Network real estate analysis
View Patent ↗A method can be used to analyze the “real-estate” performance of content items within a network site. The method can comprise determining the click distance to reach each content item and determining the performance of each content item. The method can also comprise calculating a predicted value for performance based on statistical relationship between location and performance observed in a population of content items. The method can comprise comparing the predicted and actual performance. If a content item has an actual performance greater than its predicted performance, then it may be promoted to a better location in the site and the converse for poorer performing content.
1. A method for analyzing performance of content items within a network site, comprising:
at a server computer having access to network activity information on the network site:
determining a location of a content item within the network site;
determining a predicted performance for the content item at the location;
determining an actual performance of the content item;
comparing the predicted performance to the actual performance;
based on the comparing, determining whether the content item is underperforming; and
moving the content item to a new location within the network site responsive to the determining whether the content item is underperforming.
2. The method according to claim 1 , wherein determining the predicted performance of the content item comprises analyzing a relationship between location and performance of other content items.
3. The method of claim 1 , wherein determining the location of the content item comprises determining an average number of clicks it takes users to arrive at specific network address during a session at the network site.
4. The method of claim 1 , wherein determining the actual performance of the content item comprises summing the number of session in which the content item is requested.
5. The method of claim 1 , wherein the predicted performance of the content item is determined according to:
P
′
(
c
i
)
≈
∑
d
=
1
D
[
G
(
r
d
-
L
(
c
i
)
)
·
t
d
]
where:
c i represents the content item;
L(c i ) is the location of the content item;
D is a number of basis functions used for the approximation
G(g)=g 2 *log(g);
r d is a prototypical location value for the d th basis function; and
t d is a parameter value that is chosen to minimize the sum of squared errors below over a large “training set” of content items.
6. The method according to claim 2 , wherein comparing the actual performance to the predicted performance comprises determining a real estate performance value according to:
REP
(
c
i
)
=
P
(
c
i
)
P
′
(
c
i
)
.
7. The method according to claim 6 , wherein if the content item has a REP<1, the content item is determined to be underperforming.
8. A data processing system readable storage medium having computer readable program code embodied therein, the computer readable program code adapted to be executed by a computer to implement a method for analyzing performance of content items within a network site, the computer readable program code comprising instructions executable to:
determine a predicted performance for the content item at the location;
determine an actual performance of the content item;
compare the predicted performance to the actual performance;
based on the comparison, determine if the content item is underperforming; and
if the content item is underperforming, recommend moving the content item to a new location within the network site.
9. The data processing system of claim 8 , wherein determining the predicted performance of the content item comprises analyzing a relationship between location and performance of other content items.
10. The data processing system of claim 8 , wherein determining the location of the content item comprises determining an average number of clicks it takes users to arrive at a specific network address during a session at the network site.
11. The data processing system of claim 8 , wherein determining the actual performance of the content item comprises summing the number of sessions in which the content item is requested.
12. The data processing system of claim 8 , wherein the predicted performance of the content item is determined according to:
P
′
(
c
i
)
≈
∑
d
=
1
D
[
G
(
r
d
-
L
(
c
i
)
)
·
t
d
]
where:
c i represents the content item;
L(c i ) is the location of the content item;
D is a number of basis functions used for the approximation
G(g)=g 2 *log(g);
r d is a prototypical location value for the d th basis function; and
t d is a parameter value that is chosen to minimize the sum of squared errors below over a large “training set” of content items.
13. The data processing system of claim 12 , wherein comparing the actual performance to the predicted performance comprises determining a real estate performance value according to:
REP
(
c
i
)
=
P
(
c
i
)
P
′
(
c
i
)
.
14. The data processing system of claim 8 , wherein if the content item has a REP<1, the content item is determined to be underperforming.
15. A data processing system for analyzing performance of content items within a network site, comprising:
a storage location storing network activity information on the network site; and
a server computer coupled to the storage location comprising a processor and a computer readable storage medium storing computer executable instruction, the computer executable instructions comprising instructions executable to:
determine a predicted performance for the content item at the location;
determine an actual performance of the content item;
compare the predicted performance to the actual performance;
based on the comparison, determine if the content item is underperforming; and
if the content item is underperforming, recommend moving the content item to a new location within the network site.
16. The data processing system of claim 15 , wherein determining the predicted performance of the content item comprises analyzing a relationship between location and performance of other content items.
17. The data processing system of claim 15 , wherein determining the location of the content item comprises determining an average number of clicks it takes users to arrive at a specific network address during a session at the network site.
18. The data processing system of claim 15 , wherein determining the actual performance of the content item comprises summing the number of sessions in which the content item is requested.
19. The data processing system of claim 15 , wherein the predicted performance of the content item is determined according to:
P
′
(
c
i
)
≈
∑
d
=
1
D
[
G
(
r
d
-
L
(
c
i
)
)
·
t
d
]
where:
c i represents the content item;
L(c i ) is the location of the content item;
D is a number of basis functions used for the approximation
G(g)=g 2 *log(g);
r d is a prototypical location value for the d th basis function; and
t d is a parameter value that is chosen to minimize the sum of squared errors below over a large “training set” of content items.
20. The data processing system of claim 19 , wherein comparing the actual performance to the predicted performance comprises determining a real estate performance value according to:
REP
(
c
i
)
=
P
(
c
i
)
P
′
(
c
i
)
.
21. The data processing system of claim 15 , wherein if the content item has a REP<1, the content item is determined to be underperforming.