Products Don't Spread Virally | Hacker News Reader

Products Don't Spread Virally

messymatters.com

49 pointsteaspoon15 comments

How can one breathe the word "viral" and claim to debunk the concept without looking at data from one of the many Facebook apps that actually, you know, went viral? [that's a rhetorical question, I fully realize that the answer is that they don't have any truly worthwhile data to look at]

The lone Facebook app in the study is one that the authors made themselves, and it's no surprise if you've never heard of it: while I wouldn't take AppData's stats as golden, they're usually in the ballpark, and they claim Friend Sense maxed out at 317 MAU (http://www.appdata.com/apps/facebook/7890187783-friend-sense) - to give you a sense of scale, the 2000th most popular app on Facebook had over 100,000 MAU (http://www.appdata.com/leaderboard/apps?fanbase=0&metric...). Even under the assumption that it's hard to get solid data on such low traffic apps, it's easy to conclude that no, Friend Sense is not a representative example of a successful Facebook application, let alone a viral hit...

I can't give you numbers or anything, but I don't need to violate an NDA to tell you that if you're actually at a point where you're making real money off of Facebook apps, you're almost certainly seeing non-negligible virality, and you should be thinking about it, measuring it, and incorporating it into whatever business analysis you're doing. Your viral coefficient may not be greater than 1 (and if it is, God bless you and enjoy the ride, at least until you saturate!), but trying to increase it is nowhere near as futile as these authors claim - successful apps have network structures that can be very broad and very deep, and most would not have been successful at all without them.

I guess the dynamics for something that is offered for free may differ from products that are actually sold.

I think you're confusing "virality" with "popularity". Or maybe it's a God of the Gaps argument -- if we looked at sufficiently popular stuff we'd see viral propagation!

The study considered a pretty diverse set of online stuff, with, I presume, many millions of total adoptions. Essentially none of those adoptions were due to viral propagation.

That seems compelling to me.

(PS, I'm Sharad Goel's coblogger at Messy Matters though I wasn't involved in this research.)

This piece leaves me unconvinced that some products don't spread virally. It only shows that they didn't spread virally in the environments measured. I could list dozens of products and services that I know about and use only because of word of mouth (both on and off online). The problem for marketers is that identifying viral growth after the fact is easy, but trying to spark it is very difficult.

I think this sinks "viral" for now, at least as a strict analogy to epidemiology.

I can count on one hand the number of times where I have forwarded something, or had forwarded to me, something that was "viral".

Every "viral" meme I've ever seen came from a specific broadcast-dominant system. Whether it be Slashdot, Reddit, HN and so on; I've gotten 'infected' simultaneously with thousands or millions of other people. There was no peer-to-peer spread, merely a direct fanout to individuals.

Whenever my friends and I have commented on a current meme, we've been synchronised to within about 48 hours by the propagation of memes between websites, not people.

This model doesn't sound that different from measles and schools. One kid brings it, then everyone gets to stay home for a week or two.

The article makes a similar point; there are plenty of viruses, like Hep A, that propagate in a non-"viral" manner.

Hotmail and Paypal were two of the earliest and most famous viral products.

To redeem money in Paypal, you needed an account.

If you got an email from a HoTMaiL user, there was a viral auto inserted footer.

And, most important, for these two products, they didn't attempt to track your every click, plain web urls...

I fail to understand how hotmail could be viral.

To receive the message with "viral" footer, you had to already have a mailbox. If you did, why bother?

Maybe it was viral in the sense that it was the first widespread webmail and thus was in demand?

Yes, I remember paying for an email account that was POP3 based and part of my ISP monthly fee. I couldn't (easily) access it when away from my home machine because there was no web interface and most importantly - if I changed ISPs, I lost access to my old email address. It's easy for us to forget that permanent email addresses used to be prestige things.

Hotmail with its simple tag line - get hotmail, free web based email was a pioneer.

For me, it was my very first personal cloud based service. At a time when those who had email accounts had them through thick clients like Eudora. It was a big deal. I still retain my hotmail account for sentimental reasons.

For every book or album purchased because of a personal recommendation, however, how many were bought after simply browsing the stacks, reading a review, or seeing an advertisement?

This was never how I conceptualized the idea of "viral". I basically viewed it as "taking on a life of it's own" such that it took little initial effort to get a big effect. Yes, people spreading it is a factor, but if it weren't inherently interesting in some way, it wouldn't be "contagious".

Edit: Besides, supposedly the first 50 cases of AIDS in the US could all be traced back to one person. Even actual virii don't follow the path this piece seems to artificially be trying to impose on the term

How do they know if they got every instance of person-to-person spread? E-mail, IM (3 different ones), Google+, personal communication - that's how I get the jumping cats of the internet. Unless you monitor all of that, you'll get an anti-spread bias.

The only way to be sure of the actual spread graph is to monitor an invite-only service, which of course might not be 100% representative of other kinds of services/products.

1. The 'kill marketers' puzzle was not funny.

2. There is no link to the paper it is talking about.

3. The blog post has many holes in the reasoning.

By the way I really liked the 'spreadable media' idea by Henry Jenkins and his arguments against using the word 'viral' (http://www.henryjenkins.org/2009/02/if_it_doesnt_spread_its_...).

1. I found it funny. :(

While product adoption may not be viral, perhaps product marketing is. One small blog picks up a story, which may in turn get picked up by a bigger blog, which then gets picked up by a major media outlet. The adoption profile still looks viral, even though the adopters are not communicating their preferences directly to each other through the social network.

Reply on news.ycombinator.com