Review: Birst brings DIY to BI

We’re all aware knowledge is power and time is money. But for many of us — particularly small businesses and departments without dedicated IT support — the leap from static reports of yesterday’s news to deep, analysis-driven insight might as well be a leap across the Grand Canyon.

Birst helps fill that analytics gap with a cloud-based BI suite, Birst Enterprise Edition. With Birst you really can create data warehouses in the cloud in minutes, slice your data into multidimensional drill paths, and display actionable reports and browser-based KPI dashboards with ease.

[ Download InfoWorld’s Big Data Analytics Deep Dive for a comprehensive, practical overview of this hot topic. | Cut to the key news for technology development and IT management with our once-a-day summary of the top tech happenings. Subscribe to the InfoWorld Daily newsletter. ]

The simplicity of Birst stands in contrast to competitors such as GoodData. Although GoodData’s developer-centric IDE delivers powerful features for data massaging and transformation capabilities, the power comes at the expense of complexity.

Birst provides an automated toolset that lets ordinary business people do what developers or data analysts have long labored over using traditional BI tools. Although a complicated knitting of multiple data sources will still require traditional data wrangling expertise, Birst’s wizardry is more than sufficient for a broad swath of use cases.

On the flip side, Birst offers none of the advanced formula syntax you’ll find in SiSense or the interactive charting you get in QlikView, and it’s light on predictive analytics and forecasting. At times, Birst’s Flash-based interface can feel sluggish.

But Birst flat out excels at tapping data from a broad range of sources — relational databases, Hadoop/Hive, SAP, Excel files, and cloud-based apps, including Salesforce, NetSuite, Google Analytics, and Marketo — and at guiding business users to swiftly unlock the wealth of information in that data with minimal expertise.

Loading and modeling Birst uses a Spaces metaphor to compartmentalize each deployment, allowing separate lines of business in a company to access and analyze shared data sets. Each Space provides a repository for your data warehouse, report and dashboard settings, permissioned user access rights, and the like. On first run, a navigation page divides development and administration tasks into four key phases: loading, modeling, processing data, and user access.

Navigation is intuitive enough, but it would be better served with a wizard-driven guide, or perhaps displayed in a side panel, so users retain quick access to links.

Loading data into the warehouse is easy using any of the available options. Flat files can be uploaded through the browser, while a Java app, called Birst Connect, allows uploading from the desktop. And Birst’s cloud-based Extractors tap into cloud sources such as Salesforce and Google Analytics.

The Connect tool can be run as a Windows service, so the desktop app isn’t constantly running, and both Connect and Extractors support scheduling.

Tested against a Salesforce instance, Extractor’s fine-grained controls let me pull in individual tables or entire databases. For each of the import methods, I found uploads to be fast and easy to manage.

The next step depends upon the cost of your subscription and the configuration mode of your Space: automatic, discovery, or advanced. Automatic mode will scan your data set to determine the primary keys and relationships, generate best-guess models, and populate a Space with prefab analytics. Discovery mode adds dashboards and visualizations, while advanced mode adds modeling and advanced reporting, giving you complete access to the entire Birst toolset. To get the best of both worlds, you can start in auto mode, let Birst do the heavy lifting, then upgrade your Space to advanced mode for fine-tuning.

Birst does a nice job of mapping relationships in well-defined, well-structured data sources. However, the UI would benefit from better organizational tools. Larger maps become unwieldy to traverse, and though you can override the mappings in the same interface, it’s clumsy. My manual attempts to join related sources were sometimes hindered by Birst’s hiding of data objects it considered irrelevant. (This could result from a simple misspelling in matching column names, for example.) If you prefer, you can load the data using an entirely do-it-yourself approach and map the relationships manually.

Once your data is loaded and mapped, it’s easy to set measures (like revenues or quantities) and dimensions (like time periods), and Birst makes quick work of structuring hierarchical navigation paths and grains (depths) available to the reporting engine. For example, using my loaded sales and product databases, it was a simple matter of a few clicks to drill from global sales figures down to a specific salesperson or part number by specifying the requisite paths and grains to include.

Reports and dashboards Birst packs in all the tools you need to make effective use of data: custom charting, scheduled report delivery, and browser-based dashboards. A newly added visualization interface, although still green, further simplifies chart creation through easy, guided selection of measures, columns, filters, and chart types, allowing users to interactively update the visual display in real time.

The report Designer tab maintains a list of data attributes and measures along with pre-baked, business-focused time-series measures (such as “trailing 12 months” and “quarter to date”). As a result, creating a chart or quick report is literally as easy as a click or drag of measures and attributes onto the canvas and specifying a chart type.

Birst abstracts away the underlying SQL without sacrificing the ability to run powerful SQL queries. For instance, Birst provides its own Birst Query Language for trend analysis (even if it’s only basic linear regression). Birst Query Language is SQL with added time-series functions and the ability to manipulate dimensions and measures of aggregated data.

Birst offers a good number of charting options, from bubble and funnel charts to gauges and maps. Maps, in particular, were easy to define — using either Birst’s generic maps or Microsoft Bing — and geo markers configured for underlying data display.

A search tool would be a useful addition to Designer, considering larger data sets can become unwieldy in Birst’s folder-based layout. More flexible and finer-grained layout controls would also be welcome.

For instance, Designer has a secondary Layout Mode that’s better suited than the Designer to text-based row reporting, where summation and Group By clauses can be put to work along with conditional display and custom formatting. But once you enter Layout Mode, you can’t go back to the Designer to reformat your layouts.

That said, the layout controls work perfectly fine and the underlying SQL grouping and joins are mercifully hidden from the user. Ultimately Birst produces nicely tailored, good-looking reports with minimal fuss.

Birst’s browser dashboards perform well and provide built-in prompts to update data on-the-fly. You simply click or lasso groups of data points to drill through data. All of the charts on the page autosync for a seamless presentation. Currently, users can choose a Flash or HTML5 interface, with only minor inconsistencies between the two, but Birst claims it has plans to migrate away from Flash in the future.

On the downside, dashboards can’t be created ad hoc. You must first build reports from the Designer tab, save them, then use these as widgets in a dashboard layout. Some minor layout irregularities, sizing issues mostly, required a few round trips to the report Designer for manual adjustment.

I would like to see better write-back tools for what-if forecasting, such as those available in QlikView, where easy slider manipulation adjusts graphics in real time with updated perspectives. Birst’s graphics are essentially static.

Finally, Birst reports are easily exported out of the browser — either manually or via scheduled email delivery with attachments for PowerPoint, PDF, and Excel in tow. The chance to create a quick distribution list would have been nice, but email addresses can be entered manually, or a secondary report can be structured to manage distribution.

Data in, insights out One of the challenges of the SaaS model — for both providers and customers — is managing the inevitable changes to the application. Birst rolls out updates regularly, and although you can control which engine is available to your users, you are limited to (roughly) two back-dot revs at any given time. As Birst retires older revs, users are automatically updated and pushed new features, some beta, whether they like it or not. Many IT shops would prefer tighter control over updates.

It should also be noted that Birst sells a virtual machine appliance for on-premise deployment and supports an API, but declined to provide either for this review.

Birst clearly has room to grow, but it’s just as clearly moving in the right direction. 

With support for advanced features like slowly changing dimensions (to preserve reporting accuracy in the face of customer name changes, reassigned sales territories, and other alterations to the data) and “live access” connections (to remotely query on-premises OLAP data sources and XMLA cubes for additional analysis), Birst goes a long way toward meeting the needs of analytics experts, while also helping to sever BI from its IT-centric past and place it directly in the hands of business users.

This article, “Review: Birst brings DIY to BI,” was originally published at InfoWorld.com. Follow the latest developments in business intelligence, cloud computing, and big data at InfoWorld.com. For the latest business technology news, follow InfoWorld.com on Twitter.