Sites like
Yelp and Local.Yahoo.com are a rich trove of information for businesses and organizations. The problem is making sense of it all. If you've got a high-profile business or organization (and if your marketing campaign is working), you'll probably be struggling just to keep from being washed away by the proverbial "firehose of information."
That's good news. It means your marketing is working. It means people are talking about you, your business, or your organization. Even if the word is bad news, you need to know this. And the more you know about where and how it's "bad," the more effectively you can respond to make it "good."
The bad news is you're still going to have to find a way to gather the information in a way that doesn't involve hours of tedious "cutting and pasting," collates it so you can organize it, and then makes it useable for some kind of analytics.
More good news
I've been doing some work with
McAllister Opinion Research on this challenge. McAllister recently had a large project that involved searching through 1000s of online reports – many generated by Yelp and Local.Yahoo, but also , Facebook, and myriad lesser or more geo-local consumer reporting sites. We were able to use scraper software to do the heavy lifting – going to sites like Facebook, Twitter, Yelp, etc and pulling comments along with dates and other relevant information into a vast spreadsheet.
In the end, our scrapes gave us close to 8000 datum – words, sentences, paragraphs, sometimes 3-4 paragraphs each – related to our initial search. How to deal with that? How to make it more than 1000s upon 1000s of words?
SayZu - visuals and analytics
That's where SayZu comes in. It's a way to visualize data. At first glance, you'll see a
word cloud. Word clouds are cool tools: they let you see the most used words, hence give you a sense of what the conversation is, without having to listen to every word.
There are a number of things that make SayZu different than other word cloud software. The biggest is the ability to drill down into the cloud, and the dynamic aspects of the clouds. It also contains a number of analytic tools that make it very fun to work with.
SayZu is still in beta. I've been playing and working with it for a few months, doing little demos, applying to big projects like the one with McAllister Opinion Research. Here's an example of a demo project that used scraper software with SayZu to make sense of a different "firehose" of information: the active chatter about "coffee" in the highly competitive Vancouver coffee market.
Coffee, Vancouver, and SayZu
The first task was to find a source. We chose Yelp.com (and later, Twitter, for a comparison). Second task: identify several brands for comparison purposes. We chose Starbucks, Blenz, Caffe Artigiano, and Tim Hortons.
Here's what the individual SayZu clouds looked like (top 200 words as at Feb 13, 2011 on Yelp.com):
Tim Hortons (60+ records/comments)
Caffe Artigiano (140+ records/comments)
Blenz (80+ records/comments)
Starbucks (80+ records/comments)
A general scrape of Yelp.com for the keyword "coffee," and including the previous scrapes for Starbucks, Tim Hortons, Caffe Artigiano, and Blenz. Date: February 13, 2011. (approx 800 records/comments)
A scrape of Twitter for recent (Feb 18-19) tweets with the keyword "coffee," with Vancouver at the centre of a 80km radius circle. Date: February 19, 2011. (500+ records/tweets)
Interactive and dynamic
These static images give you a sense of what kinds of things are being said. What you won't see here is the interactive and dynamic aspects of the SayZu cloud.
As an example of SayZu interactivity, visit the
Tim Hortons' SayZu cloud online. Click on a word. See where it takes you.... The cloud gives you context (ie. the most frequent words being used are the largest, words used together are clustered); drilling into the cloud gives you the specifics.
As I said, SayZu is in beta. Some bugs are not quite worked out, some features still not fully implemented. If the current beta is being "good," what you'll see here is an interactive and dynamic version of the Twitter scrape. It's interactive in that you can click on the individual words and find the originating tweets; it's dynamic in that this image will change as it picks up more tweets over time, within an 80km radius of Vancouver, containing the keyword: coffee. If you're within the target zone, test it. Go to your fave coffee bar and tweet about your experience.
Curious?
The scraper / SayZu combination is a powerful tool when analyzing online conversations. We've used it on a number of sources beyond those referred to. In one recent instance, we scraped a 100+ comments off an online news post to give a report on how a political candidate was faring. We can use the same tools to make sense of the online conversation on blogs, online newspaper articles, text-rich open-ended surveys (online and off-line), etc. I've included some links below on some other demos I've run using the SayZu / scraper combination. Have a look. What I'm seeing is that these tools apply not only to the types of campaigns we're working on – from political to customer service – but also to things like fundraising, product and services marketing, event promotion / follow-up, to name a few.
The SayZu beta is available for trial download at SayZu.com. Check it out. Let me know what you think. Tell me about some new uses you've found for it. (And yes, it is a Windows app, so I'm currrently running my old PC next to my Mac... perhaps one day they'll port it to my main machine. In the meantime I'm considering rebooting my VMware virtual windows inside my Mac...)
As an aside... Twitter and doing business
Yelp is a recognized source for consumer information. The fact that we pulled so much data with little effort using only the keyword "coffee" is a testament to its popularity as a way for customers to have their say.
But look at the numbers: approximately 800 datum pulled from Yelp on "coffee" in Vancouver, from several months of posts; approximately 500 datum pulled from Twitter on "coffee" in the same region, within 48 hours.
If I were a business owner I'd be signing up for a Twitter account
muy pronto. Then I'd be proclaiming the fact loudly. I'd be trumpeting my Twitter account as the most efficient way to communicate with my business – about the positives and the negatives, and everything in between. Third party sites are nice because they establish reputation. But nothing beats the customer telling me directly about his or her experience. And when the kudos (or complaints) become a firehose.... time for some SayZu!
hanspetermeyer
19 February 2011
