Compete, as you may know, provides analytics tools, and shows website traffic data on things like unique visitors, pageviews, visits, etc. The data is sometimes compared to Google analytics, and often tells a different story.

We reached out to Compete to discuss this a bit, and Damian Roskill, Managing Director, Marketing at Compete tells us, “It’s really like comparing apples and oranges – both methodologies have their strengths and their weaknesses.  Google Analytics doesn’t actually track people – it tracks cookies.  This means that when people change computers, or delete their cookies, you can have double counting that goes on.”

Compete’s methodology has four basic components. They are (as described in a Compete white paper):

1. Multi-source panel – combines both recruited panelists and licensed clickstream data from partners, resulting in a very large and representative online consumer panel;
2. Harmonization algorithms – proprietary processes and technology that work together to integrate Compete’s multiple data sources into a unified online
   consumer panel;
3. Sophisticated normalization techniques – unique processes that weight, calibrate and project metrics across panel sources resulting in accurate base audience measurement data;
4. Metrics that matter – analytics on site visitation, audience profiles, media exposure, search term usage, cross-shopping, conversion, competitive behaviors, and audience segmentation

“In contrast, a panel is a sample-based approach – we track about 2 million US consumers – a percentage of the total internet audience here in the US – and then we use computer modeling to project behavior from our panel to the general internet audience,” says Roskill. “The upside is that a person is a person – you don’t have the double-counting.  The downside is that you can sometimes get undercounting of smaller websites.”

“Also notice that we are currently only doing US consumers – so people often send us a Google Analytics number that includes all international users as well as US consumers – the two aren’t comparable in that way either.”

Certainly an important point to note.
 
“Bottom line, the approaches are complementary rather than either/or – what should make sense is that the two methodologies should be similar directionally.  But comparing the numbers doesn’t really provide much value.”

In my experience, they generally are similar directionally in most cases.
 
“What we think can be meaningful is combining the two approaches – essentially using a cookie-based approach and combine it with a panel based approach and use them to provide validation for each other,” says Roskill.

The fact is that you can compare data from any number of analytics tools, and you’re going to get different results. It’s when they tell dramatically different stories that you need to investigate further. Luckily there are resources explaining how data is collected. Compete’s white paper, for example explains its methodology. Google Analytics has an entire help center, and there are plenty of articles on the web discussing it.