5.05.2008

Super Tool!

Ok, so Super Tool doesn't exist, but certainly analysts can dream, can't they? @JolieODell asked me "K, so if you could imagine a real doozie of a social measurement tool, what would it do?" And so this is the start of my wishlist (I'm sure I'll be adding more as I ponder it tonight at 3am):



• Sentiment analysis: Conversation by Tonality
• Conversation by Influencer-weighted Tonality
Sentiment by topic/sentiment by topic/influencer
• Distribution: Conversation by source (and lets expand sources from Blogs, News, and Forums to social nets and media utilities as well)
• Influencer Distribution view: Conversations weighted by Influencer source
• Conversation Share of Voice competitive analysis

• Include aggregation of all UGC/media tagged with tracked keywords

• Weight data by Influencer and Type

• Expand “Types” to include (in addition to Blogs, News, Forums):
– Video/ video by Influencers (I= views, link, and embed analysis)
– Photo/ photos by Influencers (I= views, link analysis)


• Influencer Identification – authority is niche-specific, not authority in general; based on frequency/relevancy to topic/keyword/industry/brand

• Site Authority/Credibility established by post to post link analysis or site to post vs. Page Rank site to site analysis for topical relevancy/authority
– Page Rank’s power law makes it difficult to see emerging influencers around given niches/topic areas and would prefer a tool whose algorithm does not base Authority solely on Page Rank
– Ability to track emerging Influencers based on frequency/topical relevance/posting frequency/comment and/or traffic velocity to post

• Influence differentiation between authors from same site/blog: Author Tracking

• Manual exports of raw data to CSV; automated and timed exports; email alerts
– Automated graph generation would be nice too—then I don’t have to do it!

• And if we’re really talking tool wishlists then a visualization tool leveraging the comprehensive API’s of popular social utilities such as Twitter, Flickr, Digg, Google Maps etc for better data visualization would be really cool…and then there is also Goog’s Social Graph API that could turn up some really interesting data for Influencer Viz (like Facebook’s Nexus only for all of the social nets that make their data public)



And please! Add what you think you would want in a Super Tool in the comments...interested to see what people think.


Note, I've demoed/used a lot of tools in this space, and all are lacking in some capacity or other, either in feature (above is really a list of desired features with a bit on the data) or the datasets are lacking or there is a lot of indexing and analysis latency....

Also, important to note that needs are based largely on objectives. I use these tools to draw up prescriptive analysis for engagement strategies...this is different that say, someone who is in search and interested in traffic source and user journey analysis.

7 comments:

Martin Edic said...

Our SM2 social media search and analytics tool does most of what you're looking for...try a freemium account and be sure to choose advanced search rather than basic.

Jolie O'Dell said...

Ok, dude, first of all, I love your analysis of how to 'fix' PageRank. Even in strict SEO terms, at can be kinda wonky from the user-searching-for-relevant-content end because the PR can depend entirely on the skill of the SEO doing the backlink campaign. Taking all the other factors into consideration is a beautiful solution.

Now, how would you parse the differences in tonality? I mean, Scoble (for example) is a huge deal, but I haven't seen a truly favorable mention or link to his blog/Twitter in a pig's age. How does a program know when we're referencing Scoble (or any company, product, site, or service) favorably or unfavorably, and to what degree and in what context? Is it a question of creating the program that can sort through the complexities of natural language?

Fascinating stuff! At some point in the future, I'd like to interview you for an article re: social media measurement. The myth that it's not measurable/measurable enough has held back ad spend and really thoughtful marketing concepting in this area for too long, and I'd like to provide the information that would change that. Perhaps you and a few others could do a kind of interview roundtable? What do you think?

MST 1948 said...

@JolieODell hey! so yeah...tonal sentiments right now are analyzed through semantic technologies that an algorithm then churns into general positive, negative, and nuetral tonalities. i would want the tool to go a step further and weight tonality not only by Influencer but by type of content...for example a "negative" video tagged "Microsoft sucks" (with a high amount of links/traffic/velocity of traffic over time) to me would need more weight than a blog post talking about how "Microsoft sucks"...but then that I suppose an analyst could just determine.

also, a lot of the tools need "teaching" as you use them to help the algorithm better understand what you deem as pos/neg/neutral...same goes for Influencer ID's, a lot of the time an analyst will need to use a tool over a period of time, manually voting up or down a particular source in order to teach the tool who you deem as an "Influencer."

And as for that roundtable...lets do it! Lets do a Seesmic or live vlog it...its ridiculously surprising the lack of knowledge C-levels have of not only tools, what they do, but of how to use them for social media (ahem, Internet!) strategy...

:)

Anonymous said...

Explain any of this to a CEO and CMO and see how long you are in their office. Keep working on it.

copydiva said...

I am always so interested to hear the desire for better tonality measures. I personally cannot see a day where tonality will improve greatly, because tonality is so subjective - you and I reading the same post (ie, no algorithm here, just human understanding) may have a different impression of tonality - and we're better at it than machines are. Sure, what could be an improvement would be something along the lines of understanding the context of tonality - for example, in a post about 3 different brands, 2 suck and 1 is ok - having a tonality engine that could parse that would be helpful.

Mattia said...

Alisa could you please explain to me what do you mean by "the Influncer?" I was reading your wish list but without this information I cannot fully undestand it. Thank you for your time.

Ed said...

Alisa - excellent wish-list for the 'super tool'.

I haven't tried many of the services out there yet, but the feedback I have had from those that have, is that it takes some serious spade work to analyse all the data and most of it has to be done by hand...unless you want to pay up for the vendors to do it. Is that near the mark in your experience?

It sounds like the tools aren't that user friendly from the agency / marketing department side of the fence. Also, so you think an 'action' feature is missing at the moment. I haven't seen too many vendors offering the ability to engage or act on the conversations that are found. Tricky with disparate platforms but would be very cool.

Anyway - nice post, thanks.