Tag error: <txp:tru_tags_if_has_tags> ->  Textpattern Warning: tag does not exist while_parsing_page_form: archive, default
Tag error: <txp:image id=104 /> ->  Textpattern Warning: Attribute values must be quoted while_parsing_page_form: archive, default
/commentable: Testing Feedjit "Also Liked" Spider
Dinarius = digital interest
31 October 2008

Testing Feedjit "Also Liked" Spider

For two months, there have been very large, hard-to-miss boxes at the tops of all permlink pages meaning you see them on the page of an article, but not on any of the index pages. They are simply headed: Visitors to this page also like:… What follows are benefits of the widget by Feedjit.

Feedjit offers several toys for Webmasters and they’re all basic. They point to Feedjit knowing a significant amount about surfers and Internet traffic as a whole based on the adoption of their widgets from webmasters. Every visitor is tracked throughout the site and their origin is noted as well.

From 17JULY08 to 17AUG08, a metric came to our attention called, “Depth of Visit” which records how many pages each visitor looks at when coming to the site.

1 page for 730 visitors
2 to 5 pages for 2664
6 to 10 pages for 190
and 11 or more pages, 75 visitors.
Many-to-one Ratio: 4.01

In human, this means that in the month-long period half-way through July and half-way into August, 2,929 people bothered to look at more than just one page. That’s 4 times more _ than people who looked at only one page on the site and then left for greener pastures. Could we better this _and push the five page visitors to viewing 6 to 10 pages? A good challenge for any website.

Enter Feedjit’s “Live Recommended Reading” widget. We bounced the Page Popularity widget because long permlink titles broke through CSS float widths. The ‘also liked’ recommend widget was one that would also house permlink titles to similar articles, but we could yank its width from 200 to 568 pixels; plenty of space for the same long titles. Recommending similar pages is tricky: Do you base it on an indexing spider’s observations or do you base it on visitor browsing history? The spider-method, we assume, would help us lengthen Visit Depth where the spyware method would probably leave the next month’s results mostly unchanged.

17AUG08 to 17SEP08 RESULTS:
1 for 871
2 to 5 for 3806
6 to 10 for 213
and 11 or more, 93
Many-to-one Ratio: 4.72

17SEP08 to 17OCT08 RESULTS:
1 for 933
2 to 5 for 3557
6 to 10 for 206
and 11 or more, 87
Many-to-one Ratio: 4.13

With TWO MONTHS of testing, let’s look at four months total now including JUNE to JULY: So, two months without the widget and two WITH the widget to see if “Depth of Visit” increased at a faster rate than the overall site visits (thankfully, Dinarius is growing monthly).

17JUN08 to 17JULY08 RESULTS:
1 for 689
2 to 5 for 2137
6 to 10 for 164
and 11 or more, 55
Many-to-one Ratio: 3.41

The following graph shows the different depths as percentages of all total site visits. They omit robots, 11 or more page view visits and IMG hits so the percentages DO NOT add up to 100% I was looking for steep steps upward in the 6 to 10 page depth as the second two months came in. I got a surprise:

Feedjit's also liked spider widget does contribute to a website's Depth of Visit somewhat.

The graph and the little bit of raw data show that the number of 1 page visitors is increasing (more visitors to robots ratio implied) and more visitors are bumping around to 2 to 5 pages out of curiosity anyway while the 6 to 10 page geeks have remained relatively unchanged.

In conclusion, either the site is being better targetted in search engines despite fewer visitors coming from keywords or some link was promoted somewhere I’m not aware of. The Feedjit widget spike is nice to see but couldn’t account for the All/5pp Ratio step in the first two columns that equates to more than 5% On the other hand, I could claim the other 10% jump on Feedjit but don’t know if that’s truly the case. Feedjit “also liked” widget: Can’t hurt! Might help! Fully customizable and actually a logical tool.

Favorite's the ARTICLE, not the SITE.