Successful marketing requires the ability to stand out, and do things differently in order to rise above the noise. No-one understands this better than Phrasee, and their outspoken CEO Parry Malm.
Listen to the full conversation here.
You spoke at the Festival of Marketing in 2018, about the importance of ethics, especially when it comes to applying AI and behavioural data to marketing, and not using the negative emotions to encourage people to buy. Do you think since then we have seen a marked improvement in that? Or does marketing still have quite a way to go?
Well, I mean, I think you can’t really paint marketing with one brush and say it’s either good or bad, like anything in life, there are good actors, and there are bad actors, I think that there’s definitely been a real move in the marketing industry in the last couple of years to move away from these sort of dodgy short-termist tactics, you know, these double glazing tactics that you high-pressure people into buying stuff that they probably don’t want. But there’s still going to be actors who act like that, I think what has really taught me in the last couple of years is that like, we’re in a fortunate spot where we can choose who we work with, and choose who we don’t work with. And by virtue of doing that, we’ve built up a client roster of brands who are forward-thinking, and we reject brands who seem to be a bit more stuck in the past or just focused on different priorities than what we think is important.
What we do for our audience, and so what we encourage our audience to do is, I suppose it’s still the old school method is, is a lot of split testing, I wanted to get your opinion on some of the things we’ve come up there, one of the things we found that actually took us by surprise, we did it as a throwaway test was a lead gen email with a blank subject line actually did really well. And are we the only ones that have ever done that, because others have done that and had a good result from it?
Well, I can’t really comment on lead gen emails, because our technology isn’t built for sort of small-scale marketing like that. And what I couldn’t really say is, like any specific rule sets, which are sort of a one-size-fits-all, you know, “if you do a, b, and c, it’s gonna work for you.” I mean, if that was possible, and if this problem was so simple, then we wouldn’t exist as a business. I mean, we encourage our customers to test lots and test at scale. And we use the output of those tests, as base training data for various AI models. And what has shown us is that language, be it in subject lines, or Facebook ads or wherever, is insanely complicated. And while, we as humans all think that we can write the perfect subject line say, the thing is, there’s like billions of ways of saying the same thing. The human brain is not set up to explore that entire universe of different linguistic combinations and permutations, and ideas and sentiments and words and phrases and punctuation and emoji, and so on and so forth. So, I mean, yeah, split testing is critical. And it’s super important to make sure you’re putting your best foot forward. But if you’re just split testing a bunch of bad stuff, then you’re gonna find the least bad option, it’s really important to work in conjunction with high-powered predictive models, not just sort of split testing on a whim, or else you’re just going to be chasing your own tail over and over and over.
Okay, you mentioned emojis there. And I think in in your book The Language Effect, you’ve talked about emojis a little bit as well. One of the things we found is that they become less effective over time, they initially massively outperformed just a text-only subject line, do you think all sort of language optimization has that problem that there’s a degree of shiny newness in it and then people will revert back to what they what they’re more familiar with? Or is that not part of what you can test do you think?
Well I mean, emojis by themselves aren’t good and they aren’t bad. They just are. An emoji by itself is not going to make or break a subject line, it’s not going to make somebody open your stuff. We’ve run some pretty wide scale data-backed research on this topic. And we found that emojis are kind of like, Sex Panther Cologne from Anchorman, where 60% of the time, it works every time, right? Where basically, if your subject line is good in the first place, adding a contextually relevant emoji will make it better, and it’ll amplify it. But if your subject line sucks in the first place, then it will amplify it just in the wrong direction. I think it’s really important to think about emojis as amplification tools, but not as a binary “should we use them or should we not?” Because that’s just not the right question. The right question is, “Is your subject line good in the first place? Will emoji make it better?”
Does, then the same argument apply to specific personalization with an audience; that people aren’t going to respond to a bad subject line with their name in it, any more than respond to a bad one without it?
Yeah, I mean, the thing about personalization is, sticking somebody’s first name into a subject line, that’s not personalization. That’s a database match. But I do think that people have gone down this rabbit hole of personalization for the last, you know, five or 10 years saying that we’ve got to have the right message to the right person at the right time. And I think that that’s like b******t, basically. So like, if personalization worked so well, then why is the single most effective format of advertising on the planet still a homogenous Super Bowl advert, which is not personalised whatsoever? If personalization was right for all marketing, then we would no longer have any mass marketing, but the majority of advertising spend in the world is done on mass marketing. And it’s all about finding that message that resonates with the masses, because that’s how you really galvanise growth. It also allows you to create, advertising based social experiences. So, I mean, one example from the late 90s is the Budweiser “Wassup”, right? where that was a generic ad that everybody got, but g*******t. At that time, me and my buddies, were all phoning each other up and being like “Wassup, Wassup!” And it was a thing. It was like a cultural phenomenon. And you would never get that with personalization because it’s not a shared experience. So, again, “is personalization, good or bad?” There are some very, very specific use cases in which personalization is good, but they’re very specific and very point-based solutions. Broadly speaking, advertising works when it’s mass market advertising. And people seem to have lost that point, because there’s so many technologies out there offering personalization solutions. But just because there’s technology there doesn’t mean you should use it. And if you can craft a really effective mass marketing message, you’ll get much better bang for your buck from doing that, than you will from blasting out stuff to Bob and Betty.
Do you guys separate your client base between B2b and B2b? Because our bases is generally B2b. I it sounds like the answer might be “you can’t generalise”, but how much can you draw distinctions between what works for B2b, and B2c because either you need to learn the two schools seem to be either copy everything B2c does, or you see it as a completely different silo because they’re not trying to target an audience in in the same way and they’ve got completely different buying cycles.
Yeah, I’ll be honest, like that’s not something that I think about very much we sell into businesses. So we’re a B2b operator. All of our customers are large like blue chip, b2c brands like your eBay’s, your Dominos’, your Groupon’s, brands like that. But we’ve always like marketed ourselves just in in our way. I mean, a lot of people will look at it and be like, “Oh, Phrasee’s cool. It markets itself like a B2c brand.” But I mean, I’ve got no experience in doing B2c marketing, I’m a career B2b marketer. I just think like most B2b marketing sucks and is boring. And one of my big mantras from day one for Phrasee was just to never be boring, because like, there’s no reason why a B2b marketing campaign needs to be super lame and boring. Like, just because your product’s boring doesn’t mean that your marketing has to be boring, right? So we just don’t think of ourselves as like thought leadership, we think of ourselves as like punk rock perspectives, where it’s like, just because stuff is always been done in a certain way doesn’t mean it needs to be done like that. So we’ve just found what works for us. And really, I mean, lots of B2b marketing just kind of sucks. Does the world need another four-page white paper? It doesn’t, it absolutely doesn’t. So we’ve really leaned into it, we’ve created an entire content hub. And we’re just piling loads of just useful stuff into it. Because that’s what works for us. For some brands, our methods probably wouldn’t work. And there’s probably a bunch of experts who would look in our shop and be like, “What the hell are you doing? You’re screwing it all up.” But I mean, we’re not. So, you know?
If it works for you, then you crack on, basically? Yeah. Sure. Do you think part of the problem with data is that we start thinking about the why something works? rather than going “Oh, cool. That works. Let’s do it”?
Yeah, like we find that a fair amount where, you know, especially because we’re dealing with, you know, lots of language at scale. There’s an intellectual curiosity about why something is happening. But people actually scrutinise data a lot more than they scrutinise humans. Humans, on a daily basis, make bizarre choices. I mean, just think about your daily life. I mean, probably, you know, if you think about some of the choices you’ve made in the last week, none of them, or like a bunch of them won’t be rational. They won’t make sense, but they’re the choices you made. So you can rationalise them retrospectively, when you do stuff on a purely data-based standpoint, one of the challenges is that you then try to like rationalise it with numbers and try to find reasons why something happened. And you’re absolutely right, you do like lose sight of, “but it works.” The why matters to a certain point, especially if you’re dealing with stuff that sort of, you know, treads the line of ethical boundaries and whatnot, where you do need a solid justification as to why you took a certain course of action. But if you’re sending out emails to a few prospects, or if you’re writing a subject line, like, even if you do it wrong, no one’s gonna die. So, just do something that works.
Yeah, you can get obsessed, too obsessed with the spacing or who an email comes from, and signs off and things. But ultimately, it only matters what your audience think of it, unless you are your audience, which is, which isn’t the case for most marketers, then whether you like it or not, is completely irrelevant to the creative process, as long as you can justify it with results out the other end?
Indeed, yeah, I mean, I think like, like people, people try to sort of project their beliefs on their audience, right. And the problem with that is that your audience, first of all, is not the same as you and your audience will perceive your brand and your offering in ways that are different from how you aspire for it to be positioned. So, I mean, you can try to force through a message into the market, but the market is going to understand it in their own unique way, and so forth. So, you sometimes just need to kind of flex with that and not be super rigid, and like, beholden to this sort of marketing programme, which you’ve got written down, you know,
There are bits of marketing that you have to do week in, week out. Now that list is shrinking as we go on. But how do you find a balance between not throwing out something that works in favour of something new and actually embracing a new thing that can do better? That’s something I think, especially small marketing teams, really struggle with.
I guess the example here is the power of email, right? Where for like, years and years and years, I mean, you guys would know this acutely, I mean, every few years, somebody is gonna kill email, like it was blogs in the 2000s. Then it was RSS feeds, then it was Facebook, then it was Slack. I mean, next year, it’ll be something else right? And people go like, “Okay, we’ve got to get a big Facebook presence.” So they’ll hire a social media person will then build them a Facebook group, then they’ll realise that they need to pay Facebook every time they want to reach that that group. And it becomes this sort of walled garden of skimming off the top where you basically need to pay a Facebook tax, or most egregious is a Google tax, where if you go to your brand’s website, got to pay Google a tax to come up first now, or else your competition bid on it, right? The great thing about email is that it’s like first party data, like once you’ve built that data, there’s nothing stopping you from contacting that data as much as much as you want. I mean, there’s a small marginal cost with an ESP, but it’s a marginal cost, you know, it doesn’t really touch the sides. Email is super boring, it is probably the most boring of any of the online channels. But it’s also the workhorse, the most effective one. And the one that should probably get a lot more love and a lot more investment because like I said, it’s first party data, you are not beholden to these, monopolistic actors in the marketplace, who are basically eating your lunch to access your audience.
Are you able to explain how, how you how you retain a brand voice through language if you’re if you’re using purely data-driven insights, your copywriting it always seems like the argument is “Oh, brand voice has to come from a person because people understand it better.”
Yeah, so basically the question is, can you codify brand guidelines, right? If you think about a brand guideline document, it’s a set of rules which you need to follow to ensure that the language which you use adheres to your brand’s tone of voice. Now, if you have a set of rules that need to be followed, there’s an old adage that humans say, and that’s “Rules are made to be broken”, so people will break rules. If you have those rules codified in place, a machine is incapable of breaking those rules. So if you use a machine to generate content, it will not flout rules, it will not justify rules, it will not bend the rules, it will adhere to the rules exactly as you’ve written them. So actually, to adhere to a brand guideline, you’re actually better off using software then leaving it to humans on their own, because humans will always find a way to bend the rules, “Rules are made to be broken”, right? Whereas machines are not capable of doing that.
And I suppose that comes back around to sort of after-the-fact justification you say always send this email out, oh, why did you do this? Or because I think our brand voice can evolve in that way. But it might not be the rules that you’ve laid out especially as the marketing team or probably making the rules and applying them?
Most definitely. Like there’s always a way to like justify actions you know, this like post-event rationalisation is why, suddenly it’s okay to drive 230 miles to Durham to see your parents during a f*****g lockdown. Pardon my French, right. Humans are so good at like gaslighting, right? Whereas machines are incapable of doing that, so if you want to have like a codified marketing programme, or codified marketing voice, the best thing you can do is invest in software to codify that because then you remove ambiguity, you remove cognitive bias, you remove gut feeling, right?
The other objection, you’re likely to get there is that “machines are going to take all our jobs.”
The way that I’d frame that is like this: if you think back 20, 25 years, when people needed to design something, let’s say a brochure, right, they would have an illustrator, a cutting room, they would have like this big graphic area, right? And they would make this beautiful thing, right? They would do it all by hand, they would then send that off to the printer, who would then put it onto a litho screen, they would print it out, mail it out, become billionaires right? Now, that involves a lot of human stuff, right? There’s a lot of human work, there’s a lot of sort of specialist equipment needed. So there’s huge capital costs also. But then some guy said, “What if we did this on machines instead?” That guy invented Photoshop or CorelDRAW or whichever, right? So then what that did is, it democratised access to high-powered design tools. Now, at the time, there are a bunch of designers going ” this programme is going to come and take my job, I’m not gonna have a job in the future because these machines will make me irrelevant.” And people who thought like that probably did have to retrain and get a new career. But those who didn’t think like that, who actually embraced that technology, and realised the real power of it. What Photoshop did is it took away all of the mundane parts of graphics, design and so forth. And it allowed people to be creative much more efficiently and effectively. It also democratised access to it. So now there’s actually more designers on the planet than there were 25 years ago, directly because of Photoshop.
What Phrasee is doing is quite similar, right? People were always beholden to copywriters, to this sort of like dark art of copywriting. And there are some very skilled copywriters out there, these corner-office-situated, fancy-suit-wearing Don Draper types, but they are not a dime a dozen. They’re very scarce, there there’s not many of them. So would you want to have Don Draper spent all of his time writing a subject line for you? Well, no, it doesn’t make sense. What Phrasee does is, it democratises access to those mundane parts of copywriting, so good copywriters can focus on doing good copywriting. The copywriters who are fighting it are the ones who just aren’t thinking very creatively. They’re thinking very short-term, this technology is here, it’s not going away. One Luddite with a hammer in the cotton mill is not going to take down the Industrial Revolution. So the smart ones are actually embracing it. And it’s those marketers of the future, who can look out beyond this sort of mini existential crisis, which they’re having right now, they can look beyond that, and realise that actually, what’s happening is, is a revolution. There’s a Photoshop now, but for copywriting, and people who are jumping on board with that are reaping the benefits.
People following up from this podcast, you’re going to we’re going to point people towards Phrasee, we’re going to point them towards the book that you did, The Language Effect. But in terms of other resources that people can take to start thinking more creatively about copy. Where else would you point them for inspiration?
I mean, I’m not a big fan of business experts. I’m not a big fan of Gary Vee and people like that. I’d say as soon as somebody calls himself in an expert, you should turn around and run for days. I’m personally a big fan of learning from the past. Like, I mean, one book that I always recommend reading is Candide by Voltaire. Because what it does is it really outlines a difference between scarcity and rarity. And you don’t need to read a textbook to get that. But that’s hugely important. Because if your product is rare, it doesn’t matter. If your product is scarce, then you’re on to a real valuable product right there. I guess my, my point is, if you want to be real creative, then you got to get outside of these echo chambers. If all you read is like marketers talking about marketing, you’re never going to learn anything new. If you’re only speaking to people who are doing the same work as you, you’re never going to get outside of that echo chamber. You know, it’s why I read the Guardian, and the Daily Mail. To be honest, I don’t agree with either of them. But it’s really important to get as many points of view as possible. So then you can synthesise them and make your own opinions up. I mean, it’s a bit of a wishy-washy answer, to be honest, but I just like, I find the vast majority of like marketing books and business books that I’ve read, I know less after reading them that before I picked them up, and that doesn’t really instil me with huge amount of motivation.
I think it’s quite freeing that there’s no, get me all these five books you must read to be a marketer, and they’re dense tomes, but they are. No, Candide is not particularly long is it I think, I think I had a look at it before and it’s only 120 odd pages.
It’s after a big earthquake in Portugal. It’s just a lovely read. But what’s super neat about it is that when the dude went to, I think, it was Peru and discovered that the tribes there had lots and lots of gold, but they didn’t care because they didn’t value gold, gold there was not scarce. So it’s a neat concept where, if you can really find something that’s scarce, and then you can make a lot of money from it. If you find something that’s rare, then it’s basically it’s like, anthrax is rare. But there’s no demand for anthrax.