Editor’s note: Tim Macer is managing director of London-based meaning ltd. He writes as an independent software analyst and advisor.

For respondents, the Internet has changed the face of marketing research. Years ago, “marketing research” might have meant to them a mail survey or a customer comment card. Nowadays it could just as easily be an online survey or a Web-based panel. The Internet hasn’t had quite the same kind of impact for client-side researchers, many of whom still work with paper-based tables and the occasional electronic document.

But that could change with the arrival of Pulse Train’s Pulsar Web, which offers table and charting using a highly intuitive Web browser interface. For users of Dimensions, SPSS’s suite of survey design and analysis applications, it also provides not one but three different ways to pluck data from Dimensions.

The product is an evolution of desktop-based Pulsar, which first came out six years back. The desktop version still has the edge on speed, and is simpler to set up if the online analysis is destined for an audience of one or two. It also contains a few features that have not yet made it to the Web version, but Pulse Train promises they soon will.

Desktop interface in a Web browser tool

The Pulsar Web interface looks and feels identical to its desktop counterpart. In terms of speed, there is a slight lag compared to the desktop version, but not to the extent that it becomes annoying. There is no special software to install, although disappointingly, it refuses to work at all unless you use a recent version of Internet Explorer and Windows - a matter of company policy, I am told. But it does mean that users can connect immediately from the office, or even when working from home or out on the road, to what will always be the most up-to-date version of the data.

In Pulsar Version 3, the user interface, which was starting to feel cluttered and tired, has benefited from a huge tidy-up. Pleasing, crisp lines, graphics and intelligent use of screen space make the program very accessible to even occasional users, as does comprehensive drag-and-drop support. To the left of the main window, a resizable panel contains an Explorer-style tree of folders and items, which are either the questions to choose, or the tables that have already been produced, according to which of the two views are flipped into. A larger panel, to the center and right, is where all the action takes place: defining then viewing tables and charts.

In the Variables view, you can drag and drop questions into the top or side of a table, or into the “planes” to create 3-D tables of any complexity you like. Variables are organized into folders, or “classes,” created by the programmers who import the data. It is vital that this folder structure is understood properly or you will quickly get frustrated tracking things down. Unfortunately, there are currently no text search capabilities in Pulsar to make finding things easier.

Less-advanced users can be given a simplified interface by deselecting any functionality you do not wish to have.

 Smart charts

A table can be transformed instantly into a chart, and here another improvement emerges. Pulsar 3 uses a new charting engine to produce great-looking graphs with very little effort. While many struggle to get charts right in Excel or PowerPoint, this close-coupled tool overcomes most of the mundane but time-consuming problems of using tools not designed to handle crosstabs. Present it with a table containing frequencies and percentages and other values and it automatically filters out subtotals and other clutter you have to remove manually in Excel and the like.

Almost a data portal

Tables can also be output in several formats, including Excel worksheets and HTML Web pages. In fact, the tool is starting to take on the feel of a Web portal-building tool, for alongside the reports and variables, you can bind in other non-Pulsar documents such as a questionnaire as a Word file, examples of concepts or stimulus materials as Acrobat files and so on.

It also has individualized username and password-driven access control, which you can define, as an administrator, to determine who is allowed to view each survey, and even which program options you will allow them to use. For instance, you can lock down the tools available to novices so they can only use frequencies and percentages, and open up other statistics such as significance tests for power-hungry users.

The program lacks very little in the classic crosstab arena, but anyone looking for tools to help reveal latent trends, through correlation, regression, factor or cluster analysis will be disappointed: you would do better looking at mTAB, Espri or MI Pro, or, of course, SPSS. And although you can define new variables in the program, the crosstab interface should be next in line for a makeover - it is not as intuitive as the rest of the suite. Hierarchical data, never a strong point in the Pulse Train stable, is also a bit of a struggle, and is not always feasible.

The new Quanvert?

It is cost that has fueled the research industry’s mission into cyberspace, with falling response rates adding the odd boost. The opportunities for cost savings are not always so easy to see on the results side, especially since clients still usually want to see a full set of tables. Setting them up online also effectively doubles the DP work. Clients tend to see this as a value-add, and not something they want to pay twice over for.

This was where Quanvert was ingenious, as it converted Quantum tables virtually automatically. For maybe 10 percent extra work, you got 100 percent of your tables loaded into a client-driven analysis tool. It is exactly why Pulsar Web is clever: it too has this ability to load in pre-defined reports as well as build new tables and save them.

The developers at Pulse Train have skillfully navigated through the shoals of the SPSS Dimensions Data Model to provide not one but three routes from Dimensions to Pulsar without anyone having to rewrite a single question label or tab definition.

Method one lets you run directly from the Data Model in real time, though once you exceed a few thousand records, the time the Data Model takes to deliver the data becomes so noticeable that it is better to use one of the other options. Method two lets you extract the data from the Data Model and then create an optimized database, rather like a Quantum database, which is inverted to give very fast performance.

Pulsar databases can also be built from data imported via triple-s, which opens up error-free importation from a wide range of other manufacturers’ data collection tools that support this standard, such as Confirmit, GMI and CfMC. Of course, it will also interact directly with Pulse Train’s own Bellview data.

With its sights on the Quanvert user, Pulse Train has pioneered a Quantum importer, which it has called called Leap. This somersaults over the yawning chasm in SPSS’s Data Model treatment of Quantum files, which contains questions but no tables. Leap reads the Quantum spec and the Data Model to recreate both variables and tables. Bizarrely, none of SPSS’s tools let you do this. Even more bizarrely, there is no equivalent Leap importer for Pulse Train’s own Startab tabulation package: you would be better off starting in Quantum. However, as Leap is XML-based, which makes it highly adaptable, the importer could soon incorporate other tabulation sources too.

Making the Leap

Several Leap users I spoke with considered the import was virtually perfect, pointing out that the few items that it cannot recreate, such as net lines (or subtotals) and the occasionally mangled grid question, are again due to Data Model gaps.

Germany-based research firm GfK is using Pulsar Web in Germany, the Netherlands and Spain, and has been testing it at GfK-CRI Custom Research in the U.S. “We see more of our clients are asking for the facility where they can look at their own data,” says Toni Lohmann, GfK operations director. “It makes it very convenient for them - and for us as well - to know we are all working with the same set of data. As we set up the reports for the client, we also set up access for the internal clients and sales people. With Quanvert we have to send out the database individually to our all our clients, which we do not have to do with this. We just update the data on the server.”

GfK was seeking a more modern replacement for Quanvert that would still allow it to publish sets of tables to clients. “Quanvert has not had a new interface for a long time,” says Lohmann. “It takes a while to get used to it. This is not the case with Pulsar Web, which is very intuitive to use. It is a powerful tool and it provides charting which is not available in Quanvert.”

Another attraction was that Pulsar Web could sit easily within existing business processes. GfK uses Quancept for its CATI and CAPI, and Confirmit for its Web surveys, and produces tables using Quantum. In fact, Leap was largely developed by Pulse Train at GfK’s instigation.

Research International, another global agency, is also introducing Pulsar Web to some of its customers as one of a range of analytical and delivery tools. Jaw Stoute, IT/DP specialist in the firm’s Rotterdam , Netherlands office, has worked on several Pulsar projects. Comparing the tool to Quanvert and also SPSS, he says: “So far, Pulsar seems like a happy medium. It does not do some of the very complicated things but it is good for allowing clients to go in and do some additional tables, set filters and set sub-groups within their data. The drag-and-drop functionality makes it easy to use and the new charting engine allows for a lot more flexibility for people to customize the charts.”

Stoute also had used Leap to convert Quanvert data. “We often predefine many online charts; it is fairly easy to make these in Pulsar Web. Leap is number one on my list of good things. There are some small translation limitations but overall, it works well.”

Asked if using tools such as Pulsar Web can offer economies to either research company or client, Stoute says: “I do find that putting systems in place with clients does save me time. But that is not our primary concern: we see it as something we should do as a service to our clients, to allow them to get additional insights from the data.”