Interview with David Benson (Diagram.ly)
This interview is the sixth in a series of interviews with people working on interesting online projects. Big thank you to David for the interview!
How would you describe Diagram.ly in under 50 words?
Diagramly is a free online diagram drawing web application that enables to you to quickly draw node/edge graphs and to distribute the result.
What made you decide to start working on Diagram.ly?
We mostly avoided implementing it because we didn’t want the overhead of setting up and maintaining the infrastructure required to provide this in a scalable and highly available way. Amazon’s web services changed that, everything is handled for us, we just improve the application.
Another thing that made us create the application was that there are a number of similar ideas out there already, but none seem to be gaining critical mass. We enjoy the challenge of evaluating such a market and working out what the business strategy is to succeed where others are failing (not with their products, failing to gain critical mass).
How did you come up with the name?
I was slightly loathe to pick diagram.ly, because the .ly suffix does date it somewhat. But then, the modern web application needs re-inventing every 3-5 anyway, we’ll find another name then.
Being a Libyan domain, credit has to go to our domain provider in Libya, they’ve worked very hard to keep the support running, despite a country Internet blackout and their US server being removed from them.
Half of the Fortune 100 Companies use your JGraph software, were you excited about creating a free product for the public?
Creating a free product should be a lot more fun than something that’s paid for, but the temptation is always to think that you can brush off problems by saying “oh well, it’s free”. We wanted to avoid this. We’ve used too many free, but useless products, there doesn’t seem much point creating another. We have a lot of experience of selling high price components, the quality requirements are strict. We’re applying exactly the same ideas to Diagramly.
The silly thing is, customers of our component, which is considerably more expensive, are generally a polite and reasonable bunch. Free users tend to be somewhat more polarized, the unhappy ones are often just plain rude. Fortunately, we’re not the sensitive types and we’ve purposefully detached ourselves from constant interaction with users for the time being.
That last sentence probably doesn’t sound like a friendly company approach, it needs explaining. I believe that the technology overview of charging users, managing accounts and providing support is a heavy overhead during the formation stages of such an application. With Diagramly, unless a diagram previously saved doesn’t load, there’s very little reason to contact us, the application should be simple enough to use alone.
What this means is that we can focus purely on adding new features. We certainly go out to users and get feedback when we want it, but when we have 6 months worth of feature requests, we hide back in our development cave and produce the next 2-3 months iteration. Then we go through a feedback stage again and repeat. Controlling the communications and not having the overhead of the pay-for elements gives us focused development time, we just create a lot more in that time.
On visiting Diagram.ly, the application starts immediately. What made you decide to skip the traditional website? Do you think this will affect Google searches?
Two reasons for skipping a website, the first is about what “Web 2.0” applications were marketed to be all those years ago. The last 10 years is littered with headlines about which online application was going to kill which desktop equivalent. In the vast majority of cases (close to all), they have failed to do so.
To us, the whole point of a web application is if you want to replace a desktop program, you need to think about the usability at every point. Starting up a desktop application just throws you straight in, the web application should be no different, and Diagram.ly startup time is designed to be at least as good as similar desktop applications. Yes, we could still put up a separate web site, but it wouldn’t be on the root of the domain and we’d still avoid logins / registration for the reasons just mentioned.
The second reason is the application functionality and user interface isn’t something new to diagramming tool users. If we were being more innovative with the user interface, yes, we’d have a problem. There’s plenty of innovation in the application coding itself, we’re not expert user experience architects, it’s not an area I’d want to be overly experimental in, yet.
There’s been a fair amount discussion as to the effect on Google searches. It’s hard to work it out exactly since the domain was registered in late Feburary 2011. The site doesn’t rank well on possible keywords for the application, but it is a very competitive area, the current results would be expected for a “standard” web site, as for our application, given the domain’s age.
A number of so called “SEO experts” told us that you just can’t do this, you must have a web site with several pages containing the right keywords and internal links between these web pages. I don’t believe this, I don’t believe that the Google engine is so stupid, that’s all it understands. It cannot be that Google would want to encourage additional unnecessary pages in this way.
So in a way, the issue of needing a web site around the application has become a bit of a personal crusade for me. I don’t want one and I want to believe that search engines are not stupidly dependent on a fixed web site format in order to function correctly.
The application has been built on Jetty and Ext JS. Is this the technology behind all JGraph products? Did you look at any other technologies? Was it an easy decision?
What’s the purpose of Diagram.ly in terms of your business (JGraph)?
It isn’t intended to generate any money, it’s a bit more of a pet project, but one we’re trying to deliver with the quality of something that does generate cash.
Although, it’s not intended to directly generate revenue, it has created two indirect uses that make its creation very useful to us, although we didn’t foresee either whilst writing it. The first is that customers of our component want the source to Diagramly to base their own products on. This gives them a much better start and time to market, as well as reducing our support load.
The second is we get very good suggestions for usability changes from Diagramly users. These changes are made to our core product and often are well received from the component customer base.
I’m glad it’s turned out that there’s a good incentive to continue its development, that doesn’t include charging Diagramly users or the use of adverts on the site. I think this is a far more sustainable model and better for the usage growth of the application.
Has Diagram.ly got the feedback and growth you expected since launch?
It has, we had to go out and actively solicit that feedback, in cases pay for threads on quality discussion sites to get it. This was expected and the feedback has been very much worth it.
The growth has been very good too, more than we expected for an application still in need of a certain amount of user interface polish.
One of the most popular things has been the number of languages we’ve translated to. It’s often overlooked in web applications, but people just love it when it comes up in their own language.
Who do you see as your target audience?
I’ve tried to avoid the “V” word, but there is a certain amount of interest from users of Microsoft’s Visio. That said, the functionality doesn’t come anywhere near to that of Visio, Smartdraw or Omnigraffle. It’s quite silly when you see headlines such as “It’s an online Visio-killer”, it’s not, the other online diagramming tools are not, and most likely never will be.
What online application creators often fails to grasp is that desktop applications are used inside companies because those companies want the application code and the data those companies process to be kept within their firewalls. So the target audience of Diagramly is always going personal and small business use. And, obviously, another target market is existing users of online diagramming tools that don’t want to pay.
Who is your biggest online competitor?
There are four big(ger) players in the online diagramming field, gliffy.com, cacoo.com, lucidchart.com and creately.com. They are all very nice solutions, though they’ve had considerably more time to evolve.
Gliffy was first to the market around 2006, so they were the only online diagramming solution until about 2009, roughly when the other three solutions appeared / came out of beta. So when you Google you’ll generally find Gliffy at the top just for the shear volume of links in the three years where they had virtually no competition. It is a shame we didn’t launch in 2006, we had the technology but didn’t go for an end user application.
I actually prefer the user interfaces of creately, lucidchart and cacoo over gliffy, they just look more modern. If I had to pick my favourite, it would be Lucidchart. I couldn’t really pick one for the user interface, the three mentioned are all very good. My objection to Creately and Cacoo is purely that they are implemented in Flash and I’ve mentioned I don’t think Flash is up to the job for creating complex applications.
Lucidchart’s main selling point is the use of HTML 5. The funny thing with this is that HTML 5 won’t even be a standard for another 3 years. We’ve been using HTML 4 / SVG / VML for 5 years, it’s a current standard and does the job just as well as HTML 5 canvas, but we lose the publicity because we’re just not using the current buzzwords. That said, if a product other than Diagramly gains critical mass, I hope it’s Lucidchart.
What is the biggest hurdle you have faced or are still facing?
I mentioned how it’s critical to ensure the maximum amount of time for focused development, but it still feels like the main problem is getting the time to improve the application. That said, when I look at the core component development we’ve done over the past month, these will all be great improvements to Diagramly.
If the resourcing issues are more virtual than real then the biggest hurdle will be getting the Google rankings whilst trying to avoid wrapping the application in a web site. This one takes a certain amount of time, we’ll probably know if the current setup is viable by summer 2011.
Do you have any other free online applications in the pipeline?
No, do one thing and do it well. At least doing one thing is easy.
What are you most excited about at the moment?
I’m most looking forward to really seeing the advantages of our low-server load, no account management architecture kick in. It’ll mean we have a lot of users, but will still be able to scale user-wise and development-wise without the legacy of paid-user momentum. I’m optimistic that you’ll see us taking a very respectable slice of the market for just these reasons, combined with all the features being free and no requirement for logging in or registration.
Can you convince the reader to start using Diagram.ly in under 50 words?
It’s free, you don’t need to log in, you own your own data, we’re constantly adding the features people say are missing and we’ll support you if there’s a problem loading or saving.
It may not be the catchiest tagline, but this is what people actually need.
Finished reading? Check out Diagram.ly!