I was watching a pre-recorded conference session earlier today on a specific machine learning technique (Markov chain models), and the presenter said, “Oh yeah, and you can use this for X too” as a throwaway segue into the next section of the lecture. The presenter gave no thought to it, but the idea exploded in my mind brighter than a Fourth of July firework. It was so OBVIOUS. And yet, it had never once occurred to me to try using this particular technique on a different kind of data. I feel kind of silly even writing that now, because… well, it’s so obvious in retrospect. Of course, an idea is just a dream until given form, until built - but still, the idea is much more than I had when I started my Sunday morning.
This is what I mean when I say you already know the next big thing. Chances are there’s an idea - a strategy, a tactic, a technique - in your mental inventory that is proven, that works, that’s been battle-tested. You know it, you rely on it, you live it - but you haven’t moved it around to a new domain, to a new area of application. For example, this week, LinkedIn debuted its own Stories feature. I know how to make Instagram Stories. I know how to make Tiktok Stories. It’s trivial from a technique perspective to make LinkedIn Stories because I already have the experience and knowledge that I can simply move over to the new platform. What are your bread and butter techniques, tactics, and strategies? Which ones have you tried to move into a new problem domain recently? Chances are, if you haven’t, you’ve got many opportunities to take what works and make it work somewhere else. You already know the next big thing.
Let’s talk fishbowls. I recently started volunteering with one of the political campaigns in here in the United States, calling voters in battleground states to encourage them to vote early and by mail (which is the best way to vote in a pandemic). Why did I do this? Posting on social media, or even in an email list, only reaches the audience who knows me. It’s within the fishbowl, as it were. If I wanted to make a difference in this election, I’d need to step outside the sphere of people who know me to reach the people who don’t know me, and that means doing things like working with direct outreach methods - calling, texting, etc.
Even if I had a massive amount of reach, if I stayed within the fishbowl, I wouldn’t reach anyone I didn’t know. And there’s a bias. You wouldn’t be reading this if you weren’t connected to me in some way - attended an event I spoke at, signed up for the newsletter, downloaded one of my papers, etc. Thus, there’s a whole population of people who I don’t reach, who don’t see what I have to say, but are still vitally important on the particular topic of the upcoming election. So what, you may ask? The reason this matters isn’t because of politics. It’s because your marketing functions in exactly the same way.
Think about all the channels you have access to, and ask yourself: which reach existing audiences, and which reach new audiences? Email marketing? Existing audiences. Unpaid social media? Existing audiences. Remarketing / retargeting? Existing audiences. Organic and paid search? New audiences. Paid social media not to your followers? New audiences. Public relations and broad advertising? New audiences. It’s not a question of inbound or outbound; those are last decade’s concepts. It’s a question of whether you are serving the people who already know you, or reaching out to people who might want to know you. You need a blend, a balance of both, if you want to create both loyalty and growth. Where is your marketing budget? Is it entirely within one fishbowl? If so, it’s time to spread it out a little.
We all know that gaining an understanding of the attributes of your ideal customer improves the chances of your marketing and sales efforts’ success. But with this in mind, where do you start? A buyer persona is a semi-fictional representation of your ideal customer based on market research and real data about your existing customers. Buyer personas help ensure that all activities involved in acquiring and serving your customers are tailored to the targeted buyer’s needs.
Which do you want, the process or the product? How many people know how to bake a pie from scratch? Not pick up a box in the freezer section, but “you’ve got a pile of flour and other ingredients in front of you, go!” scratch. Not many. Certainly not many without having to Google it and then spend a whole lot of time on trial and error before baking something that vaguely resembles a pie. That’s the process. Now, how many people know how to eat and enjoy a pie? Probably a lot more. That’s the product. Do you need to know how to make a pie from scratch? If you’re a bakery you absolutely need to know how to make a pie from scratch. If you’re someone who is concerned about the ingredients in their food or you have a specific, severe allergy (but you love pie) then you also need to know how to make a pie from scratch. But if you’re hosting a dinner party, you don’t need to know how to make a pie from scratch. You just need to know how to use it - when to serve it, what temperature it should be at, what to serve with it. If you’re contributing to your office pot-luck, you don’t need to know how to make a pie from scratch. You pop by the store, pick one up, and enjoy. Do you see the difference? Process is about knowing the intimate innards of the pie and every aspect of how it works. Product is about making use of the pie in a broader context. This is how to think about AI and machine learning. In the case of the bakery, the pie is a core part of their services. In the case of the dinner party, the pie is just an add-on to enhance the experience. AI is the same. Is the use of AI part of your business core competency? Is it part of the secret sauce? Then you’d better know how AI works, the ins and the outs, how to build a model from scratch, the works - the process. If AI is just being used to improve some processes, then you don’t need to build it from scratch. You just need to know - like pie - what constitutes good or bad AI, how to tell good vendors from bad vendors, and how to use the AI once you’ve acquired the technology and software from the vendor - the product. That differentiation governs who you should hire, too. You don’t need to hire a baker if making pie from scratch isn’t part of your mandate. You don’t need to hire a dozen AI engineers and data scientists if making AI from scratch isn’t part of your mandate. So, which do you want? Which do you need? The process or the product?
