Facebook sucks you in because everyone you know is using it. You go to eBay to find something because you know someone is selling what you want to buy. Oracle wins in the enterprise because there are tons of experts and plenty of auxiliary software available. All these business rely heavily on the network effect: Their product is more attractive than the competition because of their market share.
Do you know what kind of software doesn’t have the advantages of the network effect? Ours. One Highrise user doesn’t give a hoot whether we have 10,000 or 100,000 customers. Jane Doe doesn’t benefit if we sign up any other customers this year. As long as there’s a sound business behind the product, she doesn’t care about anyone else. In other words, there are no network effects.
Bug tracking has no network effects
Do you know what other kind of software isn’t affected by the network effect? Bug tracking. I don’t care who else is using Trac as long as it’s great software. It doesn’t benefit us to know that the Shopify guys are using it too (short of just sharing tips and tricks). Again, no network effects.
I don’t think Spolsky notices a difference. In Does Slow Growth Equal Slow Death, he’s freaking out that a competitor (Atlassian, it seems) is growing faster than Fog Bugz and decides he has to get into market-share mode or face extinction. That unless he puts the turbo on growth, he’s going to be WordPerfect.
Bad decisions come from fear
Fear is ugly because it makes you irrational. Fear makes you jump to conclusions. Fear makes you reactionary. Spolsky’s reaction to the imaginary threat of extinction is all fear:
1) Build every feature any customer would ever want: Apparently, by having all the features anyone can ever imagine, Fog Bugz will “eliminate any possible reason that customers might buy our competitors’ junk”. That’s a faulty conclusion and a terrible idea. Software that tries to be everything to everyone generally sucks. It becomes bloated, hard to use, and in need of big up-front training. (Actually, that’s a pretty good definition of enterprise software right there).
2) Become a sales force-driven company: Hire a bunch of sales people and make them convince people to buy our software. This is even more enterprisey thinking. Side step the actual users, the developers, and go straight to management with steak and strippers. I’ve worked at sales force-driven software companies and they suck. The sales people will invariably promise more than you have and drive you even deeper into “build everything for everyone”.
Stay strong
Companies in non-network-effects businesses don’t become extinct because they only have 50% y/y growth. They become extinct because they fuck up a good thing and become their own worst enemy. They take a successful product and ruin it trying to reach for the moon. Joel, please don’t do that.
Brad Fults
on 05 Nov 09I couldn’t have said it better myself. Well done.
Ben Bodien
on 05 Nov 09I think Spolsky is smarter about simplicity than you give him credit for:
http://www.joelonsoftware.com/items/2006/12/09.html http://www.joelonsoftware.com/items/2006/12/15.html
That said, FogBugz could really use a UI overhaul. I was a customer for a couple of years and never found anything worth writing home about on the elegance front.
Rudiger Goldberg
on 05 Nov 09Excellent article.
A rational argument against the fetishism for growth.
Brad Heintz
on 05 Nov 09I’ve used FogBugz, Trac, Redmine, JIRA, RT, Bugzilla, and a bunch of other issue-tracking products that are not worth my effort to recall just now. Joel’s/FogBugz’s problem is not Atlassian, and it’s not lack of features. It’s that I’ll go back to Trac and Redmine every time, because they have all the features I need (including source control integration, email notification, integrated wiki, total configurability, reporting, &c &c zzzzzz), they’re reasonably simple to set up (Redmine moreso than Trac, IMO), and they’re free as in beer and as in speech. There’s little if anything FogBugz can do that isn’t available in one of the free products (or a plugin).
His problem is that he’s trying to sell commodity software when excellent free alternatives exist, and is depending on suits who don’t know better to keep him afloat.
Aaron Longwell
on 05 Nov 09I don’t often agree with you, but you’re dead on the money here. Nice work.
Peter Christensen
on 05 Nov 09The difference is that there are fewer companies buying bug tracking so there’s a lot more money per customer. 37s makes $1800/yr tops off of a Basecamp customer, but Atlassian or FogBugz can make several times that. If you’re fighting for fewer, more profitable customers, you need to be able to win. It’s different markets – you wouldn’t suggest to Boeing to give an all-you-can-eat deal to any airline for a fixed price.
Fog Creek could try to shrink the industry by getting rid of the per-seat licensing, but if they have a similar price structures and features as their competitors, they’d better be able to win deals.
Peter Cooper
on 05 Nov 09Sounds like common sense – the 37signals trademark – but I’m not so sure here. Spolsky’s thinking about having a competitor that’s ten times larger has a ring of truth to it. I think bug tracking software might well be a network effect business – just in a less efficient way than your examples.
If bug tracking packages has no network effect, then, I posit, nor should developer IDEs. Yet Borland had the same issue versus Microsoft. Eventually, if enough people use the same tool, the use of that tool “spreads” as people move from company to company. If Fog Creek’s competitor ends up with ten times the users, wouldn’t the “network effect” of all those developers experienced with the competitors product lead to a larger uptake of the competitor’s product as they move from post to post?
Jay Kyburz
on 05 Nov 09Actually, not true, my previous employer is in the process of moving from fogbuz to test track because its more “corporate”. Its only a few hundred licences but I’m sure they add up.
I agree with the sentiment of your article though.
Eric
on 05 Nov 09I like the fogbugz vs 37signals ping-pong that tends to bubble up every so often. I’m sure it drives lots of traffic to both sides. I bet Joel and David are chums and enjoy sharing a beer and laughing about how much sales pick up when they get into a little blog grudge.
Joel: ball’s in your court. I’m expecting a rebuttal within 24 hours.
Andrew McKinney
on 05 Nov 09‘The sales people will invariably promise more than you have and drive you even deeper into “build everything for everyone”.’
Not just over-promise, but flat out lie. Every sales person I’ve experienced seems to fall back on this with any new potential customers whose questions they can’t answer. “Uhh, yeah, it can totally do that!”
I’ve been at the implementation-end of these selections (the solutions architect told to implement what the clueless manager chose from the sales guy). These deals invariably end up with pain for both the customer and software provider.
A while back I read a comment on this blog from a conflict resolver chastising Mr. Fried for suggesting tech managers actually learn some part of the project that they’re hiring before the process. Hogwash. Had the management I worked with properly prepared for the software they acquired, even if it was asking for the development team’s input on the matter, we would have been in a much better place as a result.
Spencer Fry
on 05 Nov 09Great post, David. Very rational analysis.
Andrew Vit
on 05 Nov 09I think there is a “network effect” but in a different way. In any B2B software, the suits will talk to each other and they’ll usually end up buying whatever everyone else is using.
If they’ve never heard of it, it doesn’t matter how good it is… They might as well choose an open source alternative if they’re making their decision solely on cost/features.
JOEL37
on 05 Nov 09Well, I use FogBugz, Trac, Redmine, JIRA , RT, OTRS, Eventum, CodeTrack, Bugzilla, Outlook, Mantis, DotProject, NetOffice, CRM-CTT/Interleave, Google Bugs, Google Docs, Facebook, MySpace, Twitter, SAP and iTunes on a daily basis for issue, bug and feature-tracking. That way, I don’t miss a single feature!
And my Dad is CEO.
Anonymous
on 05 Nov 09I think you’re wrong to say bug-tracking software is not affected by the network effect. Joel’s competitor has a significant plug-in community developing incremental improvements to the product. These improvements come faster as more people use it and build their own plug-ins, which is as close to a network effect as you’ll get. FogBugz doesn’t do the same, so perhaps Joel’s approach to the competition should be finding a way (perhaps a similar way) to introduce the network effect to his product.
Daniel Von Fange
on 05 Nov 09Considering that post is the opposite of everything Joel has publicly stood for for years, it seems as likely to be a provocative, satirical post, than an actual change of heart.
Of course, successful people sometimes to stupid things to try to get even more successful. I’ve been guilty.
David Andersen
on 05 Nov 09Charlie
on 05 Nov 09@Daniel Von Fange
I got the same thing you did. It just all seems so unSpolskylike that I’m tempted to believe inc.com’s CMS screwed up and published Joel’s April 1st article waaay early.
mctotton
on 05 Nov 09It seems like spolsky has said that they are going to use their new plug-ins to solve the customer’s desire for features. It possible to have some sales guys selling the core product with specific plug-ins.
I think you can follow this strategy and still keep the core product simple. If the warehouse needs different features then the dentist office, you don’t have to subject everyone to every feature invented.
Teddi
on 05 Nov 09Joel has real problems, here’s why
1. Atlassian is offering each of their products for just a one-time flat-fee of $10. 2. Atlassian has a huge customer base. Even major software firms like Oracle, Yahoo, Adobe us Atlassian. 3. The product without a doubt looks more professional and so does their web site.
Oh, did I mention that you can get any of Atlassian products for just $10 flat-fee each.
There will also be at some point a competitor that will come out with a better product that’s less expensive. Now, when I say less expensive, I’m NOT talk about free. There will always be a group of users that want to pay for what they use. But when you have one product that’s $10 and another product (FogBugz) that’s about 100x that cost, it makes companies rethink their purchase.
Teddi
on 05 Nov 09@37signals
Out of curiosity – what does 37signals use for bug tracking software?
Dan
on 05 Nov 09To Joel: Never whine, bitch or complain.
That being said, you need to spend more (most) of your time with getting customers, and converting visitors.
Read Jason Fried’s and Bob Parson’s (Godaddy) interviews from the same publication. (The Way I Work series). Jason explicitly says that he’s responsible with all that it’s written on the 37 website. And Bob explicitly says that his main focus is on advertising, because it’s where they see the most effect on the bottom-line. The “Build it and they will come” mentality is not quie right, unfortunatelly.
Few people know about fogbugz, and nobody knows about your most recently DVD training program of $1000+. You are nuts not to promote a $1000+ product, but only throw a link on Twitter.
David Andersen
on 05 Nov 09@Teddi -
You either can’t read or are a shill for Atlassian. $10 buys you 10 users. Prices go up from there.
Mike S
on 05 Nov 09Joel has one full-time salesperson. His article expressed a need to focus more on enterprise sales, because enterprises are very risk-averse, and are very sensitive to what they perceive as ‘vendor-risk’ in a vendor that is a distant second behind the major player. This is all absolutely true and not controversial at all.
Nathan Bowers
on 05 Nov 09Insightful comment on Joel’s post: “FogBugz sounds like a bug spray and not an enterprise software name.”
There’s limited upside in FogBugz’s market, especially when their competitors are FOSS on one side and entrenched enterprisey incumbents on the other.
FogBugz should focus their lean (and according to Joel, awesome) pool of talent on something they can win at.
Lakshmanan Anand
on 05 Nov 09I dont know why david is hating facebook right from the start … !
George Morris
on 05 Nov 09I disagree… on the network-effect.
On the surface your point sounds excellent and very original, but when you think about it HighRise does have a network effect, it’s the 37Signals brand. HighRise wouldn’t be nearly as successful as it is without the 37Signals brand behind it. You are correct in isolating the network effect of just HighRise, but you can’t fully separate the 37Signals brand from HighRise. Your success in other areas has created a network of users who love your products, thus they more easily gravitate to your latest products.
Perhaps Fog Creek should focus on better engaging it’s community of fans and customers.
I am in full agreement with you on avoiding a sales-driven company and feature bloat. That is just pure reactionary decision making.
Crimson
on 05 Nov 0937 Signals’ blog, books, and quasi-celebrity take the place of sales forces and “network” effects.
SOMETHING has to get customers coming to your product (I know you don’t think it’s just your great software. :) ). For different companies it’s different things. If 37 signals ever became interested in growing faster than their current rate, then you’d have to do something different too, simple as that.
I guess I’m in the minority here, but Joel’s post seems rather logical to me.
Crimson
on 05 Nov 09For the record – I personally prefer the 37 Signals/Joel way of doing things. I like founders personally engaging the community and sharing ideas. It feels more open and honest than salesmen shoving something down my throat.
That being said, it’s OK to admit that other approaches are effective from a business standpoint.
Nic
on 05 Nov 09@DHH A little credit to omergertel’s comment on the inc.com article would have been nice.
@Teddi I guess Trac –
roger wilco
on 05 Nov 09This post reads like bullshit. One hand there’s David saying:
On the other hand, the very first thing 37signals points out to make Basecamp more attractive than the competition is its marketshare:
Bug-tracking software might not be affected by the “network affect” but real people are going to plunk down money to use it, and people are swayed by the popular choice.
Angus Bradley
on 05 Nov 09Nice article, however Roger has a point.
@DHH how many of your sales are due to recommendations, press coverage and other network effects?
Fogbugz biggest problem is it’s name. What company wants to admit to bugs? Isnt issues nicer? Add fog – what so it’s unclear and hard to see? And then add a z just to freak out the suits. Don’t add more features, just give it a nice, safe name.
Ben Mc
on 05 Nov 09@roger wilco is right on.
If Highrise isn’t affected by the ””“network affect””” then I would suggest you back that up.
Remove your testimonials, drop the “Thousands of small businesses and entrepreneurs manage over 15,000,000 contacts with Highrise.” message, heck, re-brand it as a service provided by a completely different bootstrapped company and see if even 1 person signs up.
This ””“network affect””” is a HUGE reason Highrise is successful.
Grok2
on 05 Nov 09David, in your network effect explanation you explain the network effect of brands like Facebook and Oracle and then compare it against network effect of a kind of product—you should compare it against the brand of the product…
Also saying something like “One Highrise user doesn’t give a hoot whether we have 10,000 or 100,000 customers” also doesn’t make sense - this is true of any software including Oracle’s and Facebook’s - one user doesn’t really care about the other user.
But your overall point makes sense :-).
Jamie, Baymard Institute
on 05 Nov 09Good post David.
@To the people who say 37signals use the “network effect” by having customer statistics on their marketing sites: don’t confuse “social proof” in marketing with the “network effect” David is talking about.
World's best programmer
on 05 Nov 09Um, what’s a bug?
:-)
22 year old CEO
on 05 Nov 09What’s wrong with steak and strippers?
Daniel The Scheduling Wizard
on 05 Nov 09@Jamie I agree. Network effect for me means that I am almost forced to use it because the value is in the network, not so much in the product it self.
I have friends who told me that they will NEVER get on Facebook. But after a while, they soften up, as they keep hearing about all the stuff their friends are doing. Curiosity ftw! Network effect.. in… effect.
Stating how many users Facebook has, will do nothing for me. I don’t care if 6 billion people were using it. My friends are the reason I am there.
37s have great marketing embedded in their blog and their book. But you don’t choose 37s software because other people are using it. You choose it because the service does what it’s supposed to do.
Mike Wallace
on 05 Nov 09I think Joel Spolsky has a quota to fill at Inc Magazine, so he wrote an article about something that is on his mind that Inc readers would be interested in. Relax, people.
Johan Strandell
on 05 Nov 09I think some of the commenters misunderstand network effects. You have a network-effect business when the value of your service/product increases the bigger the network is. The usual example is the telephone – if you’re the only one with a telephone it’s useless, but if everyone is connected to the telephone network it has a lot of value.
Facebook is the same – the more friends or people that are on it, the better it becomes. So for a network effect-based business market share is important. Someone could create a better social app, but without users noone would switch.
Bug tracking isn’t like this, which is what David says in the post.
But I think David underestimates economies of scale (which is what I think Joel Spolsky is talking about). A ten times bigger company can afford to hire more marketers, more coders, more QA:s, etc, which assuming equal skill etc means they can create a better product. Or simply set the price below yours, since they make up for it in scale. So even though size in itself shouldn’t be a business goal, it’s dangerous to ignore it when you’re in the software business.
Steve Bennett
on 05 Nov 09Network effect 1: The more customers, the more people producing plugins that you can use. TracWiki is a perfect example. It sucks by itself, but some of the plugins make it almost bearable. (Ok, no they don’t. It still sucks.) Network effect 2: The more customers, the more people know how to use it. Does it matter whether your next coder is already deeply familiar with your bugtracker? Maybe. A little? Network effect 3: The more customers, the more features, and the less bugs.
Mart
on 05 Nov 09I always take into account how many people are using a software i’m looking to use. The bigger the user base, the higher the probability is that the development and community around the product will stick around for many years, especially if it’s open source. However when the user base reach a certain level of magnitude, i don’t care anymore. Meaning: as long as the software is used by many users, I don’t care if it’s 10,000 or 100,000.
Todd Chaffee
on 05 Nov 09Johan, your example was excellent and shows why David’s Oracle example doesn’t work and that he doesn’t understand the network effect either.
Bug tracking software doesn’t have the network effect. Yet market share IS important for bug tracking software, as much as it is for Oracle.
I do care if Highrise has 1 customer or 10,000 customers. I’m not going to invest my time in adopting and learning a product from a company that goes out of business. Spend more time migrating data and learning to use a product from a company that stayed in business. Multiply this by however many employees I have. Which helps explain why big companies often stick with the market leader. Even open source software doesn’t get my time investment if it doesn’t have a big enough user base.
That Joel cares about market share increases my confidence in his company’s ability to survive and increases my chances of using his software as long as I know it’s not his primary driver.
David, just because you’ve worked at a few sales-force driven companies does that mean they all suck? Can you imagine a sales team that operates ethically? Then you can create it if you’re the one setting policy, goals, and rewards.
Companies don’t just become extinct because they fuck up a good product. Companies also become extinct because they stop being brave enough to take the kind of risks it takes to start your own company in the first place. Joel, please don’t do that.
Max Kanat-Alexander
on 05 Nov 09I think that this is a really interesting point, and as one of the primary developers of a major bug-tracker, I’d agree. I think that features could be developed which would create a network effect, though.
-Max
Anonymous Coward
on 05 Nov 09“Brand” is not a network effect. Don’t confuse the two.
Anne
on 05 Nov 09I’m not a fan of FogBugz and I think there are vastly better free alternatives, but I too can see where Joel’s coming from. Theoretically, it makes no difference to me if 90% of the world are using one bug tracking system, if my preference is for another. However, in reality we are group animals. Most of us prefer to stick with the crowd and do what everyone else is doing.
Eric Anderson
on 05 Nov 09There have been a few comments to this same effect, but 37 Signals marketing prominently – and smartly – utilizes “network effects” on the homepage: “Over 3 million people use our web-based apps to get things done the simple way.” Of course, using a 37 Signals product doesn’t directly expose the network to the customer and so may not be a network effect in the strictest sense, but market share presumably affects marginal decision making (ie, it’s better to say “3 million people” than “10”).
paul
on 05 Nov 09Nothing in the article you linked to even mentioned network effects. He did discuss specific historical incidents, documented in particular books which draw a theoretical framework.
You ignored all of that and argued against a straw man.
Theo Mills
on 05 Nov 09At a recent DevDays keynote Joel equated all 37signals apps to a web page with a giant text area… I’m sure it was just a friendly jab.
But he also tried to make the argument that adding every feature your customer requests will always make your product more competitive, because “the other guy is going to add it if you don’t”. He argued that the signals’ opinionated approach to less-is-more is bad for business; you’ll lose customers by the boatload if you don’t give them everything they are clamoring for.
Over the course of the keynote, he seemed to imply that all of us who pay for Basecamp or Highrise are really just caught up in the hype and can’t possibly find the products useful because they are not feature-rich. Right Joel… 37signals has hypnotized us into “getting real”, but there’s not really any truth or substance to the simplicity they espouse. We’re all zombies just like all those deluded Mac fan-boys!
However, as a purveyor of bug tracking software Joel does seem to have a vested interest in proselytizing feature creep:
The more features you create, the more bugs you have, therefore the more you need to buy FogBugz!
Emerald City Guru
on 05 Nov 09Well, so far, the companies I worked for used either JIRA or SubVersion. My current employer is a heavy user of JIRA.
I too think a more ‘suit friendly’(tm) name for FogBugz would make it more palpable for enterprise customers.
FogCreek Issue and Scheduling Help, or FISH C’mon, a good accronym gets the suits going ;)
Jeff Judge
on 05 Nov 09I understand your point of view but don’t think it makes sense to apply the 37Signals thought process to Fogcreek’s business. In a prior post Joel said that their goal was to provide the best developer tools possible. Atlassian already has a solid foundation in this market with tools aimed towards mid to enterprise sized businesses w/Jira, Greenhopper, Confluence. They’re going to pick up even more steam (via small businesses/teams) with their $10 for 10 licenses for life. I think he knows his business best and has had plenty of time to think it through, otherwise he wouldn’t publish it in Inc.
roger wilco
on 05 Nov 0937signals doesn’t promote a “network effect” on their website. They promote their market share.
The point here is that David equates the “network effect” (I am tired of typing that “phrase” in “quotes”)—the advantage some applications have when there is a large community interacting with you, a la Facebook and eBay—with market share, which is a bunch of people using your product. If I had an auction site that had three times as many users as eBay but you could only let two bidders at a time into your auction, I’d have a site with a terrible network but a great market share.
David assumes that because FogBugz doesn’t need a network it’s not going to be affected by the large market share of its competitors. That’s naive and short-sighted.
Scott Meade
on 05 Nov 09David confused the issue by talking about ‘network effect’ when the end-game Joel fears is market share. The market share observation written about by Geoffrey Moore’s has nothing to do with network effect.
Moore says that “for pragmatist customers, the first freedom in a rapidly shifting market is order and security. That can only come from rallying around a clear market leader. Once the apparent leader-to-be emerges, pragmatists will support that company, virtually regardless of how arrogant, unresponsive, or overpriced it is.”
David and svn readers: Do you agree with Moore’s assertion?
Murphy
on 05 Nov 09“It does not matter how slowly you go so long as you do not stop.” Confucius
Rich
on 05 Nov 09I found the article interesting as a long time Joel reader because it’s so counter to his previous stances. Site licensing? Keep it simple? Out the window.
I remember the selling point used to be simplicity, as in “we don’t have custom fields because they’re a bad idea”. Sounds very much like the 37S approach.
But at some point his outlook must have changed, based on the competition obviously growing faster. Customers are buying more of the other, and he’s convinced one reason is features. Don’t really know if he’s right or not; there a bunch of other possible explanations, all interrelated.
Ahmad Alhashemi
on 06 Nov 09I don’t see how word processing software can have a network effect? Most word processing software today can open files written by the others or for the others.
I think that Joel has a valid point in that if a direct competitor is 100 times bigger than you are, then you’ll be insignificant. The networking effect works both ways. If you don’t have a networking effect in your product, your current customers will not help you pull new ones in. So you’re solely dependent on direct sales and advertisements.
However, for his argument to stand, the competitor has to maintain its profitability as it grows, otherwise it will never last the eight years required and even if it does, it will not have the money to pay for 100 times the ads, etc.
So I guess growth is important, provided you keep profits in check.
Another Developer
on 06 Nov 09I think the bug tracking scene should be more afraid of this new issue tracker: http://www.jetbrains.com/youtrack/index.html made by the authors of IntelliJ. It is very very fast and efficient compared to all others (because it’s keyboard centric) – very similar to a desktop based one.
These guys innovated the IDE scene with smart coding features (now copied by all major IDEs and tools), so I think it’s worth looking what innovations do they bring to the bug tracking scene.
Ian
on 06 Nov 09very simple: Joel is doing what he want, and he don’t care what other people says. Just like 37s.
David R. Albrecht
on 08 Nov 09Funny, I read Spolsky’s piece in Inc, and said exactly the same thing before I read this article. I don’t know whether the example he cites (RDBMS software) has as strong of network effects as, say, an operating system, but it’s clearly stronger than a web app, where the value of the platform is pretty high, even to user #1.
Ian Robinson
on 08 Nov 09his was better than yours
Mark Frankel
on 09 Nov 09Staging this as a 37s vs Joel argument is fun but often obscures the real issues.
Network effects has different nuances when it comes to Facebook and when it comes to Crossing the Chasm type marketing . Discussing those differences would be useful and I suspect that David of 37s really does understand the similarities and differences more than he portrays in his post.
Different markets require different sales methodologies. Low priced, simple, small business product like those of 37s have difficulty supporting a sales staff model. More functional, higher priced products often benefit from a sales staff to highlight aspects of that functionality to different potential customers.
It’s interesting that Joel had previously stated that you need a $75,000 plus product to support a sales staff. He seems to have re-evaluated his previous erroneous position – good for him!
Features are obviously important as is evidenced by the constant roll out of new features in 37s products. All good software designers face the simplicity vs functionality trade off and it will be interesting to see how 37s meets this challenge as they continue to improve their product with new features.
Thanks to both 37s and Joel for opening up their thought processes so we can all learn, discuss and become better software developers through the process!
This discussion is closed.