Paul Blase is a co-founder of Speciate AI, a spin-out from tronc, a newspaper and online media company based in Chicago. Blase was formerly the managing of tronc’s AI and data solutions division responsible for building a portfolio of businesses and products that monetize data. Prior to that he led PwC’s Global & US Advisory Analytics Practice on applications of advanced analytics and data techniques. He is a frequent speaker and writer, with publications including the Wall Street Journal, Bloomberg and Forbes. He recently spoke with AI Trends editor John P. Desmond.

Q. What’s the idea behind Speciate AI?

The idea is that if you look at most of the data being produced today, well over 90% is in the form of what we would call unstructured data. Videos, images, text reviews, audio recordings. And if you look at the billions of dollars that companies spend on data and analytics it’s almost the opposite: 90+% is spent on how they manage their structured data, data warehouses and data marts and the like. And so our mission is to tap value in unstructured data because that’s the future. When we apply AI to that data , we think we can deliver anywhere from 100 to 1,000 times the analysis for one- tenth the cost. So it has massive potential. And based on my experience leading PWC’s global and U.S. analytics practice, companies are really just getting started on this journey.

Q. How do you incorporate AI into your products and services?

Our first area of focus is reinventing how customer feedback is collected and used to develop better products. The way AI is involved is twofold. First we are providing a solution that we call Signals that essentially takes all of the data, mostly media or text referencing products in reviews, be they consumer reviews or expert reviews and performing our Feature Sentiment analysis. We create a scorecard for their at the feature-level and allow it to be easily compared to competitor products. This gives the developer or the market researcher scores for each feature in terms of the sentiment, how positive or negative customers are about that particular feature and the frequency of comment about that feature.

So for example, if we’re talking about wireless speakers, one of the features might be sound quality. In that case, we’d use natural language processing to millions of unstructured data points to determine what do consumers really think about the sound quality? We can also measure the frequency of discussion about sound quality, i.e., how much people talk about it in all that unstructured data, we would provide a count as to the number of times it’s mentioned. So you get a sense of the relative importance of sound quality versus, let’s say Bluetooth connectivity based on the amount of conversation. We also have an application called Quality Radar, which is giving them the ability to anticipate emerging quality issues. So if the product breaks or it overheats, product and brand managers want to get ahead of it before they become major brand redefining issues in a negative way.

We also have a tool called Price Monitor, which is pulling pricing data as well as promotions from across different channels, so that a company and a product developer can see how their product price and promotion compares to competitors in their category. The difference between what we provide from traditional research providers is ours is real-time or near real-time, not lagged by several weeks. So it gives the product developer, or marketer, the ability to make more timely changes.

This is a two-sided coin. The side I just described is how companies can mine customer feedback to make better products. But we also see the opportunity to reinvent how customers can provide feedback by removing the barriers that prevent a lot of us as consumers from doing so. We have all experienced situations where you’re about to hang up and finish a customer service call, then they ask if you want to respond to a survey afterwards. You’ve just spent a bunch of time and the answer is usually no. Or we’ve all received requests to fill out extensive surveys in the mail or via email, and you open it and it says, ”This is may take 25 to 30 minutes.” The most frequent reaction is to close.

So there’s a problem and we think we can solve it with AI. Companies aren’t getting the feedback they want from customers. We have found that customers want to provide feedback but it’s too hard, it takes too much time. They don’t have the right incentives. And so, we’ve developed an app we call U SAiY.

The idea is to give consumers an app they can use anytime they want, whenever they feel like to provide feedback. They can say the product name; they can send an image of, let’s say, a broken product. Or they could send a text message or record an audio file about what’s bothering them. Or they could send a video. We then use AI to break down the information about the company, the product and the topic of feedback. We structure it and provide it back to the company.

That’s a win-win because it provides companies better insights and reduces the barrier for consumers; it makes it a bit more practical, less time-consuming. You might even argue fun. I mean who doesn’t love to rant usually or rave about products they use? Often it’s the top of mind comments we make about products that is how we really feel, not the responses we provide when the companies want us to provide the feedback.

Q. Sounds good. What industries are you targeting?

We are starting with consumer electronics for a couple of reasons. One is there’s a lot of new product and it’s really about the design and features. For product developers, figuring out which features are important to customers, which ones they should prioritize for investment and then how to message those features in a way that gets across the point of why their product is different than another, is the heart of the matter.

They are also products you can see. One of our value propositions and ways we’re applying AI is, we’re able to take images of products from Instagram pages or Facebook or other channels and apply image recognition algorithms to help pull data out of the images that is very informative to a market researcher or product developer.

For example, a company developing wireless speakers might be interested in understanding whether their product is more often used in groups outdoors or by individuals indoors? They can ask people that in surveys, but the pictures are the actual behavioral data, how the product is being used in the market. And through image analysis, we can extract insights to help them understand if the product is being used as intended and then if not, what to do about it. That is very hard to do with traditional survey methods. In essence, we’re giving them a much more scalable way, a digital way to do ethnographic research.

After consumer electronics, we’re going to expand into hospitality and consumer packaged goods. We’ve also been talking to several financial services companies, automotive companies, as well as medical device and healthcare companies.

Q. What label would you put on the market that you’re in?

When we think about our market our primary users are market researchers, product developers, and leads of customer experience interested in understanding what customers need, what is bothering them and providing the right products and experiences to satisfy them . Three traditional approaches used to figure this out are surveys, net promoter scores and focus groups.

We are in essence more insightful than traditional surveys. When you get a 1 through 10 score about how people think about a product feature like sound quality, you’re left with answering the question why? We provide you our feature sentiment scores and access to the underlying text from the reviews or media that answer the question why.

We also are more actionable than Net Promoter Scores (NPS), which are frequently used by companies as a measure of how they’re being received by customers. The frustration we’ve heard that executives have with NPS is they’re not actionable. So if they get a score that’s negative, they don’t really know what to do about it. That is an issue when 20% of your compensation is based on NPS. They don’t know why it’s negative. And so, what we’re giving them the ability to do is take action because we can pinpoint exactly what is negative and why.

Also we are more effective than focus groups. Companies spend tons of money on focus groups to bring together groups of people to discuss what they think about a new product and get their reaction or identify opportunities. That can cost anywhere from $10,000 to $15,000 per focus group. What we’re providing with our Yap! app is the ability to record that type of data, both through private studies or the multi-media feedback we’re receiving from consumers in the market. This provides data without the bias inherent in the focus group format in a more cost-effective scalable way. We do it all digitally through video.

The last thing we do is make market researchers and business analysts more productive. Many companies we speak with have them read through all the data about a product or product portfolio in the market and try to make sense out of all the unstructured data. We’re giving them a tool that reduces the amount of time they have to spend doing it by 90%. And so, what we’re also doing is helping companies’ business analysts be much more productive at analyzing unstructured data where there is more insight about their products and opportunities.

Q. Would the label on your market be intelligent market research?

Yes I think that is an accurate characterization. We are using intelligence to amplify the signals in the market data about a company’s product and service, to make their market research and product development much more efficient and effective than it would be if you had business analysts essentially trying to do all the same things that we’re doing with intelligence.

Q. What in your opinion is the outlook for the newspaper business?

The value of journalists and high quality journalism is more important than ever to make sure that, you know, people are really getting an understanding of the truth and all sides of the story. The one thing I will say about the media industry is that journalists, no matter what happens, are integral to the functioning of a free society and they need to be supported.

Q. Is there anything you would like to add or emphasize?

I would just emphasize that as companies are embarking on their AI journeys to set the bar high. In the use cases I’ve talked about that use our Product Signal solution, or our U SAiY solution to reinvent customer feedback, the bars should be set at doing anywhere from 100 to 1,000 times the analysis and analysis output at one-tenth the cost. And when you think about all the unstructured data in the market about a company or their products, or the unstructured data that companies produce be it from their own email feedback or call center feedback or sensors on products, that type of return is possible for companies in artificial intelligence. So don’t set the bar too low.

For more information, go to Paul Blase can be reached at

Source link
thanks you RSS link


Please enter your comment!
Please enter your name here