For as something as big and audacious as 'understanding web traffic' any technology company has to start somewhere – the panel approach has advantages for macro trends, strategy and big budgets and direct measurement has advantages for targeted, tactical decisions. And the funny thing is, once a technology company picks a strategy they're pretty much stuck with improving it. Branding and differentiating paints you onto a path – panels get bigger, but its still a panel; site coverage gets broader, but its still sites. The essence of a technology company can often be found in what problem the founders first solved.

For panel data companies consistency is more important than truth. I need to be assured that the data I'm looking at is derived the same way over time, segments and properties. I'm much more interested in a10% swing over time than specific counts because decisions are often made on relative changes from a benchmark rather than an absolute. For direct measurement, precision is more important than consistency because the unit of thinking is different, e.g site paths and conversion.

As to business models, seems to me that the likes of Compete, ComScore, Hitwise, Nielsen locked up the 'representative panel' approach while sacrificing specifics. For example, I tried to link TV spot logs to search traffic and got told by one of them that we don't measure that level of detail. Quantcast (and I've never spoken with them) might have chosen to focus on the opportunity around understanding specific sites. If I were pitching this idea to investors; I'd create a new category and damn the syndicated services with faint praise. “Oh, if you want to understand the 'music category' check out those guys, but if you need to identify specific sites for true relationship marketing and outreach then we're the only game in town. To do this we're going to recruit publishers, not people.'

Someday, somewhere maybe the two will meet. But until then – different strokes for different folks. Use them all!

For as something as big and audacious as 'understanding web traffic' any technology company has to start somewhere – the panel approach has advantages for macro trends, strategy and big budgets and direct measurement has advantages for targeted, tactical decisions. And the funny thing is, once a technology company picks a strategy they're pretty much stuck with improving it. Branding and differentiating paints you onto a path – panels get bigger, but its still a panel; site coverage gets broader, but its still sites. The essence of a technology company can often be found in what problem the founders first solved.

For panel data companies consistency is more important than truth. I need to be assured that the data I'm looking at is derived the same way over time, segments and properties. I'm much more interested in a10% swing over time than specific counts because decisions are often made on relative changes from a benchmark rather than an absolute. For direct measurement, precision is more important than consistency because the unit of thinking is different, e.g site paths and conversion.

As to business models, seems to me that the likes of Compete, ComScore, Hitwise, Nielsen locked up the 'representative panel' approach while sacrificing specifics. For example, I tried to link TV spot logs to search traffic and got told by one of them that we don't measure that level of detail. Quantcast (and I've never spoken with them) might have chosen to focus on the opportunity around understanding specific sites. If I were pitching this idea to investors; I'd create a new category and damn the syndicated services with faint praise. “Oh, if you want to understand the 'music category' check out those guys, but if you need to identify specific sites for true relationship marketing and outreach then we're the only game in town. To do this we're going to recruit publishers, not people.'

Someday, somewhere maybe the two will meet. But until then – different strokes for different folks. Use them all!

For as something as big and audacious as 'understanding web traffic' any technology company has to start somewhere – the panel approach has advantages for macro trends, strategy and big budgets and direct measurement has advantages for targeted, tactical decisions. And the funny thing is, once a technology company picks a strategy they're pretty much stuck with improving it. Branding and differentiating paints you onto a path – panels get bigger, but its still a panel; site coverage gets broader, but its still sites. The essence of a technology company can often be found in what problem the founders first solved.

For panel data companies consistency is more important than truth. I need to be assured that the data I'm looking at is derived the same way over time, segments and properties. I'm much more interested in a10% swing over time than specific counts because decisions are often made on relative changes from a benchmark rather than an absolute. For direct measurement, precision is more important than consistency because the unit of thinking is different, e.g site paths and conversion.

As to business models, seems to me that the likes of Compete, ComScore, Hitwise, Nielsen locked up the 'representative panel' approach while sacrificing specifics. For example, I tried to link TV spot logs to search traffic and got told by one of them that we don't measure that level of detail. Quantcast (and I've never spoken with them) might have chosen to focus on the opportunity around understanding specific sites. If I were pitching this idea to investors; I'd create a new category and damn the syndicated services with faint praise. “Oh, if you want to understand the 'music category' check out those guys, but if you need to identify specific sites for true relationship marketing and outreach then we're the only game in town. To do this we're going to recruit publishers, not people.'

Someday, somewhere maybe the two will meet. But until then – different strokes for different folks. Use them all!

For as something as big and audacious as 'understanding web traffic' any technology company has to start somewhere – the panel approach has advantages for macro trends, strategy and big budgets and direct measurement has advantages for targeted, tactical decisions. And the funny thing is, once a technology company picks a strategy they're pretty much stuck with improving it. Branding and differentiating paints you onto a path – panels get bigger, but its still a panel; site coverage gets broader, but its still sites. The essence of a technology company can often be found in what problem the founders first solved.

For panel data companies consistency is more important than truth. I need to be assured that the data I'm looking at is derived the same way over time, segments and properties. I'm much more interested in a10% swing over time than specific counts because decisions are often made on relative changes from a benchmark rather than an absolute. For direct measurement, precision is more important than consistency because the unit of thinking is different, e.g site paths and conversion.

As to business models, seems to me that the likes of Compete, ComScore, Hitwise, Nielsen locked up the 'representative panel' approach while sacrificing specifics. For example, I tried to link TV spot logs to search traffic and got told by one of them that we don't measure that level of detail. Quantcast (and I've never spoken with them) might have chosen to focus on the opportunity around understanding specific sites. If I were pitching this idea to investors; I'd create a new category and damn the syndicated services with faint praise. “Oh, if you want to understand the 'music category' check out those guys, but if you need to identify specific sites for true relationship marketing and outreach then we're the only game in town. To do this we're going to recruit publishers, not people.'

Someday, somewhere maybe the two will meet. But until then – different strokes for different folks. Use them all!