This data is a taller order than one might think. In most cases Wikipedia cannot verify which edits have been made by someone affiliated with the article-subject. Therefore they cannot separate sponsored edits from regular ones in order to collect data on them. In those cases where a financial connection has been verified, the data selection bias of only focusing on exposed sponsored edits would make the data unusable.
Meanwhile, Ethical Wiki has provided Wikipedia assessments to 75 article-subjects over the last year. While this is a small data set compared to the 23,000+ pages associated with WikiProject Companies, it's some of the best, most objectively selected, data on sponsored editing available.
I've collected and analyzed those 75 assessments and will share the data here. All the data is anonymous, as to protect the confidentiality of those I've spoken with.
The data is based on 75 current, former or prospective clients that asked about Ethical Wiki's services and received an assessment. 20 percent of them were people and 80 percent were representing an organization. 56 percent of our prospective clients were referred from another client or agency partner, and 44 percent found us online.
Work we didn't accept
In 40 percent of cases I recommended a potential Ethical Wiki client abstain from Wikipedia and not deploy our services. This is because their objectives were too malaligned with Wikipedia's rules for us to ethically meet their needs. Out of those projects we declined:
- 52% didn't meet Wikipedia's requirements for an article
- 38% felt their Wikipedia article was unfair to them, whereas I didn't feel this was actually the case
- Only 1 was a clear "bad faith" situation, where the company wanted to censor negative information and had no interest in ethics.
However, a large number of people and businesses covered by Wikipedia have a skewed perspective on how a "neutral" article should reflect on them, or mistakenly believe they are famous enough to meet Wikipedia's "notability" requirements for an article. In other cases, they may lack understanding of what kind of participation is ethical. They may want to "ethically" produce a Wikipedia article that reflects company messaging, without understanding that this is deceptive to Wikipedia's readers.
Work Ethical Wiki took on
Prospective clients that were sent to us by PR, SEO or digital marketing agency partners were much more likely (75% versus 45%) to be accepted. Out of those clients we took on, 70 percent were able to get a neutral Wikipedia article approved through formal corporate cycles. In 30% of cases, the content we provided to Wikipedia was not formally approved by the client. In all but 1 case where the content was made live on Wikipedia, clients were satisfied with the results, even in those cases where they did not see the content before-hand and had no influence over it.
In 20% of our client engagements, at least one stakeholder within the client organization argued aggressively for self-serving violations in Wikipedia's "neutral point of view" policy. In most of these cases the client - upon initial review of the first draft - immediately removed most of the negative or controversial information then proceeded to attempt to make arguments to support their deletions. In more than half of cases the situation was worked out through consulting and education.
In 7% of cases, no resolution was reached and Ethical Wiki clients took advantage of the terms of our agreement that allow them to exit their subscription contract in the event of a conflict with our ethics policy.
Almost everyone looking for a Wikipedia service want to contribute to their Wikipedia page ethically, but only 52 percent have objectives that can be ethically achieved. We've had good success rates at persuading potential clients to abstain from Wikipedia, once we explain Wikipedia's rules and give them perspective. I think this is a strong indication that article-subjects want to do the right thing, but merely have a skewed point-of-view and objectives that conflict with Wikipedia's.
The data also suggests that Wikipedia astroturfing services breach Wikipedia's principles and policies at a much higher ratio than there is actually demand for. This may be explained through a data selection bias, as our firm name, "Ethical Wiki" will attract a certain kind of client. However, my observation has been that vendors are often creating promotional articles for clients that actually want neutral pages and doing so covertly for clients that actually want to do the right thing.
Additionally, the amount of resistance to neutrality was lower than I expected. The majority of our clients, 80%, only have reasonable input and do not lobby us to make bias edits on their behalf. Again, there may be a data selection bias, as these clients have had extensive up-front consulting from Ethical Wiki.
Better and more data is needed, but the results are interesting.