A while ago we made public our intention to build an insights generator for Google Analytics, Omniconvert Insights.
The response has been overwhelming so far: >4000 websites signed up for early access, giving us read-only access to their data from GA to generate data insights.
As we’ve been using Similar Web to analyze the traffic for various websites, we decided to dive a bit in this huge amount of data to compare it with real data from GA and understand how accurate is Similar Web in reporting data for websites of various sizes.
That’s why, we’ve gathered data since September in terms of:
- Number of sessions
- Bounce rate
- Time on site
But, is this data reliable? Can we apply this to all the websites?
For sure, no!
That’s why we decided to segment the data, like we do in GA for visitors. Only, this time we felt like in a fractal. The websites are now the unit, not the visitors.
Side note: thanks a million, Andrei & Gero for doing this hard and repetitive work to make this data possible:
- The more sessions a website has, the more accurate the data in Similar Web is. Therefore, you can rely on the data for websites that have >100k sessions in GA;
- The time on site is the most reliable parameter from Similar Web;
- Doubling the number of sessions that you find out from SW will not give you accurate results for most of the segments. Except for the 10-100k websites, which represent a third of them;
- Similar Web (as well as this analysis) can give you good indicative numbers;
- Less is not always more.
P.S.: Omniconvert Insights is to GA what Waze is for traffic optimization. Free early adopters spots are available until tomorrow.