Working with data is one of the best ways I’ve found for understanding a client’s business and their customers. I’d love a chance to talk with you about finding creative data opportunities that could help you and your clients. It’d be great to pitch you “data project ideas”, the same way we pitch TV spots or Social. Maybe some of your creatives have questions about ways data could be used in an app. (Feel free to call or email me your questions. I’d love to be a knowledgeable resource for you.)
Admittedly, working with data can be complex and complicated. If what you’re wanting to do is beyond my skill set, I’d be glad to try to find someone or some technology that could help you do that, even serve as a Data Producer. Here are a few examples of Creative Data projects I’ve done or want to do:
I’m an Emmy-Winning Copywriter who has worked on every type of product in every type of media over the last two decades. I was Creative Director on the Acura National Brand (two years) and NFL on Fox (two years). I shoot video, edit, do motion graphics, and some web development. Sometimes, it takes knowing the accounts, the people, the goals, and the data resources, in order to find great Creative Data opportunities. While there are a lot of great companies that can crunch big data for insights, what I want to do is much more creative and conceptually oriented.
I've done a couple of projects with an awesome researcher who does campaign concept testing for various schools and colleges. Often using online, “Over-time” focus groups. Some of the things I've done that have been helpful were:
- Establish numerical scores based on participant’s comments, using a Sentiment Analysis algorithm. This helped her give the agency (and their client) hard numbers for each campaign and how much they were liked or disliked. Clients appreciated that the scoring was tangible, solid numbers they could point to.
- Helped her distinguish the parts of the winning campaign (from the three-minute video they tested) that seemed to have the most impact.
- Wrote suggestions (based on participant comments and years of being a creative) about possible next steps to improve the leading campaign. I also helped her write the presentation summary for the agency. It was extremely well-received.
Skype and platforms like itracks.com will also let us interview and focus group using video, which I can edit for presentations and web content.
I created a web app using the Artificial Intelligence of IBM Watson (and Node.js) that can transcribe video and audio to text with 90-95% accuracy.
Note***: In general, text mining is probably more useful when the amount of text/data you have is beyond your ability to read it and parse it. Like the next example that utilizes 10,000 Customer Comments.
A prime example of looking for “Creative Data opportunities”. I freelanced as a Copywriter on a new business pitch. The company we were pitching makes t-shirts and and other items with custom logos for group and corporate events ($200 million annual revenue). I discovered that any customer who placed an order was automatically directed to a section of the website to write a comment. Something the company was quite proud of (with a 97% approval rating). I proposed writing a web scraper in Python to scrape over five-years worth comments from their site, analyze the text to see what specifically their customers were most excited about and didn’t like. We would then use this info to write creative strategies and make ads from it. It didn’t make it into the pitch so I later did it as a simple (rough) spec project. Click here to see it, along with some examples of how we can use web scraping, and text mining algorithms to turn customer comments into rough ad strategies and concepts.
I freelanced as a Copywriter on a new business pitch for a financial product. The company we were pitching was two years late to market and was coming out with a product exactly like their competitor’s. I proposed web scraping all the online reviews for their competition’s financial product, since it had been on the market for two years. Using Sentiment Analysis to sort out the good and bad reviews. And text mining a creative SWOT analysis on them to see what the competition’s customers were liking (and not liking). I/we could then make content that promoted the positive aspects, AND make ads (and real business changes) to show that potential customers won’t have those problems with our financial product. (Did not make it into the pitch.)
Wouldn’t it be great to pitch your clients Creative Data ideas the way you pitch them creative ad concepts?
Google Analytics (and other packages) will tell you the URL of the source page that sent people to a website. Often, that referral URL is as generic as "Google Search/Organic". But sometimes those referral sites, mined creatively, can offer interesting insights and themes that can be used to create unique strategies, concepts and content.
I’m currently up for a project working with a researcher who is going to conduct 24, one-hour, qualitative video interviews via an online service called iTracks.com. I would actually conduct half the interviews. And then transcribe the video (if it isn’t done for us) using the IBM Watson app I built. I would then text mine the transcription, edit the video and help write the presentation. In many ways, this is a great project for me. Since there’s too much text to really read through and explore thoroughly. Plus, I edit video. So I can bring a lot of different talents to this project.
I studied Cinematography at UCLA. I shoot video and photography, edit in Premiere Pro and work in After Effects – even have my own cameras, lights, sound equipment. I’ve shot tons of interviews, often by myself. I find you get better answers when there isn’t a big crew around. I also like interviewing and learning about people in their environment instead of sterile focus group rooms. I’ve shot a number of “customer journeys”. But also, product teams and management teams. I find that no one person knows the whole brand story, but collectively they do. I love to make that collective story come together in an edit.
I’d love to go out to stores or dealerships on the weekends and find out how people arrived at their decisions. That’s valuable data.
In fact, I’d love for a car brand to send me around the country interviewing shoppers, buyers and sales people. I could do relevant research, create beautiful video content, and concept from what I/we discover. It’d be great for Social Media, especially Snapchat and its focus on locality and immediate experience. We could even cut instant radio from the audio. We could post stuff almost “day of”. It would probably pay for itself in terms of content, PR, dealer appreciation and the brand’s very public interest in its customers.
If you write a lot of emails or blog posts, you probably have a pretty good sense from your analytics of which ones got the most love and had the most impact. And which ones didn’t. I’d love to use text mining, and algorithms like Naïve Bayes, to get hints and guesses about what’s working and what’s not. (Here’s where my Copywriter skills come in.) Then use what I learn to write new more impactful emails. What we learn can also be applied to other areas for your client, website, brochures, banners, etc.
Let’s say you’re Subaru. And your tagline is “Love. It’s what makes a Subaru a Subaru.” We could build an app where a person takes texts/emails/posts from a prospective, new or current boyfriend/girlfriend and paste them into the app. The app would give them back a score based on a Sentiment Analysis algorithm – a rough estimate of how emotionally positive or negative that person feels about them. Essentially, a score from 1-100 about how much that person likes you.
Or, for example, I recently saw a presentation where a data scientist used a Recurrent Neural Net to build an app that will actually write a pretty good (sometimes accidentally funny) song, on any subject you choose. A song that actually rhymes. You can even telling it how much cursing you want.
Can we search for new and interesting data we could show to a potential client? Data that would show them how we are thinking about their product/service like no one else has. Perhaps build a simple R Shiny app they could interact with? In addition to working with data and writing, I shoot video, edit and do some motion graphics in After Effects. Is there a video we could shoot, customer interviews, man-on-the-street, or just some kind fun content that we could use to get a potential client to talk to you?
I recently worked as a freelance copywriter on a new business pitch for a tech company (intentional NDA vague-speak). As is often the case in technology, it’s pretty much a commodity business. They all have access to pretty much the same hardware, software and people. It would’ve been a great data project to really spend time scraping websites, blogs, video, reviews, etc., text mining and figuring out a great, differentiating brand strategy. In fact, they had hired a big brand identity agency to help them figure out their brand. Only that wouldn’t be ready for months.
What they needed from us was an interim brand campaign that converted. So I wrote strategies and concepts based on quite a bit of technical knowledge I have in that area. But, in my constant search for “Creative Data” opportunities, I discovered a dozen nicely shot, but boring, customer testimonial videos. Many of them viewed by only handful of people. As I talk about in another section, video IS data. So as part of the pitch, I used my editing and motion graphic skills to turn about 25 minutes of video into a 60 second, very strategic spot. I’m proud to say, we won the pitch.
We live in an era where companies can now deconstruct, copy, and reconstruct a rival’s competitive advantage in less than six months. For many brands, the idea of having a “Unique Selling Proposition” is quickly becoming a quaint idea from the past. That means that what you say (Data/Information), and how you say it (Creative) has become both more difficult and more important.
On the bright side, the world has also become one big digital focus group. With millions of people offering terabytes of opinions each day in the form of blogs, tweets, posts, YouTube videos, Snapchat stories, podcasts, digital books, song lyrics, and emojis. We tend to think of data as numbers and stats but 80% of the Internet, the world’s largest database, is unstructured text.
Sure we can Google our way through a lot of that to arrive at creative strategies and concepts. And we should. It just doesn’t scale very well. At some point, the mind blurs.
Fortunately, computers don’t get bored. Text Mining can let us work with 10 to 50 to 100 times more information and pitch us notions that surprise us, that are a lot more like the form we want to wind up with – 3 to 5 word phrases that are more like headlines or taglines. So if you have a vague intuition about what might work, web scraping, text mining and video interviewing can get us to more specific directions. Remember, good data, like good writing, is specific. If creative people have access to more specific info, their minds are more easily able to riff, than if they have vague notions. (Plus, it can help break them out of what Creativity expert Edward De Bono called “self-organizing ruts”.)
An interesting phenomenon… language can be thought of as the Legos that we create thoughts and ideas from. So by working with more specific info, our minds seem more capable of recombining it into new ideas.
While I do love text mining, if you or your clients have databases, I’d love to do what many journalists now do – explore those databases using the SQL programming language to ask questions of them. To “solve for interesting”. Data has meaning. It can tell interesting stories. And jokes. For a class project, I once asked the US Census data, “Which religion is most likely to shoot you?”. And asked the Lahman Baseball database (with over 100 years of data), “Which city has the most players who were born, lived and died in that same city?” A testimony to it as a great city to live and play in. (Philadelphia, btw.)
Here’s a short, fun TED Talk about using databases for storytelling by a great data hacker/iquantNY blogger Ben Wellington. He really brings the idea to life.
What I do isn’t all about text mining. (Though admittedly, it’s one of the things I most love doing.) It’s also about munging numerical data to find the product’s advantage(s). There’s a lot of technical data about products available on the web (and in a company’s “gray material”). It’s difficult because things like proprietary hardware and open-source software make it hard to arrive at a true one-to-one comparison. (This can require the help of a tech expert from the company.) I believe it’s often worth it though, since that’s important decision-making data for buyers of these products and services. It’s even more impactful if I can turn that info into an engaging ad or content.
This is the type of project I would love to do but it could be very involved. Let’s take the top two or three luxury car brands (or specific models) and work to compare and understand them from multiple perspectives. And hopefully develop strategies for differentiating your brand.
· Perspective 1): What does the brand/model say about itself? Text mine all relevant materials and website pages.
· Perspective2): What do customers say about it? From the brand loyal to haters. Some of this could come from customer comments, company databases, YouTube, Twitter or Reddit-type fan groups.
· Perspective 3): What do reviewers say about it? There is no shortage of writers and bloggers on the web. Not to mention, many video reviews on YouTube. (I don’t mean to be redundant but I created a web app using the Artificial Intelligence of IBM Watson (and Node.js) that can transcribe those to text with 90-95% accuracy.)
There are millions of people expressing their opinions each day. Twitter, in particular, seems to be a platform for sharing thoughts and ideas. While Twitter no longer allows you access to the Twitter firehose without paying GNIP, the 1% data flow it does allow is actually a pretty fierce deluge of info (180,000 Tweets-per-hour). Not only can we mine that, we can take what we learn and produce video and photography that we put it back out there as relevant Social Media content.
Have some suggestions for Creative Data Opportunities? Please don’t hesitate to call, email or tweet them to me.