<< by on November 12th, 2009
This session was full of tools and thoughts on how to effectively measure social media and online reputation.
John Marshall of MarketMotive was first up and discussed web analytics and social media. John chose to focus on referers, sample size and revenue attribution.
John said that web analytics packages grew because of the referer report. However, referers are an “act of grace” (or luck) that it exists. You don’t really have a God-given right to the data. Referers have problems. For instance, IE and Firefox, as browsers, you get referers. If it’s a url shortener like bit.ly, you will get a referer. However, if the customer’s interaction happens on installed software, like TweetDeck, there is NO referrer. Same is true with Outlook as well. If an RSS reader is using installed software, again, no referer.
What this means is that we end up with “unknown unknowns” in the data. Lots of big holes.
The second problem is low sample size or click volume. For instance going from 8 clicks to 12 clicks is NOT a 50% improvement. The numbers are just too low to be meaningful.
The third problem is revenue attribution. John said that even if you are tagging your URLs, social still has a weak intent to buy, while search has a strong intent to buy. Don’t expect a revenue-based ROI report for social media. Find other meaningful ways to measure social media.
Because there are multiple failure points that can happen, it’s nearly impossible to really measure true ROI from social media.
Revenue attribution solutions include commercial tools like ClearSaleing or use the Google Analytics custom variables, a new feature in beta.
What would Van Halen do?
John said that Van Halen is famous for the “no-brown m&m’s” rider in their contracts — meaning that they wanted M&M’s, but no brown ones. Van Halen’s equipment and setup was very complex, and the roadies needed to read a complex manual. To test if they were detail oriented, a question asked of them was if there had been brown M&M’s in the snacks.
“When you’re going to lose anyway, lose early.” Don’t waste time on a bunch of things that won’t work!
Adam Proehl was up next and discussed “Getting Beyond the Geeky Numbers”.
Adam started with tools — the bests tools can still build a bad house if you don’t have a plan. So first, understand your business objectives and your customers’ objectives. To understand your business objectives, you have to understand your corporate goals. Then set objectives for online marketing and measurements. The rules:
- Be specific
- Be measurable
- Be realistic
- Be relevant
- Set a timeframe
“We want more leads” is NOT a business objective. Be more specific!
Also understand your customers’ objectives — why did they visit the site? what are they doing on the site? how did they get there? To get started, ask and answer questions. For example — “We want to be on Twitter to ______.” Are your customers/prospects even ON Twitter?
Next, establish a methodology:
- Site objectives
What tools to use and what to report? Examples:
- Mentions: Radian6, HubSpot, Raven
- Sentiment: Addictomatic
- Share of Voice: ScoutLabs, Keotag, Trackur
- Influencers: Web analytics and monitoring tools
- Traffic: Web analytics
The numbers also never give the full story. For instance, with sentiment, if someone says “Successful call on 3rd try, nice network AT&T!” — the “nice” term is being used sarcastically, making it more difficult to measure sentiment accurately.
Some useful and free tools to use:
- Addict-o-matic: generates a dashboard for social media
- ShareThis: Integrates with Google Analytics and Omniture
- Snip-n-Tag: Firefox add-on; attaches a Google campaign ID to your shortened URL — NICE!
- GA? : Firefox add-on; shows Google analytics problems
- Better Google Analytics: Firefox add-on; see social media submission pages on Digg and the like — also REALLY NICE.
- Enhanced Google Analytics: Firefox add-on; detects unusual spikes in traffic
- Outbound link tracking from Google Analytics
- Excellent Analytics: Excel plug-in
- Site Scan GA: scans your site if you have Google Analytics code issues
- Tweeting too Hard: pokes fun at Twitter posts
No one tool can measure everything — understand the limitations of each.
Last up was Dan Zarrella of HubSpot discussing retweets as a quality signal. Web search is too slow to determine quality. Social media produces a high quality signal quickly, but many are small and niche. Social networks have challenges, such as Facebook, which is too closed for real API access.
So what’s the best solution? Look at retweets. Retweets often have links in there, so likely they will be a vote of popularity for search engines. Many of the accounts in Twitter are not complete or filled out.
So what algorithm can make RTs a good quality signal? Divide the number of retweets by the number of followers of the retweeting users. Example: (retweets)/(followers of those that retweeted)
Dan also mentioned TweeterAuthority of the users that retweeted you. Also the words in the retweet may be important.