Friday, November 30, 2007

Viral Marketing, Facebook and some new research

Since I posted about FB's Beacon service - where you can have your purchases from selected sites listed in your FB news feed - there has been lots of talk over the merits of the system.

I personally don't believe it's useful. At least not for me. I have many friends with many different tastes and would prefer an actual recommendation from one of them than a link to everything they purchase.

Of course, that's not to say the system has no merit. Exposing your product this way has to be better than basic display (banner) advertising. How much better is the question.

Surprisingly, there was some recent research done that the FB guys and gals could look to for a bit of insight into this area. The Dynamics of Viral Marketing is a research paper that was written this year on data collected from a large online retailer of books, music and video between 2001-2003.

The data they looked at was recommendations (15million in all) made by purchasers on the site. It was actually a referral program. Which means when you bought something, if you emailed it to a friend and they ended up buying it as well, they got a 10% discount on their purchase and you got a 10% credit.

Compare that to the FB program where there is no real incentive and no real recommendation (you're not targeting an individual friend with it), all things being equal the FB program results are probably going to be worse. The problem is that 'worse' results than the referral program are probably 'no' results.

A few of the things they discovered looking at almost 15million referrals were:

  • The probability of a recommendation working decreased with repeated interaction - so we tend to ignore people who spam us with product recommendations. I am sure the same is true for multiple purchases on FB news feeds.
  • The probability of a recommendation working if we receive it from multiple sources quickly reaches a (relatively low) threshold - the verbatim conclusion from this one was "...individuals are often impervious to the recommendations of their friends, and resist buying items they do not want". Couldn't agree more.
  • Highly connected people who recommend a lot ('spreaders', 'mavens', 'sneezers') tend to have significant diminishing returns - they seem to have an influence over their immediate friends, but the wider network discounts their recommendations the more often they receive them (point 1).
  • Smaller, more tight-knit communities coalescing around a single interest were more conducive to viral effects - that's definitely not FB. That's more like EBay where small groups of collectors and such show a strong interest in certain product areas.

I think the key point for FB is that last one. Viral effects are more pronounced in 'communities of interest'. What I would like to be able to do is actually HAVE a newsfeed from my friends that I can specify exactly what I am interested in at the time.

For instance, if I need a new laptop, let me look at all laptop purchases from my friends in the past 6 months? Or a new phone? New summer clothes?

The key is I have control. I don't feel like I am being spammed. And I am looking for products that I have an interest in. Seems much more useful.

Digg this
Sphere: Related Content

Wednesday, November 28, 2007

Sweating the small stuff when conveying information

It's an absolute must to sweat the small stuff when conveying information - any information.

I just came across this from the Microsoft Performance Point site. It's a description of the different system requirements for the Performance Point product.

Try to figure out if the Planning Business Modeler and the Monitoring Server have the same hardware requirements? Impossible, right? Or at least extremely hard.

This is either multiple pages or an inventive dynamic page. It's not a long list of thin columns.

Yes it's only one page out of many, and yes, they are only system requirements. But it's the small stuff you have to sweat. This one pages speaks volumes as to how difficult the help system is probably going to be, or how difficult it will be to find help full-stop.

It's the extra 10% of effort to produce an easily readable page that counts. Without it the other 90% is wasted. Yes that sucks. But that's just how it is.

Digg this
Sphere: Related Content

Sunday, November 25, 2007

Facebook and the death of the Public Space

I was interested to read about the Beacon announcement from Facebook - basically, a way people can share what they have bought or registered for on different websites. I am just waiting until the first abuses of this start rolling in - and they will

Add this to all the ads, the branded pages, company pages, etc., and FB starts to feel like the rest of the web - crowded with branded content.

It reminds me of Naomi Klein's No Logo. She wrote that back at the turn of the century bemoaning the loss of public space to brands. Not a hell of a lot seems to have changed. It's just all moved on line.

Not that I am against branding or branded content, but now as part of the news feed in FB I have to read what all of my friends are buying? Some of my friends (no one reading this post of course) are blithering idiots and I wouldn't buy anything they recommend (I love 'em though!). I have different interests and tastes than many of my (more normal) friends as well. I'm not friends with them because we buy the same brand of shoe-shine.

Recommendations are powerful because they are personal and targeted. A friend has the same problem you have, you recommend something that helped you. Spamming seemingly random purchases in a FB news feed is not really the same thing.

It's the never-ending need to intrude on public space because of the perceived 'value' in the distracted attention of the 'crowd' that got us into this mess. There is no value in 'intrusion' anymore. Hasn't that lesson already been learned?

Naomi Klein started her book with a profound quote from David Ogilvy (that most sensible of ad men):

As a private person, I have a passion for landscape, and I have never seen one improved by a billboard. Where every prospect pleases, man is at his vilest when he erects a billboard. When I retire from Madison Avenue, I am going to start a secret society of masked vigilantes who will travel around the world on silent motor bicycles, chopping down posters at the dark of the moon. How many juries will convict us when we are caught in these acts of beneficent citizenship?

Pretty topical for something written in 1963.

Digg this
Sphere: Related Content

Saturday, November 24, 2007

Blogger Social '08 in NYC

CK just put another post up on her blog about Blogger Social '08!

The poor girl is out on the streets promoting this thing. It just took one look at the obviously 'captivated' crowd of suites to move me to post about it here. She definitely knows how to get the sympathy vote!

Seriously though, it looks like a great event. If you're a blogger, you're welcome to come. Head over to this site to check out the event, times, etc. It's NYC based. They already have a good contingent of bloggers from places far and wide signed up.

Kudos to CK and friends for putting this together. It's a lot of tough work to get one of these things off the ground.

Digg this
Sphere: Related Content

Friday, November 23, 2007

Has the PC and Apple guy campaign Jumped the Shark?

I recently saw a couple of new installments in the Apple vs PC guy campaign. Here, you have the PC guy hiring a PR person to 'smooth' over some cracks. Here, you have the poor PC guy trying to promote Vista, but things turn out bad.

Now, I'm a fan of this campaign. It has very pointed humor that cuts through to the things that are indeed irksome about Windows. It doesn't make me want to switch to Macs, but I can attest that it has had an impact on other Windows users I know.

However, there just aren't many places you an take a campaign like this. It's dependent on popular Mac/PC issues. When those run out it feels like the juxtaposition is grappling at straws. The latest installments are a good example of this.

I find Vista easy to use. It has some quirky traits, but it doesn't take a lot to get over those. It's far superior to XP. Most of the problems people have is in upgrading and third party software and drivers, not the OS itself.

Comparing Vista to the new Mac OS is fine. But bashing Vista just for the sake of bashing Vista is, well... nasty. And that's not what the Mac guy is about. If anything, his persona is overly nice. He's always trying to help.

By all means compare the security environments. Poke fun at the bloated PC 'freeware' . Laugh at the abysmal peripheral compatibility. Make fun of the nasty reviews. But when the substantive comparisons run out and the best thing you can say is 'you suck', it's time to stop.

Don't jump too far over that shark Apple, you were doing such a good job.

Digg this
Sphere: Related Content

Tuesday, November 20, 2007

Visualizations as Metaphors

There is a mini 'industry' of professionals who worry about data visualization. Which is great. Someone has to! I have a few of them in my Interesting Links on this blog - check them out if you have some time (Many Eyes, Swivel, Information Aesthetics, DSA Insights, Perceptual Edge).

Part of what they discuss and debate is efficient ways to convey insight from information - similar to some of the posts on this blog, but their concern is more with the physical display. What's the best way to arrange data in a graph? What is the best graphical representation? What types of annotations are acceptable, etc.

However, having read many of the thoughts and postings in this area, there is another way to look at visual representations of data that has less to do with efficiency of display, and more to do with impact - visualizations as metaphors.

I love metaphors. My previous boss was a metaphor giant - which can be a double edged sword when swung too liberally. But as a way to add impact to data, the visual-as-a-metaphor is invaluable.

Case in point. Here is a post on Juice Analytics regarding the evils of pie charts. The image is:


Now I love the guys at Juice, they are very sharp and smart. And their reasons as to why this is a bad chart are valid - too much visual for too little data and the 3d is kind of pointless.

They also quote some nasty comments made about the poor pie chart. One that caught my eye was from Eric G. Myers's Improving Customer Experience blog:

To my mind, the best use of a pie chart is when you have one value that is overwhelmingly larger than the rest and you don't want the audience to focus on the actual values, but just bamboozle them with the overwhelming size of the leading segment. Of course, this seems to come close to embracing the old adage, "There are lies, damn lies and statistics."

Eric, you hit the nail on the head! You just hit it with the wrong end of the hammer. This is EXACTLY the right way this type of visual should be used. You throw it up on the screen and scare the living daylights out of the audience with the large, overwhelming pie piece that dwarfs all other pieces! If it's competitive market share, then the imposing piece of pie is a looming metaphor for struggle and toil in the face of incalculable odds. If you arrange it just right, you can make your tiny sliver (of market share) even smaller and seemingly in danger (if it's all in 3d) of being eaten alive by your competition - the impending Cheesecake of doom!

Ok, so that's a little over the top, but you get the point. The image becomes a powerful call-to-arms to combat the invading competition. Or perhaps a reality check for the task ahead? In any case, it's more impactful for delivering your message than any table or set of numbers ever could be.

Just like great design is full of metaphors, great data visuals can be too. Especially if they exist to make a point or further an argument. Not all data displays are static reports delivered as cold, hard facts. Which means not all visualizations are flawed if they don't follow the rules for cold, hard, factual static reports. Something to keep in mind.

Edward Tufte calls this the greatest data visualization ever. I think I agree. But not because it crams more information into less space than anything else could. It's the fact that it LOOKS like a jagged, disorganized, unplanned and unforeseen tragedy.

Digg this
Sphere: Related Content

Monday, November 19, 2007

Video Games and Community Based Marketing

Information Aesthetics just posted some interesting data looking at certain statistics in Half Life 2 - a popular video game. The statistics make for some interesting reading if you are a video game enthusiast - play time, average completion time, heat-maps of 'maps' showing difficult and easy areas, etc.

What's really interesting though, is that you can get this information at all. The information is used by the game's designers, Valve, to understand what is and isn't working. It's used to tweak and refine further versions or updates. This is like Microsoft releasing its Office QA data. Or Apple its OS usage studies (Apple does do those, right?).

But then, video game companies, particularly ones that make PC based titles, are unsung heroes of community based marketing. Community based initiatives are passe in the video game industry. You can't release a major title without a solid community based strategy. And this isn't just about forums for complaints or questions - it's encouraging fan sites, player development of additions or modifications to the game, community days and events etc.

Even development is, in some ways, a community based task. Video games tend to have massive betas with extremely good mechanisms for player feedback. Most developers keep a close eye on player reactions when changes or additions are made.

Valve releasing data on play time etc. is just one example of this type of mentality - the more information we give out about the product, the better ideas, help and support we get back from the player community.

It's a great way to run a business.

Of course, not all the lessons are easily transferred outside of the industry. Gamers tend to be pretty fanatical. They play for the fun, the escape, the community. Users of business productivity apps are harder to engage. But not impossible.

Digg this
Sphere: Related Content

Sunday, November 18, 2007

Great post on how to demo software

Just a quick link to a great post over on Joel on Software about his trials and tribulations running a demo tour of his latest creation, Fogbugz.

His point about triangulation is powerful. You can scream at the top of your lungs to try and convince someone of something and often come up short. But when someone else whispers your argument to them, they believe.

A good point for anyone selling anything to keep in mind.

Digg this
Sphere: Related Content

Friday, November 16, 2007

Metrics wish list

Scott over at Artificial Simplicity just posted about some metrics he wishes he could measure. I just got off a phone call with a very intelligent ex colleague of mine who was also dealing with metrics he wishes he could measure.

On the call, my esteemed ex colleague brought up the Mystery and Puzzle notion so eloquently explained in a Gladwell op-ed piece for the New Yorker. A great idea to apply to anything you want to try and understand - be it marketing metrics or conundrums in general.

To quote directly from the Galdwell piece:

The national-security expert Gregory Treverton has famously made a distinction between puzzles and mysteries. Osama bin Laden’s whereabouts are a puzzle. We can’t find him because we don’t have enough information. The key to the puzzle will probably come from someone close to bin Laden, and until we can find that source bin Laden will remain at large.

The problem of what would happen in Iraq after the toppling of Saddam Hussein was, by contrast, a mystery. It wasn’t a question that had a simple, factual answer. Mysteries require judgments and the assessment of uncertainty, and the hard part is not that we have too little information but that we have too much.

The distinction is a great one. Too often we think we're trying to solve a puzzle, when in fact there is no real solution or vital missing piece. What we have is a mystery.

I would venture that the vast, vast majority of metrics/measure we use to look at consumer behavior are mysteries or parts of mysteries. Yet, in most cases, we believe them to be solutions to puzzles. Why? Because puzzle solutions are far easier to understand. You end up with a final solution, a factual outcome, a 'result'.

I think Scott asks some great questions, but they are all mysteries - and hence tough to solve.

The Brand Utility Mystery
A mystery if their ever was one - but a great thing to try and understand. Scott describes utility as more than 'usability'. It encompasses a brand's ability to understand and anticipate. Amazon being the gold-standard.

There is no one score than can encompass this though. It's just not possible (I've seen it tried). The big problem is tacit knowledge. It's too hard to divorce what a person knows from their perceptions of how easy or useful something is. Give the same utility test to top users of Amazon and Barns and Nobel and you will get the same result. You find utility in what you use often, for the very reason that you use it often.

So what would you need to do to get to this concept? Probably multiple measures that segment out new, recent and older users (to try and control for tacit knowledge). Apply different questioning techniques to each segment. Look at adoption times for new functionality across these groups. The goal would be to build up a body of insight, not a single measure. Less sexy, less useful... but it's a mystery!

Shareability/Network/Viral Effects
Another pure mystery, and Scott treats it as such. I'd love to know this one, I'd make a million dollars. An interesting question is why is this a mystery? Why is it so hard to predict what will and won't go viral? I think one of the reasons is that network effects are significantly more important than content. But you can't control network effects - who sees it, at what time, who they send it to, who they in turn send it to, what they say about it, etc, etc. There are way more great pieces of content (or conversations) that never went viral than there are networks with no content.

Participation
Again, a great mystery. What does encourage participation in a community? A host of factors have to come into play here - networks, interest, time, branding, incentives, hype, content, functions, fun, etc. The list is long. I'd love to see someone even try and model it! If you broke it down into manageable chunks - importance of just incentives for instance - you might learn something. As a broader question though, it's unknowable.

Linking Sales and Branding
This was the conversation I had with my ex colleague. Again, this delves into the realm of mystery. It also has to be the single most common question you get from Marketers - why are my sales going south but my brand metrics are strong? (or vice-versa). The truth is that there is never a clear relationship, never a stable relationship, and it's never solvable - it's not a puzzle. I worked for one of the fastest growing phone companies in the US in recent years and saw sales skyrocketing and brand metrics tanking - it made perfect sense. But it wasn't a situation that was typical and nothing you learned from it was useful to anyone other than that specific company.

Mysteries aren't 'general'. You can't derive learnings and apply to situations that look similar. I've blogged about this before, but if you really want to understand uncertainty, read The Black Swan - possibly the best book I've ever read that deals with what is and isn't knowable, and what to do about it.

Most of Marketing deals with mysteries. Beware people that package mysteries as puzzles - a common technique used by Marketing Research companies to sell you research. Beware consultants and industry pundits that package their experience as laws.

Digg this
Sphere: Related Content

Sunday, November 11, 2007

E-Filing Data

Interesting graph over on Swivel showing the proportion of people in the US using e-filing for tax returns.

Percent E-file Reaches Record Levels

Interacting with government/local government via the Web is really an idea who's time has come - and not a moment too soon.

The interesting thing about this graph is that as an adoption curve it's not that steep. You would think that network effects would have resulted in a steeper climb - as more people e-file, tell others how easy it is, they do it, tell others, etc.

I remember seeing some data on online banking back in the early 2000s and you could clearly see the rapidly increasing adoption of that service. My guess is there is probably a lot of built in inertia - accountants you have always used, not trusting some of the early services, cost.

I'd love to see data on paying parking tickets via the Web - that's a service I have been using a lot recently :(

Digg this
Sphere: Related Content

Friday, November 9, 2007

Guidelines for setting useful Marketing KPIs

I'm doing some KPI setting work for a client at the moment (I do freelance marketing/research/strategy related stuff on the side to support my bad habits) and it got me thinking about Key Performance Indicators (KPIs) in marketing.

Marketing KPIs are different. Many are 'softer' than more common revenue or profit targets. 'Softer' means they don't reflect things that have a direct and consistent relationship with revenue. For example, as a marketer you might be interested in the number of people who know about you (your Brand Awareness). Measuring this is useful because a higher number of people aware of you means your marketing is working. But what does this mean for sales? Harder to say. High Brand Awareness can lead to higher sales, but not all the time. It gets complicated.

This is why Brand Awareness is a 'soft' measures - it's useful to know, but the relationship between awareness and sales/revenue/profit is not always clear.

Marketers tend to live in this murky world of 'soft' measures for many of their efforts. And because of this, they need to pay close attention to how they measure things.

If I had to come up with a framework for Marketing KPI setting, I'd start with two broad areas:

1. Measure Selection - this is really a no-brainer. You have to at some point decided on what you want to measure. This is also hard to give general advice for because it's so situational. As a guide though, pick measures that are the closest things to direct consequences you can find. And if possible, build in feedback loops for these consequences. For instance, you don't want to set awareness targets for a viral campaign, just not going to work. Instead, measure referrals on your website with quick questions at the end of the purchase process - "Did someone recommend our product to you"?. Set baseline and measure the delta.

The general hierarchy of measure selection goes - direct consequences or immediate expected results (something was passed on, taken up); diffusion effects or secondary effects (I heard about that but didn't do anything); and tertiary effects or lingering attitudes (I thought it was good/cool/nice/interesting).

2. Measure Calibration - this is less commonly done well in my experience. The key here is to think about the absolute value of what you are measuring, how its value will reasonably be expected to change based on what you are doing, and if any change you see can be assessed as good or bad.

'Absolute' value (or 'real' value) means try and think about things in terms of concrete numbers, not percentages or in relation to other measures. If your brand awareness is 25% and the market is 100million people, 25million people are aware of you - this is important to keep in mind. Why? Because it affects the second point - how you expect this number to reasonably change. If 25million people are aware of your brand yet you propose tactics that, at most, might affect 500k people, and 25% of those are, on average, likely to be aware of you anyway, how many people can you expect to affect? Not too difficult to work out that you should be looking for less than 1% increase in awareness.

Of course for tactics that small, you're probably not going to have awareness goals - which brings us back to point 1!

In terms of assessing 'good' or 'bad' movements, this comes down to effectively using historical data and trends (although it depends on the measure as well). Markets are very fluid. All the time people are hearing about your brand, contemplating it, buying it, rejecting it, etc. This is all background noise. The general turmoil of life and commerce. You need to filter this out to see whether your tactic had an effect over and above this 'noise'. Admittedly, not always easy to do.
In fact, none of this is easy to do. It requires deep knowledge of many aspects of your market and your brand. And if not done carefully, is likely to get you in trouble as much as it can help you.

To bastardize a classic Simpson line - 'Information, the cause of, and solution to, all the world's business problems"...

Digg this
Sphere: Related Content

Monday, November 5, 2007

Some reliable Radiohead figures released by comScore

This recent press release from comScore shows the number and distribution of downloads for Radiohead's new album - the one they released online in a 'pay what you want' method.

Worldwide, 62% of people downloaded the album for free, 38% payed for it. Of those who payed for it, the average amount 'given' was about $6USD ($8 in the US). This is pretty much what I expected and puts to bed the rumors that the average price across everyone was $8. The 1.2 million was actually the number of people who visited the site during the time it was released.

The distribution of the sums given was pretty much as expected as well. Most people payed between $8-$12 with only just under a third paying more than $12. The $12-$20 price bracket isn't broken out further, but I would guess most of it sits between $12-$14.

So what does all this prove? There are some fanciful reactions to it on the comScore page. One pundit (Michael Laskow) reckons it is the death of the industry as few bands have the tenure and history of label support Radiohead does and therefore, for them, giving away music is not viable. It's actually the opposite. Giving music away is the only way future bands will create a fan base. Jim Larrison of Adify thought few people would pay and those that did would only pay a few bucks. Jim doesn't know anything about consumers.

My take is that is proves one thing only - that given worldwide publicity and a stated experimental intention by one of the biggest bands in the world, fans will come to their aid and successfully average their collective intentions to produce what is seemingly an optimal average price - $8 feels like what an album should cost if you strip away all the overhead. It does not prove that if you do this again you will get the same result and it certainly doesn't usher in the start of a new business model.

If other bands start doing this and it becomes prevalent, the average cost of 'pay as much as you like' albums will trend towards zero. Fans are making a point at the moment. They want to believe themselves that there is intrinsic value in music and they are willing to pay for it. But at the moment they have but one choice - Radiohead. As more choices appear (if this model takes off), this will spread their goodwill thin - they will either give less to each band or give most to one and take from others.

This is not a viable business model for the music industry.

Digg this
Sphere: Related Content