Allen and Larry discuss where direct to consumer marketing was 5 years ago, where it is today, where they think it will be in 2022 — and what you should do now to prepare.
(Podcast Transcription)
Allen: Hey to all you data-driven marketers out there looking for new ways to reach unique prospects and better engage audiences. This is the seventh — yes seventh — podcast for the 2 Guys and Some Data series. Giving you the nitty gritty advice you need to actually make more money. I'm Allen Abbott.
Larry: I'm Larry Kavanagh.
Allen: Today we're going to talk about the future of direct to consumer marketing. What current technology and trends we're seeing that are going to make a big impact on how we engage our audiences. A few weeks ago, Larry spoke about this topic at NEMOA. For those of you who are unfamiliar, NEMOA is the National eTailing and Mailing Organization of America. The theme this year was Reimagine, really rethinking the rules of marketing and therefore the traditional tactics we're still using to develop our strategies and marketing mix.
Larry: That's right, there's been so much change in the last decade: computing power, advances in algorithms, new marketing technology, printing capabilities, AI machine learning. So much more is possible now, but it's time for marketers to start catching up to what's currently available and planning for what direct to consumer marketing will be in the future.
Allen: It's something we're pretty passionate about at NaviStone and we've talked about this many times. Larry, you believe there are four major forces driving this change in marketing technology right?
Larry: I do. They are, first of all, advances in artificial intelligence and machine learning. Number two, the ability to integrate data across media. Third, the shift to people based marketing. Number four, the capability to individualize marketing.
Allen: We've talked about individualized marketing and consumer focused marketing in previous podcasts and our blogs, but let's break these four points down so we get the entire picture. Where do you want to start?
Larry: Well we should absolutely start with AI and machine learning since it allows me to say Skynet and Terminator on our podcast. Maybe more sensibly because they are already having a profound change across the whole economy, not just marketing.
Allen: Right, in the last decade or so AI and machine learning have gained in popularity, partly because of IBM's Watson and Google's Personalized Search. I think it was only five or six years ago that Watson competed on Jeopardy and won. Is there a difference between artificial intelligence and machine learning?
Larry: That's a great question and you've actually given a great example of each. AI is the overarching name for the field of computer science, which builds computer programs that kind of imitate or act like humans, can process and do tasks that previously we thought only humans could do. The technology behind self-driving cars is a great example of AI. Machine learning, however, is just a subset of artificial intelligence. In a sense, what machine learning is all about is it allows computers to build it's own gut instinct. Allen as you know, humans — you and I — we build up our gut knowledge by seeing the same type of problem over and over, trying different solutions and seeing what works best.
In machine learning we give a computer a training set of data that allows them to do the same thing. For example, if we were trying to use machine learning to have sort of a gut instinct about what the price a particular house will sell for we might provide the computer data that would include things like what neighborhood is the house in, how many rooms does it have, how old is it, how many square feet are in the house? Then the machine learning program will assign a weight to each one of these data points, guess a price, then measure how far it's guess is off from the actual price, make corrections and guess again. It keeps on repeating that process, further refining what it actually is producing for a guess until it turns out to be pretty accurate.
Allen: Wow, that's pretty amazing. Does Google use machine learning in search?
Larry: Well, as you know, everything with Google is a little mysterious, but they revealed last year that they are using a machine learning software they call RankBrain to continually improve search results. Google has a tremendous training set of data. It processes more than three billion searches a day. One of the examples that Google has given about how it uses RankBrain, or sort of what effect RankBrain has had, is that RankBrain was able to figure out that the answer to how many tablespoons are in a cup is different in the US than in Australia because of different units of measure. Machine learning began appearing in web marketing about five years ago and it is rapidly becoming more common.
Allen: How so?
Larry: Some web ecommerce product recommendation engines use machine learning, I mean certainly Amazon does. Criteo, the biggest supplier of retargeting ads, say they collect 25 terabytes of raw data everyday, and they show one billion display ads a month. All of that becomes their training data set. The web is a great place to capture a lot of interactions with consumers, see results, and then use that as training sets for machine learning software.
Allen: Those numbers are just crazy. It's amazing how fast data captures storage and analytics have grown during the past five years. Where do you think AI and machine learning will be five years from now, and how's it going to impact direct to consumer marketing?
Larry: I'm a little contrarian here because I really don't think that AI is going to show up in consumer marketing in that timeframe. Ultimately it's going to be very big, but I think in the next five years I think the focus is really going to be on machine learning. In part because we're already using it today, particularly in places where there are large audiences and lots of data that we can use for training sets.
I think you'll see it used in direct mail, in email marketing, and website optimization. I saw a really cool presentation on using machine learning to figure out what's the optimal price to sell a product at on a website. I could see that as a huge application for machine learning.
Allen: You piqued my interest here, you said direct mail. This can apply to direct mail as well?
Larry: Of course it can. I mean if you think about it, companies are mailing millions of pieces a year. That's some great training data right there that frankly nobody's using today.
All right, so last week Allen you put me on the spot with a trivia question, now it's my turn. While I'm really tempted to give you a question about Terminator movies, I'll stick with a data question. Did you know that researchers are trying to find a way to store data on DNA?
Allen: Wow, well I'm still trying to figure out a way to store all my passwords somewhere I won't lose or forget them.
Larry: Well here's the question for you then, Allen. How much data can fit on a single gram of DNA? I'll give you a little time to think about it, but I'll also give you a hint. Although really I think this is just frankly giving it away. The average weight of a base pair of DNA is 650 daltons.
Allen: Can I have the Terminator question instead maybe?
We've covered AI and machine learning, what about the other three forces reshaping direct to consumer marketing: data integration and use across media, people based marketing, and individualized marketing?
Larry: Right, well in the past marketers had to identify shoppers by transaction channel — web, call center, store — and consumers had to make a purchase in order for marketers to capture any really valuable data. That's not the case anymore.
Allen: No, it's certainly not. Today when a consumer goes online their web browsing data can trigger emails and retargeting display ads. Cell phones can tell us someone's location. A person’s response or non-response to email marketing can be used to make decisions about sending direct mail.
Larry: Yes, getting a complete view across channels opens up a world of possibilities. For instance, if a customer has unsubscribed from your email marketing list you can reach out to them by uploading their email to Facebook. Certainly we've talked a lot about using browsing data to understand intent and applying that across other media. Another way marketers are using data across media is through geofencing, which involves using the longitude and latitude coordinates of a cell phone to make a marketing decision.
For example, the Uber app can recognize that I've landed at JFK Airport and tell me there are five Ubers nearby. Domino's Pizza can show digital ads with coupons to people that they know are in hotels near airports. If I'm selling margarita mix I can geofence a Jimmy Buffett concert and deliver display ads just to those folks. Marketing to people based on location is an entirely new field that is just beginning to open up new marketing possibilities.
Allen: Wow, I guess there's really no place to hide anymore. By integrating and using data across mediums you can really start marketing to people rather than to channels.
Larry: Exactly, even five years ago marketing departments were set up by channel. You had a direct mail marketing department, digital marketing, email marketing, even had social media marketing.
Allen: We still see that today, but it is starting to shift. Marketing departments are looking toward people or consumer based marketing to deliver a more cohesive marketing campaign.
Larry: You're right, they are co-targeting digital ads to direct mail, they're also uploading lists to Facebook and creating “Lookalike Audience.” Marketing isn't and shouldn't be limited to a channel because consumers aren't limiting their interactions with your business to one channel.
Allen: Where do you think marketing to a person or audience segment will be in five years?
Larry: First of all, I hope we get a lot smarter about which channel we use primarily for a particular consumer. Right now the mantra appears to be if you can market to someone in any channel just do so. As you know, we've seen data that indicates individuals are more or less responsive to different ad media. We can make marketing a lot more efficient by determining a medium preference at the consumer level. I think we'll also start to see dynamic ads on the TV shows we watch. The next generation of TV is the generation after 4K, will use a technology that allows for each TV viewer to see a different ad, much like you do in streaming video today.
It's already possible to target a group of people using a particular search phrase, not just the search phrase used, but a particular group of people using that search phrase. Bricks-and-mortar stores are beginning to match anonymous website browsers to the people who make in-store purchases and then looking back to see what marketing activity drove that website browsing that eventually led to that in-store purchase.
Allen: This leads to another aspect of people based marketing, individualized marketing. We've talked about that a lot in our podcasts and our blog, and it's only really become an option across all media in the last few years.
Larry: Right, in the old days we had audience segmentation, catalogs and direct mail version based on prior purchase history and rudimentary triggered emails. People who visit category X get this email creative.
Allen: As we've discussed, that's not enough today. Marketers must individualize marketing if they want to make an impact with consumers.
Larry: Well we're able to create dynamic display ads, some dynamic email, abandon browse, new products and categories you've shown interest in, some dynamic website interaction. In five years from now I believe we'll be sending dynamic direct mail, which will be the biggest change in direct mail since the coops.
Allen: Yes, you know it's amazing, I've been in this business a very long time and there's been so little innovation in direct mail during the past 25 years. Using intent data from website visits will revolutionize what many thought was a dying industry. There's a few things marketers should do now to prepare for the changes in direct to consumer marketing.
Larry: First thing to do is to make sure you're keeping your consumer interaction data. This means storing website browsing information, call center activity, social interactions. If you're a brick-and-mortar retailer, offer free wi-fi, and then when someone visits your store and uses your wi-fi, you'll be able to retain information about what mobile device it was that visited as well. If you're using direct mail today, this is a pet peeve of mine, bring it at least into the 21st Century. Almost no one should be using RFM segmentation today. At the very least use a simple algorithm. As well consider smaller, more frequent mailings to maximize the hotline effect.
Allen: If you're not using direct mail you really should consider adding it to your marketing mix. Use your own web and retail store anonymous browsing data to trigger daily direct mail postcards.
Larry: As we talked about, I think that location signal is one that's going to really explode as far as marketing goes. You should definitely start testing digital advertising based on physical location. Geofence locations that are likely to attract your type of shopper. For example, if you sell auto parts for classic cars, show digital display ads to people attending classic car shows.
Allen: If you haven't already done so, you should really start applying a people-oriented mindset and set of metrics to all of your marketing. The days of separate measures for different programs is gone. Start by defining a primary marketing channel for each group of customers on your file. If they're active in email use that channel, cut back your marketing spend in other media. If folks who have opted out of email or are non-responsive, try advertising with Facebook or Google. You should even try retargeting with some direct mail.
Larry: Yes, and make individualization a part of your marketing efforts now. Start thinking of ways you can make your emails more dynamic if you haven't already. Test dynamic printing like catalog covers or other formats like a tri-fold or a postcard.
Allen: Larry, five years isn't that far away. A lot can happen. Look back at the last five years, so much has changed and for the better.
Larry: You're absolutely right. A while ago I asked Allen if he knew how much data a single gram of DNA could hold. Do you have an answer?
Allen: Well I don't, but I do know how many parsecs it is from the ice planet Hoth to the moons of Endor, if that helps?
Larry: Not so much. A single gram of DNA can hold 215 petabytes of data. One petabyte can hold 13.3 years worth of HDTV video. Of course, storing data in DNA isn't cost effective — yet — but who knows, five years from now we could be sticking our finger into a data port to download a movie.
Allen: Wow, finally a way to store all the episodes of Seinfeld.
Larry: There you go. Now that we've gotten a few podcasts under our belt, we're ready to start opening up to our listeners. If you have a marketing question for us let us know. Between the two of us we've had a lot of experience in the industry.
Allen: To kick off our new Q&A section we have a question Larry received from a round table discussion he took part in at eTail West.
Larry: Yes, there was a great question around how IoT, that's what people are calling the Internet of Things, how that will impact marketing. What I told the person who asked the question, there's really a couple of categories of IoT devices. First one is wearables, literally devices that you wear, think of a Fitbit. It may sound a little bit scary, but my guess is that wearables will become another source of location data just like cell phones, and that location signal, I think, is going to be really big. Another category of devices are things like Alexa. Amazon is keeping track of the questions and tasks you ask Alexa and they plan on developing advertising content around that, much like paid search ads respond to your Google query.
My favorite example though of how IoT devices can affect marketing is what Domino’s is doing. They're a really interesting company from a technology standpoint. They've created an easy order button, it's literally this button in a little pizza box that they give you. It is connected to your cell phone. You can setup on your cell phone what your favorite order is, what your type of pizza is, and by just pushing that button it will automatically send a signal to order a pizza.
Allen: That's amazing. Not sure about any of those for me, though if I could push a button and have a bourbon placed in my hands I might go for that. Well if you have questions for us email them to cheers@navistone.com.
Larry: That's it for this episode. Thanks for listening to two guys ramble about the future of direct to consumer marketing. If you like this topic you'll probably find our blog Bridging Digital and Direct Marketing interesting. You can find it and more resources at navistone.com/blog. Again that's navistone.com/blog. We'll be back in a few weeks to talk about how to take your retargeting offline and why that's a good thing. I'm Larry Kavanagh.
Allen: I'm Allen Abbott, thanks for joining us today.