Bug tracking isn't a network-effect business | Hacker News Reader

Bug tracking isn't a network-effect business

37signals.com

85 pointsaditya50 comments

It doesn't have to be a network-effect business for customers to prefer the popular product as a "safe" choice.

This prompts the question, if you're entering a field that has a 'safe' choice, how do you position to compete?

I'm sure there were other bug-tracking systems available when FogBugz was being developed, yet FogBugz has stuck around and is perceived as being a 'safe' product now -- perhaps because of Fog Creek's longevity?

1) Create something significantly better than the safe choice has.

2) Convince a few well respected companies to take a chance on you.

3) Use that social proof to get everyone else to follow their lead.

4) Profit.

Just to add a little to your excellent advice:

#1 could be an intangible improvement over existing offerings, such as better customer service, better (or no) contract terms and/or pricing, etc.

Something that's been talked about here before is offering localized versions of popular services, such as the dating sites tailored to the local culture of Japan (or was it China?) due to the large difference in the way that dating work over there as apposed to western countries. This could make #2 much easier to accomplish, if your offering is a better match for a specific subset of possible customers, but I guess that's the definition of finding a niche market, huh?

Marketing will say it's a brand position problem. Development will say it needs features done this way. Sales will say you need costumer X. Lawyers will say you need patents. Every role has an opinion, because they think the solution lies on what they know how to do best.

And you know what, they're all right. The problem is how to find out who's more right than the others so you can allocate resources properly and coordinate the changes.

I think your statement might be self-contradictory. If having more users (by itself) can bias a new user, then the product being sold has a network-effect.

EDIT: I think the term I (and we) are looking for is positive feedback loop. That's ultimately why network effects cause a race to grow the fastest.

Network effects tend to be substantially overestimated. For instance Metcalfe's law is widely cited, but more careful analysis of actual behavior suggests that n*log(n) tends to be a better scaling estimate than n^2.

Moving on, be careful about moving to a sales force driven strategy. When you're selling with a sales force you have to cover the overhead of an expensive sales process. This is one of the big reasons why a lot of software is priced low enough that someone can pay for it with a credit card, and lots of software priced over $50,000 but very little between those two numbers. Also once you go down the enterprise software route the quality of your sales process matters more than the quality of your software. A good read on that is http://lists.canonical.org/pipermail/kragen-tol/2005-April/0....

I'm not saying that this is a bad strategic move or the wrong strategy for your company. But you should be aware of the consequences.

I did an internship at a largely sales-driven company (Everyone gets an email when someone closes a big sale), and it suuucked. Especially when everyone was really anxious and uneasy because Customer X wanted his bug fixes Right Now.

Unfortunately, you attract top sales talent with big money, and they get lethargic if the big money doesn't come from commissions, and they tend to over-promise to make the sale. This can piss off the developers and makes it not "a place developers want to work". You definitely have to maintain a balance.

The idea that comes immediately to mind is profit sharing. When Joe closes the big deal, everyone gets commission. Even if it's not a significant amount, it'll let your developers continue to feel valued.

At least that'd make me happier (Edit: Assuming I were in that situation).

Thanks for the link, that was a good read! Reminds me of my days at a mid sized regional bank. The amount of money they spent on implementing "Portals" was outrageous, and included licensing, training, consultants, and opportunity costs (which were huge if you consider what they could have done with all of the developers, testers, and sys admins that were otherwise tied up in the project). I don't think they've implemented Portals yet.

A social proof business?

Every business is at least some network effect, in case of bug tracking,

We convince every new client we sign up to use Unfuddle unless they have a very strong preference for any other. I am sure I must have sold >5 new clients to Unfuddle this way, and those people sold Unfuddle to a few more contractors who to few more clients who to ...

'Network effect' has a distinct meaning above and beyond 'positive word-of-mouth'.

Sure. But how "

    I recommened to 5 people, who recommended to

    5*(avg no of ppl influenced) * (decay factor people), who to

    5*[(avg no of ppl influenced) * (decay factor people)]**2

   ...

" not a network effect as long as [(avg no of ppl influenced) * (decay factor people)] > 1?

Now you're conflating 'network effect' with things like 'viral loops' or population dynamics.

Again, they're distinct. 'Network effect' is the term to use when you're emphasizing that each new adopter increases the value of the installed base. A 'decay' factor doesn't usually come into such analysis.

To say that any product with enthusiastic users has a 'network effect', because the users tend to influence other potential customers, robs 'network effect' of its specific meaning.

In business, up to a limit, money has a network effect. More money, especially more revenue, lets you move faster and take more risks. Only true up to a limit, okay, but disposable income is agility.

I think I'd actually say it has a "reverse network effect" (oh, christ, i hate talking like that).

Seriously, though, allow me to illustrate:

If the value of $1MM to you is N. And you add a second $1MM. The value is almost certainly going to be less than 2N.

We're talking about the utility function of money. With enough money, it's definitely the case that it goes down, but I think it goes up faster than the apparent value for a bit as well. £20k is more than twice as useful as £10k is more than twice as useful as £5k. Where the crossover is, though...

Money can let you take more risks, market better, do things faster and larger -- but none of those capabilities are 'network effects', any more than a 'mountain lion' is a 'wedding ring'.

Geez, people, is there some sort of contagious aphasia going around? Somebody call the CDC.

I agree. And oddly enough, 37Signals does, too, even they don't seem to know it. :P

Go to http://highrisehq.com/ , scroll down a bit until you see the second large green button, there you can see the text: "Thousands of small businesses and entrepreneurs manage over 15,000,000 contacts with Highrise."

If it doesn't matter how many people use Highrise, why that text then?

Well, because this fact is indeed important, as the blog says "As long as there’s a sound business behind the product, she doesn’t care about anyone else." That's of course correct but a lot of people use the amount of other people using the product as a way to determine how sound of a business there is.

I'm always amazed at 37signals' ability to snipe other people in their industry, and find it more than a little repellent.

Also, of COURSE products have a network effect! Haystack attempts to be a marketplace for design work; if they cannot draw traffic to it then it fails.

Edit: changed Highrise to Haystack

Think you mean Haystack? Highrise is the CRM...

I don't think 37s lives or dies on Haystack. It's not even listed on their front page as one of their products. You picked a poor example.

I didn't say that. I said the product dies if it cannot get traffic.

And you're right. Haystack is, in fact, a network-effect business. But Haystack is an insignificant part of 37signals business, and trying to characterize them by the way Haystack works is not a lot different then trying to characterize them by how the 37signals Job Board works.

You know what? I read and wrote in a hurry, and my point is nonsensical and irrelevant. Forgive me.

Just because their product doesn't have a strong effect, it doesn't mean their customer base doesn't.

If ten times the number of people are raving over their competitor eventually, that'll have a big effect against them (or, at least, put them on the back foot).

There's an enormous network effect with bug tracking software. If you learned a software as a junior worker at one company, you're likely to choose the same one when you're in a more senior position at the next one.

Developers move from company to company, and take their knowledge and preferences with them. When you're setting up a new company or department, you're going to pick the program that your team knows.

My employer bought an instance of a certain bug tracker because one of our vendors was using it and we needed to keep their data when we changed vendors. A year later, the whole division uses it.

Actually I would argue that 37signals thrives on the network effect. A network of anti-enterprisey, web-savvy, tech trendy folks but still the network effect. I think maybe joel was wondering is if they were in danger of falling behind in their particular network.

This reminds me of something I heard about Motorola, and their decline after the Razor phone peaked. They learned that the phone business is more like the fashion business than the electronics business.

With software related to the dev process, look how fast Subversion went from cool to uncool. It took just one viral Linus video.

I think there IS a bit of a network effect and Joel is correct to have concerns.

He says he doesn't have to know who used Trac in order to use it. Well, I disagree, and I'm with Joel on this one. Whether or not a project has users is one of the things I look for when picking software. More users often (but not always) means more developers, and some sort of guarantee that the product won't be abandoned.

This isn't as critical with open-source software (I can install a bunch of bug-tracking tools, use them for a few days, and keep one I like), but if I have to pay to try these products, the popularity factor gets more important. Even if there's a no-risk trial (still have bad memories of trying to cancel that AOL subscription way back in the day).

The 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.

My feeling is that Joel is really just getting tired of his comfortable, but small, operation. If he keeps doing what he's been doing he'll probably be relatively small forever.

37 signals is in a similar situation. They're really small relative to what's possible. As far as I can tell they're on course to be small forever, in terms of revenue.

You already have a good product. You know there's a market. You know people are willing to buy it. Now what do you do if you decide you want to accelerate your growth a lot?

I doubt you have any idea how much 37s (or Fog Creek, for that matter) makes in top line revenue.

You're implying there's a chance they make enough to put them outside the realm of "relatively small". Are you suggesting either of them make many 10s or 100s of millions per year?

37s made 1.5MM on a job board, and 6 digits on an ebook ("getting real"). Would I be surprised if they had an 8-digit top line? No.

What's "relatively small"? Plenty of 100-person VC-funded companies never get that kind of revenue.

So, that's where we differ. I consider low 8 digit revenue to be "relatively small".

I made it a point not to base size on headcount, since I agree that's meaningless. I'm talking strictly about revenue.

Low 10's of millions ain't bad when you've only got a handful of people working for you.

I don't have a reference readily available, but I believe I heard/read recently that their revenues are in the $xx million/year (maybe $15M?).

Sounds a lot like what the Ingres guys must have said. "It doesn't matter if Oracle has a bigger network, there'll always be a market for us."

Of course there's a network effect. Oracle gets market share, third parties build stuff that relies on Oracle, Oracle gets more market share. Your bug tracker gets market share, people build source control and build system integration around it, it gets more market share.

Swapping out bug trackers is both theoretically and practically as easy as swapping out databases.

Especially if a third-party or a competitor provides a relatively easy conversion tool to take your data from one product to another

Everything which has a hinterland and ecosystem around it benefits from network effects. If you're going to throw up walls at the web-browser edge, sure, but...

Highrise, in particular, benefits from form processors, iPhone apps, analytics tools all outside of highrisehq.com. Nothing exists in isolation - everything in software services is a network-effects business. You can argue over how strong network effects are, but they're always present.

The main point I heard in Joel's article was not his desire for network effects, it was his proposed move into an enterprise model.

In the comments here for the Inc. article he said something about being good at marketing but terrible at sales. Being reliant on one or the other of these seems like a fundamental quality for a company.

To be fair, I don't think Joel was going for "become sales-force driven", growing a salesforce is not the same as letting them run the show. As long as they have great engineering which Fogcreek seems to have.

Unfortunately, it is hard to say if these fears are unfounded or not.

An application with lots of users seems likely to get better faster than an application with few users, if only because a larger user base means a) bugs will turn up faster, and b) patterns of use will suggest feature improvements faster.

Basecamp's recent switch to global logins is evidence that their products benefit from the network effect.

Sorry David. Selling bug tracking software can be a network-effect business if your product is a platform: http://confluence.atlassian.com/display/JIRA/JIRA+Plugin+Dev...

[edit: http://confluence.atlassian.com/display/JIRAEXT/JIRA+Plugins]

An application platform does not make a business network-effect driven. Having plugins doesn't make your product create entertaining, compelling, or useful content to "suck in" non-users. Facebook and eBay have this because it's baked in to their DNA.

Basecamp either

That's one of his points. At the risk of sounding non-HN-esque, rtfa.

You can say what you want about DHH, but he's 100% correct here.

Reply on news.ycombinator.com