Professional Documents
Culture Documents
Forward:
The project below represents an exercise in editorial content strategy for the website Refinery29. The first many
pages of Part I are based on quantitative analysis of visitor data. The final section of Part I (beginning at the bottom
pg. 5) embarks on a more qualititative analysis, which is perhaps more suited to content strategy at Huge.
Part II (pg. 7) is wholly qualitative and contains my editorial critique of a minimally-shared facebook post and my
recommendations for improving it.
Enjoy!
1) Increasing pageviews
My specific recommendation for getting more pageviews, and thus revenue from ad sales, is a two-parter.
i.
Publish more Slideshows
ii.
Publish more content, specifically articles, that have multiple pages
Right off the bat I noticed that Slideshows generated the most pageviews of all the entry types published in the
month of May, 2013. Slideshows comprised just under a third of the content published with 31.6% of the total
number of posts, yet they account for 83.5% of the total pageviews among content published that month. They also
comprise a significant number of the total visits to these articles, 41.6%. (See below)
Entry Type
ARTICLE
Pageviews
# of
Visits
%total PV
posts
%total
%total
visit
posts
4,844,885
2,432,595
727
15.79%
56.28%
61.71%
LINKS
31,361
9,195
48
0.10%
0.21%
4.07%
SHOPPABLE_ARTICLE
13,624
2,797
0.04%
0.06%
0.51%
SHOPPABLE_VIDEO
56,804
31,116
0.19%
0.72%
0.34%
25,607,804
1,797,806
372
83.46%
41.59%
31.58%
SLIDESHOW
VIDEO
126,683
48,720
21
0.41%
1.13%
1.78%
TOTAL
30,681,161
4,322,229
1,178
100.00%
100.00%
100.00%
Taking a different view of the data shows average pageviews and visits among the Entry Types, and their %
differences from the overall averages.
Entry Type
ARTICLE
# of
Avg
posts Pageviews
Avg
% diff from
Visits
avg PV
Rate
727
6,664.2
3,346.1
-74.41%
-8.80%
70.12%
48
653.4
191.6
-97.49%
-94.78%
76.14%
SHOPPABLE_ARTICLE
2,270.7
466.2
-91.28%
-87.29%
55.83%
SHOPPABLE_VIDEO
14,201.0
7,779.0
-45.48%
112.01%
64.27%
372
68,838.2
4,832.8
164.30%
31.72%
25.00%
LINKS
SLIDESHOW
VIDEO
21
6,032.5
2,320.0
-76.84%
-36.77%
69.89%
TOTAL
1,178
26,045.1
3,669.1
0.00%
0.00%
56.02%
As shown in the above table, Slideshows are by far the most successful on average across both pageviews and visits
(also bounce rate). Page views and bounce rate can be partially explained by the high much higher than average
number of pages comprising a slide show. The average number of pages for slideshows is 30.8, 1.3 for articles, and 1
for all the other entry types. It does not, however, explain the much higher than average visits among slideshows,
which will be touched on later.
After the above, the second part of my recommendation for boosting page views seems much more obvious: Post
more multi-page content in lieu of 1-pagers. The data supports this by showing a general trend towards more
pageviews as the page count increases.
The relationship holds up for the entire
set of May, 2013 posts (see figure to left),
but also when looking at just the articles.
This is important because articles make up
most of the stories published for the
month, and also because slideshows by
nature have multiple pages and we
already know they get a lot of pageviews.
The important thing to note here is the
line in red, which models the overall
trend that increasing page count increases
page views. The R2 represents the
tightness of the fit to the data. The positive value means that it more precisely represents the trend in the data than
would a simple horizontal line for the mean. Though the value is small (R2=1 would be perfect), this is most likely
due to the small size of the data set, as evidenced in the graph below.
These relationships still held true when
accounting for outliers (*see final
pages of project for the data and
description of the techniques used for
determining outliers).
The problem here is that the articles
and other content that do have
multiple pages are much like
slideshows themselves, and not every
post can work in that format.
2) Increasing visits
i)
One way to increase the number of visits was already mentioned, namely boosting focus and
output of slide shows.
Thus, my recommendation and explanation centers on how to gain more visits to articles by looking at the kind of content
that is already successfully bringing in visits.
ii)
Based on the analysis below, my recommendation for increasing visits to articles
(and in general) is to double down on the following content categories
a. how-tos,
performance in terms of average visits, News and How-To. News Articles comprise nearly 20% of the
total published content for the month, but cover just of 35% of the total visits.
The How-to articles numbers (as shown in the first of the above tables) are modest by comparison; further
exploration revealed that it could be a potentially very impactful category.
The second table the just shows the above average performers among the subcategories that relate to my
recommendations. The most significant point being the last line, which shows celebrity-news articles as not
only having a much higher than average number of visits, but also that it also that it makes up a large chunk
of total the overall total stories. This is what we want to see. The goal is to have high performing
content make up a significant number of total publications. The larger data pool also ensures that these
findings are more statistically reliable and should hopefully hold water in the future.
On the other hand, only 2 how-to/what-to-wear (one of which could potentially qualify as a moderate
outlier) so the data isnt good. The analyses that follow explain why I think its a good bet, however.
The first table below shows the entire population of content (all entry types) in terms of their broader
categories/verticals. The lines highlighted in blue show the verticals that successfully drew in higher than
average visits.
A closer look at beauty, how-to, and news into their subcategories is just to provide further insight and to
start to paint a picture of the kind of content the succeeds on refinery29
Category Path
beauty/fitness
Pageviews
Visits
# of
%total PV
posts
%total visit
%total
Avg
posts
Pageviews
Avg Visits
avg visits
Avg.
Bounce
Rate
386,096
58,044
1.26%
1.34%
0.34%
96,524.0
14,511.0
270.60%
295.49%
51.61%
beauty/wellness
1,609,983
309,660
26
5.25%
7.16%
2.21%
61,922.4
11,910.0
137.75%
224.60%
53.07%
how-to/i-diy
1,562,561
136,836
5.09%
3.17%
0.51%
260,426.8
22,806.0
899.91%
521.57%
61.88%
how-to/styling-tips
1,600,261
122,269
12
5.22%
2.83%
1.02%
133,355.1
10,189.1
412.02%
177.70%
27.22%
74,017
15,438
0.24%
0.36%
0.17%
37,008.5
7,719.0
42.09%
110.38%
47.21%
3,380,052
1,282,893
129
11.02%
29.68%
10.95%
26,202.0
9,944.9
0.60%
171.04%
70.15%
news/beauty
296,343
198,383
26
0.97%
4.59%
2.21%
11,397.8
7,630.1
-56.24%
107.95%
78.50%
news
430,862
108,432
20
1.40%
2.51%
1.70%
21,543.1
5,421.6
-17.29%
47.76%
69.04%
how-to/what-to-wear
news/celebrities
While theres a lot of data that seems to support my recommendations for increasing visits by increasing output for
high performing content categories, these just represent averages. The majority of the categories have stories that are
successful and others that underperformed as well.
My 3rd recommendation(s) is based on a more qualitative approach, though my conclusions are informed by my work
with the data.
- Use the following guidelines headlines and subject matter.
- Produce more instructive content How-tos, style tricks, life hacks, etc.
Looking at top performing stories shows that many of them relate to common themes.
1) Relatability
a. While most of the top performers related to things (people, issues, etc.) that I personally know
about and think about,
b. The right column (worst performers) was rife with things and people Id never heard of and dont
think about.
2) Curiosity Gap
a. Top performers on the whole had headlines have the right about of information (bait) to get me to
bite without giving the whole article away.
b. Underperformers on the right were often either too vague as to not incite my curiosity, or
conversely too informative, taking away the incentive to click and read since I already got the gist.
3) Celebrity:
a. In the left column, top performing content related to primarily, young current buzzworthy names.
Many of them considered style icons, notorious beauties and/or it-people.
b. Celebrities mentioned in headlines in the right column were not. They are primarily lesser-known
names, and if not, people that couldnt be described as pioneers of modern culture.
4) How-tos for the refinery29 readers
a. Many of the top performing stories were instructive, particularly in ways that seemed easy and
nonspecific enough to replicate while appealing to modern set of priorities. (I myself wrote a DIY
cut-off instructional this past summer for fun)
b. There were far fewer explicitly educational/informative headlines among the least visited posts,
and those that were often related to specific stores that made it seem too particular to be achievable
Lastly establishing an emotional connection is another good way to attract readers. Phrasing headlined in order to
maximize Shock/surprise (the OMG feeling), outrage, sympathy for a perceived villain or what I like to call the you
go boy/girl! feeling is a potential way to increase traffic and thus revenue.
Thanks,
Nick