What's Wrong with the Brand Monitoring Leaders?

Breaking it down to the bare bones:

1) All of the text mining leaders that cater to brand monitoring (not the BI or BAM guys), are, without exception, focused on sentiment scoring - the weakest of all metrics. Other than vendors like Buzzlogic, which takes an interventionist approach to marketing support via the 'influence' channel, sentiment only got a foothold because it is deployable via HSD's, hand scored dictionaries.

2) To understand why this is so, we must realize that these very early sector leaders specializing in Brand Monitoring of public corpora are all venture backed. This private equity binding imposes a 'time to market' condition. Therefore, the earliest ventures went with the best understood methods that seemed demonstrable. Other technologies for issue detection would have taken a longer development life cycle exceeding the 'time to harvest' most Series A rounds demand.

3) When one peels back the covers from sentiment analysis, it shows not only linguistic scoring weaknesses (even across large samples, although this helps), but it hardly provides the fine grained kind of qualitative analysis that would allow some type of steering or advisory when co-integrated with other variables - sales, stock levels, incentive program offers, etc.

4) Good words, bad words, neutral words - blah.....all of the HSD's used by the current 'leaders' (the term is used advisedly), are non-standard. Even test query sets run against the same data show non-repeatable results...the text scoring engines are not mature. Surprisingly, the Open Source tools are very robust (lingPipe for example.)

5) Compare this to the giants of brand monitoring, even for modest samples, and Arbitron, ACN, and Gallup always get repeatable results. The brand owners know this, and having tried some of the Buzzmetrics and Cymfony campaigns, few have opted for continuing campaigns. The returned data is just not compelling.

6) The TNS acquisition of Cymfony was a repudiation of the brand monitoring model, and the two rounds of private capital that Cymfony raised were barely covered by the sale. TNS is an advertising reporting service bureau, not a brand consultancy - it came from the opposite side of the spectrum. Cymfony could not find a buyer amongst the brand equity giants.

7)   Like lemmings, all of the sector entrants went towards this model of butting heads with the brand equity giants - creating a 'recurring campaign model', using long leads, non-reusable reporting dashboards, etc. Unsustainable.

8) The fat part of the market is in the intermediaries - the mid-market that acts as recipients of the brand decisions as they play out in the market. Retailers, distributors, small and medium multi line  durable goods purveyors.

9) These brand intermediaries were never approached by the text-mining sector entrants, because they could not be served with the type of technology that these companies had available, that is to say, sentiment scoring using non-standard dictionaries.

10) These mid-market intermediaries can be served by providing fine-grained guidance of the consumer's perception of product performance and interaction outcomes.  There are technologies that can provide this type of detection - eBay is rumored to be investigating modular dynamics to create a system to detect redress  issues.

11) The noise reduced (via statistical algorithms) and scored outcomes can be co-integrated with the retail program incentives (floor plan financing, coop advertising) and be placed within a 'decision support' framework.

12) A web-based, moderately-priced, service architecture, can be innovated with care. Such a system can generate many streams of revenue beyond subscription fees, for the use by mid-market intermediary distributors, retailers,. and someday.....consumers.