Share of mentions isn’t the same as share of voice
(A thread)
(A thread)
Share of voice (SOV) is a strong tool in advertising planning and budgeting. There’s convincing evidence that brands whose SOV consistently exceeds their market share (eSOV) can grow that share.
There’s increasing evidence that “Share of search” is a strong predictor of future market share.
ipa.co.uk/news/binet-pre…
ipa.co.uk/news/binet-pre…
However, “social SOV” (which for structural reasons usually means “share of Twitter mentions” isn’t predictive.
It’s is trivially simple to increase share of mentions on Twitter. This alone should make one take stock.
It’s is trivially simple to increase share of mentions on Twitter. This alone should make one take stock.
SOV is a measure of visibility; and mentions don’t map well to visibility.
The problems with confusing “mentions” with “visibility” are manifold. Some key issues are as follows:
The problems with confusing “mentions” with “visibility” are manifold. Some key issues are as follows:
1. There is significant ‘follower inequality’ (the median Twitter account has c. 60 followers; if you have 1,000 you’re in the 96th percentile), 
2. The algorithmic feed, the sheer volume of content, and the short half-life of a tweet’s (15-30 minutes) mean that only a small fraction of an account’s followers are likely to see a given tweet.
3. A simple single measure like “Share of Mentions” doesn’t account for the fact that the “social” in social media speaks to soft values like reputation, authority, and trust.
Consider mentions from:
* ‘egg’ account with no followers
* ‘Russian bot’ (with bot followers)
* member of the public
* journalist with a broad interest remit
* analyst with a relevant interest remit
* client
* politician
* celebrity
* expert (author)
* ‘egg’ account with no followers
* ‘Russian bot’ (with bot followers)
* member of the public
* journalist with a broad interest remit
* analyst with a relevant interest remit
* client
* politician
* celebrity
* expert (author)
Market researchers are fond of saying that Twitter provides excellent qualitative data; but at such volume that it’s tempting to treat it as though they were quantitative data. We’d support this POV.
We’d suggest that Twitter can provide at best — a proxy for public opinion (for very large audiences, and very macro themes). At worst it can be a distraction.
However, for certain topics and certain audiences, Twitter can help us understand the discourse.
Notably, Twitter is particularly attractive to people who have (or require) a platform associated with their role: journalists, politicians, activists, analysts, experts.
Notably, Twitter is particularly attractive to people who have (or require) a platform associated with their role: journalists, politicians, activists, analysts, experts.
We sometimes recommend identifying and profiling these accounts based on metrics like “relevance”, “activity”, “on-topic activity”, “in-group engagement” etc.
A combination of machine-led data collection backed up by ethnographic research can identify influential voices, themes, and potential allies.
Previous projects have demonstrated that analysing tweets from a small, manually-selected sample (e.g. topic editors and journalists from quality and business press, key bloggers, and market analysts) can yield predictive results (editorial themes, coverage, trending stories).
TL;DR
Don’t use Social Share of Mentions as a target. In social media analysis, “who” usually matters much more than “how many”
Don’t use Social Share of Mentions as a target. In social media analysis, “who” usually matters much more than “how many”
• • •
Missing some Tweet in this thread? You can try to
force a refresh
