Chartio Gives Tools to a New Generation of Database Jockeys

9/6/12Follow @wroush

Even though data is more important than ever—whole businesses are built around collecting it, analyzing it, and repackaging it—you don’t see a lot of CEOs boasting that their companies are “data driven” anymore. That would be like saying that your car is “road based” or that your refrigerator is “cold driven.” It’s so obvious that it goes without saying.

But what’s remarkable is that so many people whose jobs depend on managing and analyzing big data still have clunky or expensive tools for doing so. Take Dreamhost, a Brea, CA-based Web hosting company that’s home to more than a million domains. Co-founder Josh Jones says that to get basic data about business performance—for example, how quickly customer support requests were being resolved—“we would always make these one-off graphs of things…We are a pretty technical company so a lot of people knew how to do this a little bit, but [the graphs] were all disorganized in different places, and it wouldn’t be clear what each graph was of. People were reinventing the wheel.”

That’s the home-brew approach, which has obvious disadvantages. At the opposite extreme are complex business intelligence platforms from companies like IBM, Microsoft, SAP, and Oracle that can cost hundreds of thousands of dollars per year and require full-time data analysts to manage.

Both types of tools are built around the idea that a business’s core databases are complex, fragile, and inaccessible, and that querying them is a challenge akin to divination. Hence the whole industry of “data warehousing,” the premise of which is that information from an operational database must be copied, cleaned, and catalogued before managers can even start asking questions about it.

Chartio demo dashboard

But there’s a startup in San Francisco called Chartio that hopes to throw this whole creaky model out the window. (Which might cause a bit of a traffic jam on the street below, given that their window overlooks a noisy eastbound on-ramp to the Bay Bridge.) Chartio founder and CEO Dave Fowler says getting actionable information from a database shouldn’t take a data warehouse or a priesthood of data scientists.

“Our vision is that there are only two interfaces that a database needs,” Fowler says (he’s second from left in the team photo above). “One is the application itself. If you are Facebook, then you see Facebook. The other is the administrative and analytics interface, where you can see how many signups you’re getting, how many sales you’re making, et cetera.”

At most Web-based companies, in other words, the public should see the public interface, employees should see the administrative interface, and that’s that. It’s a setup that will sound familiar to anyone who runs a social media or e-commerce site or publishes a blog. But until recently, getting a basic administrative dashboard really did mean either building it yourself, relying on the rudimentary tools available from sources like Google Analytics, or investing millions in an enterprise business intelligence suite like IBM’s Cognos or SAP’s Business Objects.

But those approaches don’t work anymore, Fowler argues. “Today it’s a different time. People are collecting exponentially more data, and even younger companies need to see this kind of data—companies that don’t have $10 million budgets for this.”

To suit the new times, Chartio, which emerged from the Y Combinator startup accelerator in the summer of 2010, has a different approach. For one thing, its tools are simple: even non-technical employees can use Chart.io’s drag-and-drop interface to build charts and graphs, easily specifying which values they’d like to see on each axis. (Under the hood, Chartio translates those actions into SQL queries and runs them against the specified database.)

For another, it works on operational databases directly, with no data warehouse in between. “We skip the data warehousing part, because whenever you upload your database and try to munge the data together you lose the timeliness of it—it’s always three days old, and to us that means it’s dead,” Fowler says.

Chartio is also vendor-agnostic. Unlike Cognos, which works best on an IBM DB2 database, or Oracle’s business intelligence tools, which are obviously specialized to work on Oracle databases, Chartio can draw from any MySQL or PostgreSQL database. It can also connect to Google Analytics and to databases hosted on Heroku, Rackspace, Amazon’s Relational Database Service, and other cloud services.

Fowler, a Minnesota native and former IBMer who went through Y Combinator twice (his first, failed startup was called Socialbrowse) says the big-data business has been dominated so far by companies that want to get rich filling up corporate data centers with servers and storage devices, then charge again for the databases, data warehousing, and business intelligence software needed to make it all work. “We’re saying that what people have missed out on is making a great interface that is easy for people to use.”

Dreamhost has been using Chartio’s Web-based data visualization service since it was in early beta testing. “It wasn’t really our business” to spend lots of time making graphs, Jones says. “I came across Chart.io, and it was exactly what I was thinking would be useful as a generic service.”

Dreamhost uses Chartio to track things like revenue per user, the progress of its monthly billing process, the supply of IP addresses it can hand out to new customers, and the alacrity with which tech-support staff resolve complaints. “There was one case where we found a huge spike in customers leaving,” Jones says. “We actually saw it on a graph before anyone had heard about it in tech support, and it turned out that there was a bug that was disabling customers. We looked into it before it got worse and before all these people started writing in. When you have the visuals and you are watching the data, you see these things.”

Chartio's office in the ClockTower building in San Francisco

Chartio’s basic selling point is that creating a new chart is so easy. Once you’ve connected Chartio to your company’s database, it understands which values are stored in which tables, and lets you specify which dimensions should appear on a chart’s axes by dragging and dropping from a list. Once a graph has been rendered, users can control what range of data shows up using simple filtering tools.

There’s plenty of depth for the veteran database administrator, but creating or modifying a chart or graph is so easy that beginners can learn what types of visualizations are actually useful through experimentation. “We want people to make mistakes and play around with it,” Fowler says. “That’s the way you discover things and have insights.”

It’s impossible to break anything, since Chartio isn’t copying or hosting its customer’s databases (that’s the old-school approach). For each visualization, it establishes a secure “tunnel” between a database and its own servers, pulling in live data as needed to complete specific queries. The hard work at the seven-employee company revolves around creating the visual chart-building tools, along with the code that translates their parameters into SQL queries that make sense to specific databases and hosting services.

A typical Chartio customer settles on a selection of important charts and graphs and assembles them into live dashboards for each functional team, Fowler says. “A company typically has about 17 dashboards,” he says. “One for their ops team, one for tech support, one for marketing, one for sales, and so on. So the people responsible [for each function] have the data they need and can create the questions they need to ask.”

Chartio offers tiered pricing based on the number of dashboards in use, the number of employees accessing them, and the number of databases connected to them; the service costs $85 per month on the low end and $485 per month on the high end. The startup’s private beta period ended in March, and since then its user base has been growing at 30 percent each month, Fowler says. (He wouldn’t share the raw numbers.) “I would like to grow faster, but it’s a pretty good rate,” he says.

Chartio has raised about $4.4 million in venture backing, most of it in the form of a Series A round last year led by Avalon Ventures and Bullpen Capital. Fowler says he doesn’t feel that the startup is in competition with the big business intelligence providers, but rather with engineers who might otherwise build specific queries and charts themselves, not realizing that there’s an affordable alternative that saves time and facilitates collaboration.

“Our philosophy is that if someone isn’t using Chartio, the reason is that they haven’t heard of Chartio,” Fowler says. “We feel like we’ve reached product-market fit. The huge shift now is getting the word out.”

Wade Roush is a contributing editor at Xconomy. Follow @wroush

By posting a comment, you agree to our terms and conditions.