Permalink
Fetching contributors…
Cannot retrieve contributors at this time
216 lines (167 sloc) 16.7 KB

Further social media usage analysis

We're seeing strong engagement from social media so which pages are being linked to and what conversion rates are we seeing? Are particular pages on the site generating interest or do is the interest spread out across the site?

Answer

TL;DR: Focused on a small number of request pages, but there are doubts about the accuracy of the conversion rate data. Note that this looks as what Google Analytics tags as "Social" rather than the t.co or facebook.com referrer data.

Analysis

We looked at data from Ask The EU for 16 March - 15 May 2016 (to match the reporting period of our earlier report "Are some sources of traffic better in terms of people making requests?"). This is based on Google Analytics data and "Conversion" refers to the completion of the new request process without distinguishing between a single person making multiple requests and many people making one request each.

Baseline Traffic analysis

Our traffic to the site - measured in terms of sessions - looks like this, with Social providing less than 5% of the traffic but with clearly the highest conversion rate of any traffic channel.

Channel Sessions % of Sessions Conversions Conversion Rate
Organic Search 6,152 57.19% 83 1.35%
Direct 3,178 29.54% 125 3.93%
Referral 1,004 9.33% 27 2.69%
Social 424 3.94% 48 11.32%

Source: Google Analytics - Acquisition > All Traffic > Channels

Drilling down further shows that most of this traffic has come from Twitter and Facebook.

Social Network Sessions % of Sessions Conversion rate
Twitter 268 63.21% 15.30%
Facebook 140 33.02% 5.72%
Reddit 8 1.89% 0.00%
Delicious 3 0.71% 0.00%
LinkedIn 2 0.47% 0.00%

Source: Google Analytics - Acquisition > All Traffic > Channels > Social

So we can immediately discount Reddit, Delicious and LinkedIn for our current purposes as they do not have any associated conversions.

Twitter

Top referrers by conversion

Referral Path Sessions % of Sessions Conversions Conversion Rate
t.co/iwDGVPUbVD 76 28.36% 1 1.32%
t.co/CWyoh6R6Df 36 13.43% 3 8.33%
t.co/J6x93pE4Cy 28 10.45% 26 92.86%
t.co/Oaq1xTs5bt 15 5.60% 2 13.33%
t.co/UG2cL2mHnQ 13 4.85% 2 15.38%
t.co/VrmZfcM4d6 11 4.10% 3 27.27%
t.co/Nq26nf9lnJ 2 0.75 2 100.00%
t.co/YWW1xSVTcZ 2 1.45% 2 100.00%

Source: Google Analytics - Acquisition > All Traffic > Channels > Social > Twitter > Secondary dimension : Referral path

Luckily for us, Twitter provides unique links via their t.co link shortener, which is passed through as the Referral Path so in most cases we can fairly easily see where a link was originally posted (although a bit of work is needed) to find out what page it's referring to. The table below adds in the site link and original Twitter details, where available (presumably data that is not available has either been deleted by the original author or is associated with a private account).

Referral Path Site link Original Tweet Tweeter
t.co/iwDGVPubVD Commissioner meetings - anti-contraband tobacco agreement with PMI Letter via DHLExpress? But @access_info got 📧 last week 😒 #BerlaymontSnail strikes again 🐌 [link] Ask The EU official account
t.co/CWyoh6R6Df EU Action on Migratory Pressures - targeted update on Turkey Freedom of information: more confidential details on the EU's action plan with Turkey on migration @AskTheEU [link] Keegan Williams
t.co/J6x93pE4CY Is glyphosate safe? We have the right to know! Is #glyphosate safe?Follow our document request to @EFSA_EU w/ @MicheleRivasi @HeidiHautala @BartStaes @FoodRevEU [link] Jávor Benedek, Hungarian MEP
t.co/Oaq1xTs5bt Documents linked with "First Reading" Agreements in co-decision procedures 'Documents linked with "First Reading" Agreements in co-decision procedures' [link] via @asktheeu Emilio De Capitani
t.co/UG2cL2mHnQ Is glyphosate safe? We have the right to know! Ist #Glyphosat sicher? Wir haben ein Recht es zu wissen [link] via @AsktheEU Martin Häusling, German MEP
t.co/VrmZfcM4d6 documents assessing the situation in Poland under the Rule of Law Framework I asked the EU Commission documents re: situation in Poland under the Rule of Law Framework [link] @EUClinic @Access_Info Alberto Alemanno
t.co/Nq26nf9lnJ Background Paper - A Digital Trade Initiative in the WTO unknown unknown
t.co/YWW1xSVTcZ Sion Simon MEP Expenses unknown unknown

(Note the 2 glyphosate links are slightly different - the first one goes to the /new link)

Possible flaws

It is noticeable that the 2 highest conversion percentages come from links which end in /new which is the last page of our "Make a request" goal. These may be false positives. Adding in the number of pages per session throws some doubt on the accuracy of these numbers:

Referral Path Sessions Pages/Session Conversion Rate
t.co/iwDGVPUbVD 76 3.05 1.32%
t.co/CWyoh6R6Df 36 6.67 8.33%
t.co/J6x93pE4Cy 28 1.11 92.86%
t.co/Oaq1xTs5bt 15 4.47 13.33%
t.co/UG2cL2mHnQ 13 3.31 15.38%
t.co/VrmZfcM4d6 11 7.45 27.27%
t.co/Nq26nf9lnJ 2 1.50 100.00%
t.co/YWW1xSVTcZ 2 1.00 100.00%

Source: Google Analytics - Acquisition > All Traffic > Channels > Social > Twitter > Secondary dimension : Referral path

The 3 highest conversion rates are recorded as only have between 1 and 1.5 pages per session, admittedly this in an average but particularly as in two of these cases there are only 2 sessions, this isn't looking good. However, as a counter-argument, it should be noted that t.co/VrmZfcM4d6 - with a more plausible conversion rate of 27.27% (with a 7.45 pages/session average) also links to a /new format link so there's no straightforward answer.

Discounting all 3 of these unlikely entries entirely puts Twitter's conversion rate at 4.10% which is still the highest in our sample (even if only just) but less eyecatching than the figure we started with.

Who is Tweeting?

Sticking with our small sample of 8 tweets seems unlikely to be helpful so we've cast our net more widely to answer this one by searching Twitter itself (the search query starts earlier than our Google Analytics sample set in the hopes of catching a higher number of tweets, although interestingly this still does not include all of the public tweets we started with). Tweets that pointing to the blog or the front of the site - or have typoed the link - have been ignored.

Referral Path Site link /new? Tweeter Requester Requester Tweet?
t.co/xwOON1rsKW Jobs in automotive industries Yes Jens Jürgen Korff (@jejko) Jens Jürgen Korff Yes
t.co/T5KhiSuA9Z Circulation Fiche Adonis No ACT (@ACT_ACT) Arun Dohle Maybe
t.co/NkWUdMH9tF Database used to record entries at the EU Yes Alba Gutiérrez (@albagutierrezf) Alba Gutiérrez Yes (possibly via the site)
t.co/oRY5mYOz4R Frontex report about data of refugees to which EU Comission Acknowledges Yes Alba Gutiérrez (@albagutierrezf) Alba Gutiérrez Yes (possibly via the site)
t.co/Ofmza5S9pA Details of meetings with tobacco industry or packaging industry No steve white (@lostexpectation) Colm Fahey No
t.co/wjGq8EfuQ1 Correspondence Against Child Trafficking (pdf) No ACT (@ACT_ACT) Arun Dohle Maybe
t.co/SfYiZCubx5 Unicef No ACT (@ACT_ACT) Arun Dohle Maybe
t.co/iwDGVPubVD Commissioner meetings - anti-contraband tobacco agreement with PMI No AsktheEU (@AskTheEU) Pau Salarich No
t.co/oqcNqHMTO1 Meeting Mimica -Lumos No ACT (@ACT_ACT) Arun Dohle Maybe
t.co/23CVehP7Sr Source code for European Data Portal Yes Ross (@rossjones) Ross Jones Yes (possibly via the site)
t.co/d95X7nN43m Apple and TTIP lobbying No Elza van Swieten (@confluencemedia) Vicky Cann No
t.co/UG2cL2mHnQ Is glyphosate safe? We have the right to know! No Martin Häusling, German MEP (@MartinHaeusling) Heidi Hautala, Benedek Javor, Michele Rivasi and Bart Staes No
t.co/bxHbQ4M3EQ Frontex report about data of refugees to which EU Comission Acknowledges (pdf) No Keegan Williams (@keegan_aw) Alba Gutiérrez No
t.co/CWyoh6R6Df EU Action on Migratory Pressures - targeted update on Turkey No Keegan Williams (@keegan_aw) Michael Bauer No
t.co/2fIZ6OhJv9 Is glyphosate safe? We have the right to know! No guendalina de sario (@guendalinads) Heidi Hautala, Benedek Javor, Michele Rivasi and Bart Staes No
t.co/i3pq5Ldnyp [DG Trade contacts with industry lobbyists about EU-US trade negotiations](DG Trade contacts with industry lobbyists about EU-US trade negotiations) No jaka (@jcrgordon5) Olivier Hoedeman No
t.co/Tlsdz8SwuV Safeguard Mechanism: United Kingdom and free movement Yes NiccoloMilanese (@NiccoloMilanese) Niccolo Milanese Yes (possibly via the site)
t.co/UljjzLy0OO Does the United Kingdom government make Payments to the IMF No Abdul Hai (@AbdulHai23) GEORGE TSIGARIDES No
t.co/HTbRn6jyr8 Latest available Operational Plans for Operations Poseidon Sea and Poseidon Land, Triton, Alexis, Good Will and Serbia & Albania (pdf) No Keegan Williams (@keegan_aw) Amélie POMEON No
t.co/GoleyxQqK5 London City Airport No Abdul Hai (@AbdulHai23) Abdul Hai Yes

Data from: https://twitter.com/search?q=asktheeu%20since%3A2016-01-01%20until%3A2016-05-16&src=typd

This gives us 20 tweets to look at. Of those, 6 are definitely someone tweeting their own request and a further 4 look like they might be, the rest are people finding interesting things are sharing them (with the possible exception of the 2 tweets - one from the maybe pile and AskTheEU's one which we've said probably doesn't count pointing out that email is much more efficient than the waiting for physical copies of documents in the post). Which gives us roughly a 50-50 split.

However, although not all users who share their own request use the /new form of the url (the one we suspect upsets our analytics magic), all the /new links in this random sample are from users sharing their own requests.

Facebook

Landing Page Sessions % Sessions Conversions Conversion Rate
Is glyphosate safe? We have the right to know! 7 5.00% 7 100.00%

Source: Google Analytics - Acquisition > All Traffic > Channels > Social > Facebook > Secondary dimension : Landing Page

Unlike Twitter, the Referral Path data from Facebook doesn't really helps us so we're having to make do with the Landing Page. From a Google search, the main page linking to this request appears to be a public group page called "Stop Monsanto before it's too late!".

User journeys

It's at this point that something odd happens - either the user journey data is incomplete or the analysis above is returning false positives (or possibly a mixture of these).

For example, the Behaviour Flow report shows 181 sessions that start with the glyphosate page from t.co/J6x93pE4CY, and our only solid Facebook lead - 80 with /new and 101 without.

Of the 80 "new" sessions, 75 leave straight away and the rest only look at one more page before disappearing.

Of the other 101 sessions, only 22 make it to a second page and 16 stay to have a proper look around. No-one seems to make a request (and one appears to be the request owner) who adds a follow-up.

So unfortunately I have not been able to determine which requests were made as a result of visiting the social media links, or even if the data is correct.

Conclusions

Just 6 request pages appear to be commanding all the attention here. However the lack of evidence of subsequent requests made in the Behaviour Flow report and the possible false positives we found in the data, it's hard to be confident about the conversion rates reported by Google Analytics.