The current state of text mining/analytics marketing?
One thing that didn’t go so well at the Text Analytics Summit was the marketing panel. Indeed, when we wracked our brains afterward, Mary Crissey (who was on the panel) and I could only think of a single observation that was actually made about marketing. Namely, she referred to a core truth of marketing: Just selling features doesn’t work (nobody cares). Just selling benefits doesn’t work (you’re not differentiated). What you have to do is sell the connection between your features and desirable benefits.
So I’m going to try to gather some useful observations on marketing here, filling the gap that the panel left. Key questions I’d love input on include:
1. Which feature-benefit connections do you see customers easily accepting?
2. Which feature-benefit connections is it harder to get them to believe?
3. How are customers defining text analytics market segments?
4. What do they see as the key issues in each segement?
5. Which application areas are showing growth even beyond that of the market overall?
I’m particularly interested in comments from the larger vendors that are selling into multiple parts of the text mining and text analytics market. But everybody else’s input would be warmly appreciated too.
The comment thread to this post is open for business!
Comments
6 Responses to “The current state of text mining/analytics marketing?”
Leave a Reply
1. Which feature-benefit connections do you see customers easily accepting?
SAS text mining customers tells us they really like the exploratory analyses capabilities of our software. Frequency counts of key words are widely accepted as a first step in reporting data. The most appealing features are those that deliver “unexpected” value — benefits resulting from finding connections between individual words. Everyone accepts the fact that words carry information and are eager to operate within a unified BI approach, but most remain leery of complicated technology that might be difficult to learn. That’s why interactive interfaces with flexible diagrams and a variety of graphical or visual options are welcomed enhancements to text analytics.
2. Which feature-benefit connections is it harder to get them to believe? Its not so much an issue of don’t believe, but often the text is gathered in a happenstance fashion (with no real purpose) so it cannot be easily integrated with related structured data such as demographic information and purchasing behavior for a well rounded view of the customer.
SAS delivers our text analytic technologies bundled in with SAS Enterprise Miner as an add on to our predictive analytical data mining solution. There is far more quantifiable benefit to using text for prediction of such things as satisfaction & churn. Yet there is a certain amount of intellectual curiosity or appreciation for advanced analytics that must be in place before Mr. or Ms general business guy will be willing to start asking the tough questions and explore the underlying WHYs of his daily adventures.
The data doesn’t always exist in a format ready for historical trend analysis or predictive forecasting so careful preprocessing is essential. Sometimes additional data values or input points must be added as the text data is captured and stored — and it is not easy to have customers change internal processes to improve results.
3. How are customers defining text analytics market segments? For SAS Text Miner, our biggest segments include financial services, manufacturing, public sector, Telco, retail.
4. What do they see as the key issues in each segment? CRM scenarios are common in our Text mining success stories — especially as businesses seek the competitive edge in today’s dynamic marketplace.
5. Which application areas are showing growth even beyond that of the market overall? Early warning for manufacturing especially in warranty scenarios, insurance fraud for financial, churn for Telco.
Thanks, Mary! Interesting point about the “exploration” focus. I think linkage analysis is generally at that level. E.g., I’ve posted elsewhere about Cogito, and from what I can tell that’s pretty much how their technology is used too.
Actually, for #3, I was asking about the definitional wars — text mining vs. knowledge extraction vs. knowledge management vs. whatever.
Curt Monash
And do you have any specific examples you’d care to share of the unexpected discoveries?
Curt Monash
Olivier Jouve emailed me to say he was having difficulty posting the following comment. CAM
————————————-
1. Which feature-benefit connections do you see customers easily accepting?
Obviously the main trend is Sentiment analysis and Fact/Relationship extraction.
Customers want to go beyond term or even concept extraction which provide only little value when merged with predictive models.
I agree with Mary. Our customers are not linguists or text analytics experts… so they accept only solution that can be easily customized and maintained.
2. Which feature-benefit connections is it harder to get them to believe?
That a standalone text mining application will fix their business issue.
3. How are customers defining text analytics market segments? Actually, for #3, I was asking about the definitional wars text mining vs. knowledge extraction vs. knowledge management vs. whatever
I don’t think customers really mind about those definitions. They have tons of text in their databases/blogs/etc and want to get value from them.
Most of them have now understood that a search engine approach is not relevant for discovering and understanding customers’ behaviour.
About segments — biggest segments include Market Research, FBI (Finance/Bank/Insurance), Telco, Public Sector, Life Sciences, Manufacturing, Higher Education.
Main business solutions are around Marketing Effectiveness, Fraud Detection and Prevention, Risk Management, Enterprise Feedback Management, Public Health and Safety, Administration and Institutional Research, Scientific Research.
5. Which application areas are showing growth even beyond that of the market overall?
Enterprise Feedback Management (including insights coming from external/internal blogs), Fraud Detection (claim analysis).
[…] We have discussion going in the comment threads to a couple of posts. Mary Crissey of SAS and Olivier Jouve of SPSS responded to this one on sales/marketing of text analytics, and customer response. (Mary: There are followup queries for you.) And Olivier and Glenn Fannick of Factiva offered appropriate responses to my half-serious comment about the French presence in text analytics. […]
[…] If you have any thoughts on these subjects, please share them in the comment thread! (One catch: Comment spam is really bad these days, often overflowing Akismet’s measly 150 message spam buffer. If your post somehow gets lost in the trash, I apologize deeply in advance and implore you to contact me directly.) Please also see last year’s post-Summit thread on text analytics marketing, and this observation on major text mining applications. […]