Scientific is probably the leading provider for the academic community.
However, the original science for measuring influence in a linked network was developed by
Wassily Leontief who devised something called the
Input-Output Analysis. It was originally (and still is) used to measure how sectors of the economy directly and indirectly influenced each other.
Leontief won the
1973 Nobel Prize in Economics for this specific work.
Now, you may doubt that knowledge gained in pre-Internet times more than 60 years ago can provide any useful input to explaining how to measure online influence.
But in much the same way as engineers at NASA draw on
Isaac Newton’s work, those who shape the on-line world draw on Leontief’s.
In 1965 Random House published a book by W. H. Miernyk titled “The elements of input-output analysis”. The book deals, as the title says, extensively with Leontief’s work.
This book is cited as a source in the article “Measuring the relative standing of disciplinary journals” by P. Doreian (1988).
Doreian’s article then goes on to be cited as a source by Jon M. Kleinberg in his article “Authoritative Sources in a Hyperlinked Environment” (1998).
Then, same year, an article, which you may have heard of, cites Kleinberg’s:
Lawrence Page, Sergey Brin, Rajeev Motwani, Terry Winograd. The PageRank Citation Ranking: Bringing Order to the Web.
There you have it: From Leontief to Google in 4 easy steps. (There may be shorter/more paths).
Ok, end of part 1.
In part 2: The list of the 80 or so most influential PR blogs measured using citation analysis including their relative influence. Stay tuned.