This is a paper I wrote in Fall Quarter 2009 for the Master of Communication, Digital Media program’s “Research and Methodology”core class. The assignment was to propose a research project, with the grade being given on the form and thoroughness of the proposal. The CRED+STAMP metric I propose is one way to approach the problem of measuring campaign effectiveness through social media. I think it has value and could be pursued. However, I offer it with the caveat that more effort went into the form of the paper than the substance.
Brook Ellingwood
COM 529: Research and Methodology in Digital Media
December 2, 2008
Instructor: Hanson Hosein
Abstract
A proposal for a project to develop a new metric for measuring the change of conversational topics in social media channels coinciding with advertising or marketing campaigns.
Project Summary
The CRED+STAMP project is an effort to apply techniques of data mining and textual meaning analysis to social media sources. It seeks to determine the value of conversations in the social media space to companies engaged in campaign-style marketing efforts. These campaigns may support commerce or other activities.
Inspiration for CRED+STAMP can be summarized by a quote, attributed to department store magnate John Wanamaker: “Half the money I spend on advertising is wasted; the trouble is I don’t know which half” (“John Wanamaker Quotes,” 2008). At the root of this complaint is the simple fact that human behavior can be directly observed, but human motivations must be inferred, even if they are self-reported via polling.
CRED+STAMP is thus a proposed metric to increase the amount and accuracy of inferred data available to marketers and advertisers. Each letter in the acronym is the first letter in a word used to discretely measure aspects of the social media sources being analyzed.
CRED, the X axis:
- Commentary
- Rumor
- Expression
- Distraction
STAMP, the Y axis:
- Site
- Topic
- Account
- Message
- Phrase
The value of this information to marketers can be enormous, especially for those who are looking to ease acceptance of social media within organizations that have been resistant to it. In this paper, I will describe in some detail a proposed methodology for using text data mining to demonstrate measurable outcomes of using social media.
Project Details
Genesis
My proposal for a CRED+STAMP metric is heavily influenced by my professional experience, and bolstered by research showing that others have similar experiences. The “Social Media Resistance Survey” (Pereira, 2008) showed that 26% of respondents felt their organizations were resistant to new media and technologies because of an inability to demonstrate Return On Investment (ROI) and 31% due to the lack of ROI on applications.
At the 2007 Community 2.0 conference, I made a note when one main stage panelist suggested that asking for the Return on Investment of social media was the wrong question. Instead, he suggested, the right ROI for social media would be “Return on Information.” Similar ideas can be found in the blog posts of professionals working in social media:
…the consistent response I get is “Show me how to make money with it”. To this response I ask the following questions:
- How much money do you make with email?
- How much money do you make from your phone conversations?
- How to you measure the ROI on marketing materials, sales brochures, attendance at industry events?
- How much money do you make on a sales call?
- How much money do you make from your web site?
The consistent response is similar to a deer staring into the headlights of an oncoming car. (Deragon, 2008)
The problem with trying to determine ROI for social media is you are trying to put numeric quantities around human interactions and conversations, which are not quantifiable. (Falls, 2008)
These comments anecdotally illustrate the statistic that 53% of those surveyed felt their organization lacked a basic understanding of new media and technologies (Periera, 2008).
CRED+STAMP is a proposal to create a metric that can provide some insight into the value of social media in measuring campaign effectiveness. If CRED+STAMP could be an effective tool to bypass resistance to new media and technologies based both on concerns about ROI and lack of understanding, then it could be very useful to those looking to persuade organizations to use social media.
Instead of looking at monetary return, CRED+STAMP looks at conversational return by asking the question: “What are people saying about us and our products?” Without the knowledge inferred from knowing what people are saying, campaign results are seen with Wanamaker’s blind spot blotting out the middle. They may or may not make money, but either way, understanding what motivates behavior remains unknown.
Goals
Having concluded that Return on Investment is not a fruitful goal, my initial goal for the CRED+STAMP proposal is to promote conversation about the feasibility of extracting meaning from conversations in social media space.
If it seems worthwhile, CRED+STAMP may be further developed as a methodology, and the technical issues involved in applying it to real-world data can be quantified.
Should study of CRED+STAMP methodology and supporting technology show feasibility, a further goal of developing a business model based on offering CRED+STAMP analysis as a service would be justified.
Historical Context
The emergence of user-generated content as a major element of Internet content has clouded analysis of content quality (Agichtein, Castillo, Donato, Gionis, & Mishne, 2008). While CRED+STAMP as currently conceived has no concept of “quality” it does share some of the same challenges Agichtein, et al. describe in programmatically determining the best content submitted to Yahoo! Answers:
The main challenge posed by content in social media sites is the fact that the distribution of quality has high variance: from very high-quality items to low-quality, sometimes abusive content. This makes the tasks of filtering and ranking in such systems more complex than in other domains. However, for information-retrieval tasks, social media systems present inherent advantages over traditional collections of documents: their rich structure offers more available data than in other domains. In addition to document content and link structure, social media exhibit a wide variety of user-to-document relation types, and user-to-user interactions. (Agichtein, et al., 2008)
Given the general economic nature of technological societies, I suppose some sort of promotion of goods and services has been present in electronic media almost from beginning. However, the current advertising age on the Internet can be said to have begun in 1994, with the first clickable banner ad, widely credited to TANGENT Design/Communication’s “Have You Clicked Here?” banner on Hotwired.com as part of ATT’s “You Will” campaign (Dabitch, 2004). From the start, a major selling point for clickable advertising was that it provided performance-based pricing plans.
Performance-based advertising is advertising, typically used in electronic media, for which the advertiser only pays when a measurable outcome is achieved. The primary models are “pay-per-view” in which the desired outcome is displaying the ad to a user, and “pay-per-click” in which the desired outcome is for the user to click the ad and be taken to the advertiser’s target site (“Performance-based advertising,” 2008).
Performance-based advertising has succeeded because the hypertext-based user interface of Web browsers provides convenient mechanisms for circumventing Wanamaker’s conundrum. Where the pricing of newspaper ads, for example, is based on newspaper circulation numbers, an advertiser has no guarantee that any given reader will even open the newspaper to the page on which the ad appears. By contrast, user interaction with ads served through the Web are much more traceable. It can be known if a page with an ad is viewed, and it can be known if a link in an ad is clicked.
“The benefits of online advertising become even greater If you have a performance-based component to it,” said Vivian Wu, a VP at TA Associates, a private equity firm. …[The] next generation of online advertisers is creating models that can provide more specific data points. “If you achieve interaction with your customers, you should be able to quantify that” with conversion rates and ROI measurability, Wu said. “But we’re still in the early stages of online advertising.”Brand marketers are still less comfortable than direct marketers with the concept of performance-based marketing, being more accustomed with the glossy ads and the banners. “They will eventually be more comfortable with the action-based and performance-based stuff,” Wu said. (O’Conner, 2005)
AdWords and AdSense, Google’s pay-per-click performance-based advertising services, have leveraged very simple quantification of behavior to a powerful revenue model. A user uses a term in a search, in an email, on a blog, or in some other context were Google AdWords is enabled. The use of the term triggers the serving of ads. The user clicks on one of the ads, and the advertiser then pays Google a negotiated price for the click. More sophisticated advertisers then have the ability to track that user’s behavior as they navigate the advertiser’s site, and can accurately determine that X number of clicks on a Google ad triggered by keyword Y will result in conversion metric Z.
Google AdWords extracts meaning on a a fairly primitive basis, by allowing advertisers to set values for keywords by bidding on them. This results in a 1:1 correspondence between keywords and message. At this moment, if I search on Google for “The North Face” an ad for REI.com is in the #1 spot, and an ad for Backcountry.com is in the #3 spot. However, if I search Google for “North Face” they switch positions, with Backcountry.com taking #1 and REI.com taking #3. This shows the relative value each company has placed on these terms with their keyword bids.
When other sites subscribe to Google’s AdSense revenue model, they embed the ability for Google to analyze the content on their site in this same simple fashion. If you have a blog with a Google AdSense feed and use the word ‘North Face’ in a post, the Google AdWords engine will go displaying the advertisement with the highest bid for that keyword match just as it did when displaying ads directly on Google search results.
But a key difference is that now the phrase has context, which it did not before. If I go to Google and search for the phrase “North Face rocks” none of the advertising I get with my results is for REI.com, Backcountry.com, or any of the other retailers on the less-specific search, except for The North Face itself, which one suspects did not bid on the term expecting many results. If I search for the opposite phrase “North Face sucks” I get no results at all. Apparently no one wants to associate their name with that negative search term.
But one of the links in the search leads to a thread entitled “The North Face..(sic) SUCKS” on a discussion forum (“The North Face.. SUCKS,” 2005). On that page Google is serving two ads for competitors of REI.com and Backcountry.com. These companies almost certainly did not bid on the phrase “The North Face Sucks,” but Google’s matching algorithms have not been designed to do a lot of contextual analysis, so the ads get served on this page whether the retailers would like it or not.
Even in the simplified context of a pure search query, this model absolutely cannot tell the advertiser is why the user used that term in the first place. What is the motivation? If you run a successful commerce Web site maybe 1% of the people who mention your best-selling product in an online context are interested in buying it, but what about the other 99%? Why are they talking about it? When they talk about it, do they also talk about you? When you run an advertising campaign featuring that product and sales go up, does the online conversation change in some way that corresponds to sales? What else are people who talk about the product talking about? Do the things people say online about your product contain clues to other keywords you should include in your performance-based advertising plan?
The outcome of performance-based advertising models is that marketers and advertisers were drawn to the emerging Internet market with the promise that they would be able to directly track ROI. However, there is no applicable model for measuring the ROI of a social media campaign, which makes social media less attractive as an marketing tool.
Even companies that don’t engage with social media directly — or engage, but not effectively — can learn from monitoring social media conversations and then apply those learnings to other business decisions.
Description, Part One: The Graph
CRED+STAMP proposes increasing the value and decreasing the cost of inferred knowledge about marketing and advertising by data mining conversations in social media settings. While the approach outlined in this paper can’t provide a measure of the ROI of social-media related expenses, it can provide companies the ability to measure the visibility of target messages within social media contexts, providing more data for accurately inference.
Illustration 1: White noise is all pitches at the same volume. In the jargon of audio engineering, “white noise” is the sound of all the audio frequencies humans can hear, playing at an equal volume. This term has been popularly appropriated to refer to ongoing chatter from which no sense is being extracted, which is an apt description for how many advertisers and marketers perceive the Internet at large.
As in audio, the chatter of white noise may be filtered by tonal ranges. An audio pitch may correspond to single note, while in CRED+STAMP analysis the four CRED tones correspond to one of four categories of conversation:
- Commentary. Commentary is a statement of belief, opinion, or action characterized by the use of the first person pronoun and strong verbs: “I think…” “I know…” “I like…”
- Rumor. Rumor is hearsay, or commentary that may be stated in the first person but is not backed up by first person credibility: “I heard…” “My friend said…” “REI is…”
- Expression. Expression is creative speech, or speech that doesn’t fit patterns of commentary or rumor. It is the first of two categories by default, which exist to capture communication which may be about the topic of analysis but can’t be further classified.
- Distraction. Distraction is the second category by default. This is any speech that doesn’t pertain to the topic of analysis in any way and therefore is not analyzed further.
Fine-tuning these definitions is a task best suited for someone with a background in linguistics, and would be an early step in any actual development of CRED+STAMP as a methodology.
In CRED+STAMP analysis, volume is an absolute number of CRED pitch determined by data mining discussion at five discrete levels:
- Site. A site is a channel or domain. Examples might be Twitter.com, Facebook.com, or YouTube.com.
- Topic. The term “topic” is borrowed from message boards, where a topic might be “Crosscountry Skiing” or “Mountain Biking.” In other contexts, topics may be defined according to the architecture of individual sites, according the needs of the particular CRED+STAMP analysis being performed: Twitter topics might be all tweets by the followers of a given account, or all tweets by the followers and second-generation followers of the account; Facebook topics might be the news feeds of all the friends of a given account; YouTube topics might be all the comments on all the videos returned by a site search for “snowboard.”
- Message. A message is a single post by a single account.
- Phrase. A phrase is an identifiable part of speech with a subject, a noun, and a predicate.
Illustration 2: CRED pitches graphed to show relative volumes at a single STAMP level. Account. An account corresponds to a user name. While it may be generally assumed that accounts exist in a 1:1 correspondence to individuals, it’s safer to think of them as representing a single “voice.” The voice may be that of a family that share the account, or it may be one persona used by an individual with multiple personae, such as a “switch Twitter,” who maintains one account for personal posts and another for professional posts.
Description, Part 2: The Baseline Analysis
CRED+STAMP analysis would be done on a per-campaign basis. First, the general terms of the topic to be analyzed would be identified.
Assume REI is to introduce a new three-season tent called the “Hart’s Pass” with an ad campaign, and wants to track the effectiveness of the campaign in reaching users of Twitter. A CRED+STAMP analyst might track the following words and phrases “Hart’s Pass,” “Harts Pass,” Heart’s Pass,” “Hearts Pass,” “three-season,” “three season,” “three-season tent,” “three season tent,” “tent,” “REI,” “REI.com,” etc.
A baseline analysis would establish how those terms were being used on Twitter in the month before the campaign launch, providing as much additional context to them as possible. All the posts made to Twitter during the baseline period would be loaded into the CRED+STAMP database via the Twitter public timeline API.
The first parsing of the data would be to flag posts as either containing the target phrases or not. Those that don’t contain the keyword are irrelevant, or Distraction, posts. Through text data mining, the non-Distraction posts are grouped into the Commentary, Rumor, and Expression buckets. As the processes for each bucket are the same, we’ll follow an example analysis through the Commentary bucket.
First we want to establish a number representing the Commentary rating for the Site — the CS number. Like all the discrete measurements in CRED+STAMP, this needs to be an open-ended number, providing an absolute value which can be meaningfully compared to the CS numbers for any other sites.
Without the input of a trained statistician, I am hesitant to suggest how this number might actually be calculated, but for the sake of illustration we can imagine it to be a something like
X+X∗((X/T)∗100)
where
X=Number of posts labeled as 'Commentary'
and
T=Total number of sample posts
This formula will be consistent for all CRED+STAMP number pairs and is essential to understanding how CRED+STAMP can be used for comparison against a baseline, or against a sample taken from any other site. Because the number is calculated from both the total and the percentage, it provides an absolute reference for comparison.
The messages that make up the CS group will be “spidered” and those that are part of a chain of replies will be grouped as Topics. Each standalone message will be a topic of one. Each topic gets a DT score, using the same formula used to calculate the CS score except now
X=Number of topics labeled as 'Commentary'
Now each post is grouped by posting Account and given an initial CA score where
X=Number of accounts labeled as 'Commentary'
Following that, each Message is given an initial TM score where
X=Number of messages labeled as 'Commentary'
Finally, each message is broken down into constituent Phrases, and given a CP score where
X=Number of phrases labelled as 'Commentary'
The process as described is then repeated for the remaining three tonal ranges, R, E, and D.
By this point, it’s clear that the analysis needs to account for the very likely possibility that whatever unit of amplitude is being measured (S, T, A, M, or P) may very well contain within it text that fits into multiple tonal ranges (C, R, E, or D). How best to handle this mathematically is best left to an expert in the field, but we can visualize the solution by picturing the X-axis of our graph curving around on the Z-axis so that the tonal ranges overlap.
Illustration 3: Overlapping CRED pitches roughly illustrate how at each STAMP level, quanta of data may contain more than one pitch. Description, Part 3: Setting Goals
With the baseline set, it becomes possible to set some goals. Lets look at setting goals for the target phrase “Hart’s Pass” in our hypothetical example.
As a place location well-known to campers and hikers in Washington State, we might find “Hart’s Pass” to be mentioned occasionally in our baseline sample, although a search during the drafting of this paper returned no results. (“No results for hart’s pass,” 2008) For the purposes of this campaign, all these references are Distractions. But looking at the context in which the phrase appears, we can see the difference between baseline use of the phrase and the way we want the campaign to influence the use of the phrase.
A Distraction use of the phrase as a place name might use it as the subject of verbs such as “to go” or “to visit”: “We went to Hart’s Pass…”, “My friend visited the Hart’s Pass fire lookout…” As each CRED+STAMP analysis is customized to a campaign, identifying these uses and feeding them back into the text data mining engine as Distraction examples can help in fine-tuning the algorithm.
For this campaign there are two sequential goals. The first is to introduce “buzz” around the use of the phrase “Hart’s Pass” in the context of a new tent, beginning two months before the tent is available for purchase. This means we are looking for increases in tonal range “R” for Rumor, reflecting contextual uses like “I saw an ad for the REI Hart’s Pass tent…” or “The Hart’s Pass looks interesting…”
The second goal is to monitor reaction to the product after it becomes available for sale. For this, we are looking for increases in tonal range “C” for Commentary, reflecting contextual uses like “I bought the Hart’s Pass…”, “I used the Hart’s Pass…”, “Looked at the Hart’s Pass in the store, but…”
This illustration of the use of the discrete CRED+STAMP numbers shows how some measure of campaign effectiveness can be gathered by monitoring and data mining social media sources. It is predicated on the assumption that having your brand or products talked about is a predictor of increased conversion. As with choosing the right keywords to purchase in Google’s pay-for-performance advertising programs, knowing how to stoke the conversation is up to the advertiser or marketer. All CRED+STAMP can do is focus the thinking around goals, and report back on how well they were achieved.
Project Plan
I’ve sought to envision a metric that would would have practical value, using a methodology of my own devising. Input from specialists in data mining and statistics, as well as users of analytics in business would be helpful in further determining its usefulness and shaping how it might be developed. This would be achieved by conducting interviews and sharing this proposal for comment. I envision that I could complete this phase alone.
Pending more input from specialists, the best approach to actual development of CRED+STAMP would be to pursue it as a cross-departmental academic project at The University of Washington. A provisional team plan of graduate students would be:
- 1 Project Manager
- 2 Computer Science majors
- 1 Linguistics Major
- 1 Statistics Major
Technical resource needs would be minimal, but would include access to a networked computer with significant amounts of memory and a good processing speed. It is possible that open source data mining solutions such as GATE (“GATE Home,” n.d.) or RapidMiner (“RapidMiner Community Edition,” n.d.) would provide acceptable starting points for the project, possibly influencing decisions about computing capacity and operating system.
Budget
The initial phase of further information gathering could be undertaken as independent study at no additional costs above those already incurred for academic enrollment.
The estimated budget for the development phase of the project is $2000 for computer hardware. Staffing will be student labor working for credit, and network connectivity will be provided by piggybacking on existing systems.
Personnel Qualifications
My interest and qualification in solving the problem of value metrics for social media is rooted in my 18 years of professional media experience. After receiving my BA from The Evergreen State College, with an emphasis in Media and Performance Communications, I moved to Seattle and worked in sound, film, and live theater. By 1993, I was working on interactive CD-ROM projects, active in the local online Bulletin Board Service (BBS) community, and beginning to explore the possibilities of hypertext just as the Internet entered the graphical era.
In 1996, I began working with Starwave Corp. At the time, CEO Mike Slade described his own experience with Starwave this way:
“When I joined Starwave, I viewed it as a software-company type of management challenge. As I’ve gotten educated, mostly by the people I’ve hired, I’ve realized that Starwave is really a next-generation media company.” (Malone, M. S., 1996)
As a Content Producer at Starwave, I was expected to use the company’s analytics tools to measure the performance of my work and guide my decision-making. Later, after Starwave was purchased by Disney and merged with Infoseek, one of the leading search engines of the day, I dove deeper into user search behavior and used it to inform content programming choices we made on Disney’s GO.com portal launch.
In 1999, having developed an interest in indexing, linking, and publishing information on the Web, I left Disney for a small startup in San Francisco called Invisible Worlds. One of the company founders was the editor of the specifications for POP and SMTP, the protocols that e-mail uses, and the focus was on data mining, and intelligent tagging with XML. Invisible Worlds didn’t succeed as a company, but it was an extremely educational experience on both the technical and business sides.
Following my return to Seattle in 2001, I worked variously as a freelancer and contractor, finally settling into a role as the Online Production Supervisor at REI, managing a team of up to six Technical Producers. There, I advocated for the creation of dedicated Information Architect roles, data-driven decision making, flexible publishing models, the adoption of social media as a natural channel for customer support, and reorganization to follow media staffing best practices. I even succeeded in getting some of it implemented.
In early 2008, in recognition of the nature that much of the work I had undertaken didn’t fit in my job description, my title was changed to Information Architecture Manager. After leading a very successful “frame-off” rebuilding and redesign of the REI Web site, I chose to leave the company in pursuit of other opportunities.
Currently, I am contracting at POP, an award-winning interactive agency, while I work towards my Master of Communication in Digital Media.
Bibliography 1: Sources Referenced in this Proposal
Agichtein, E., Castillo, C., Donato, D., Gionis, A., & Mishne, G., (2008, 2 11-12). Finding High-Quality Content in Social Media . Eugene Agichtein’s Publications, Retrieved November 17, 2008, from http://www.mathcs.emory.edu/~eugene/papers/wsdm2008quality.pdf
The quality of user-generated content varies drastically from excellent to abuse and spam. As the availability of such content increases, the task of identifying high-quality content in sites based on user contributions—social media sites—becomes increasingly important.
Dabitch, (2004, 10 27). Banner ads tenth birthday!. Retrieved November 30, 2008, from AdLand Web site: http://commercial-archive.com/node/114815
Deragon, Jay (2008, 10 30). Is Social Media ROI Important?. Retrieved November 28, 2008, from pitchengine Web site: http://mediapitch.ning.com/profiles/blogs/1625905:BlogPost:7636
Mr. Deragon has over 25 years experience in working as a strategic consultant for numerous industries such as wireless, entertainment, capital markets and technology.
Falls, J (2008, 10 28 ). What Is The ROI For Social Media?. Retrieved November 28, 2008, from Social Media Explorer Web site: http://www.socialmediaexplorer.com/2008/10/28/what-is-the-roi-for-social-media/
Jason Falls is the director of social media for Doe-Anderson, a brand-building agency in Louisville, Ky.
GATE Home. (n.d.) GATE Home. Retrieved November 28, 2008, from GATE, A General Architecture for Text Engineering Web site: http://gate.ac.uk/
The home page of the GATE open source text data mining application.
John Wanamaker Quotes. (2008, 10 1). John Wanamaker Quotes. Retrieved November 30, 2008, from ThinkExist.com Quotations Web site: http://thinkexist.com/quotes/john_wanamaker/2.html
A non-authoritative listing of quotations without attribution. At the time of retrieval, it listed two different versions of the Wanamaker quote in question, demonstrating its folkloric status.
No Results for Hart’s Pass. (2008, 11 28 ). No results for hart’s pass. Retrieved November 28, 2008, from twitter Web site: http://search.twitter.com/search?q=hart%27s+pass
A search for the phrase “hart’s pass” on twitter that returned no results.
O’Connor, C. (2005, September 19). Behind Search Engine Craze Lies Advertising. Investment Dealers’ Digest, 71(35), 7-8. Retrieved November 29, 2008, from Business Source Complete database.
A slightly-dated article on search engine advertising intended for a general business audience.
Pereira, F. (2008, May 24). Social media resistance survey – Preliminary report. Retrieved October 7, 2008, from http://www.communitelligence.com/blps/article.cfm?weblog=73&page=537
The Institute for Communication Technology Management’s Dr. Pereira conducted an original survey of business professionals about the resistance to social media at their companies. The findings were presented at the Executing Social Media conference in 2008. Although the powerpoint presentation available online provides little information on methodology, the data is thought-provoking.
Performance-based advertising. (2008, November 27). Performance-based advertising. Retrieved 03:15, November 28, 2008, from http://en.wikipedia.org/w/index.php?title=Performance-based_advertising&oldid=254498839
Wikipedia entry provides general background on the subject.
RapidMiner Community Edition.(n.d.). RapidMiner Community Edition. Retrieved November 30, 2008, from Rapid-I: Report the Future Web site: http://rapid-i.com/content/blogcategory/38/69/
Home page for the Rapid-I company’s open-source data mining application.
Malone, M. S. (1996, 10). Starwave Takes the Web … (Seriously) . Retrieved November 30, 2008, from Fast Company Web site: http://www.fastcompany.com/magazine/05/starwave.html
A Fast Company cover story on Starwave Corp.
The North Face.. SUCKS. (2005, 3 28). The North Face.. SUCKS. Retrieved November 28, 2008, from IH8MUD.com Web site: http://forum.ih8mud.com/chit-chat-section/39601-north-face-sucks.html
Off-road enthusiasts openly express their opinions about camping gear on a forum.
Bibliography 2: Additional Source Material
Facebook for suits. (2008, September 27). Economist, Retrieved October 6, 2008, from Academic Search Complete database.
The Economist reports on LinkedIn and Xing, two social media networks geared towards professionals. It further touches on possibilities for Facebook or other networks to displace LinkedIn and Xing.
Global research from Trampoline Systems reveals 88 per cent of businesses ready to deploy enterprise social networking (2008, June 24). Retrieved October 6, 2008, from http://www.trampolinesystems.com/news/press+release/14
UK-based vendor of social computing software Trampoline Systems released the results of a survey of executives in the US and UK on their views regarding the internal use of social media software to impove communications.
Herzog, A. (2008, July 29). Wary, Risky Business Tackles Social Media. Retrieved October 5, 2008, from http://www.ariwriter.com/2008/07/wary-risky-business-tackles-social.html
A roundup of data from a number of studies and articles, which are promising paths for further research.
Jones, A. (2008). Studies Suggest That Enterprise Social Media Will Change the Face of Business. Retrieved October 7, 2008, from http://www.econtentmag.com/Articles/News/News-Feature/Studies-Suggest-That-Enterprise-Social-Media-Will–Change-the-Face-of-Business-50353.htm
A review of studies on social media adoption by Trampoline Systems, AIIM, and Forrester this overview provides useful information, plus new directions for research information.
Kho, N. D. (2008, April 4). B2B Gets Social Media. Retrieved October 5, 2008, from Communication & Mass Media Complete database.
Much discussion of social media focuses on business-to-consumer and internal uses. This article provides another perspective by surveying use of social media in business-to-business (B2B) interactions.
Livingston, G. (2007, October 8). Corporations Have Anti-Social Media Cultures. Retrieved October 7, 2008, from http://www.livingstonbuzz.com/2007/10/08/corporations-have-anti-social-cultures/
This post on Livingston’s blog makes a number of assertions about the reasons businesses may resist social media, many linked to other blogs. As part of a shared narrative amongst social media practitioners, including quantative data from some sources, his opinions reinforce and clarify issues.
Stoller, J. (2008, May). Online social networking arrives in the office. CMA Management, 82(3), 46-47. Retrieved October 6, 2008, from MasterFILE Premier database.
Attitudes towards privacy and legality of social media in business are briefly explored.
The Digital Millennials (2006). Retrieved October 7, 2008, from http://www.resource.com/thoughtleadership/gen_y.aspx (Please see following notation.)
The Digital Millennials: R U Ready? (2006). Retrieved October 7, 2008, from http://www.myspace.com/ruready4us
Kelly Mooney, President and Chief Experience Officer at Resource Interactive, compiled data from a number of sources to present a portrait for marketing professionals of the “Digital Millenial” generation, the leading edge of which is defined by those who graduated from high school in 2000. Mooney unveiled her findings as the opeining keynote at the National Retail Federation’s Shop.Org annual summit in 2006, where the audience largely consisted of members of the “Baby Boom” generation. Her slideshow from the presentation is hosted on a My Space page and additional content has been published on the Resource Interactive site.
Wang, L., Baker, J., Wagner, J., & Wakefield, K. (2007, July). Can a Retail Web Site Be Social?. Journal of Marketing, 71(3), 143-157. Retrieved October 6, 2008, from Communication & Mass Media Complete database.
Focusing on marketing retail customer interactions through social media, this article mixes both original research into user experience and summary data from other studies. The bibliography sources are worth further examination.
Post a Comment