A Playbook for Programmatic Creativity
On October 26th I gave a talk at the iMedia event in Austin, Texas on the topic of programmatic creativity. I was going to post the slides to Slideshare afterwards, but the slides themselves don’t make much sense, so instead I’m posting both the slides and my speaking notes here. The copy is a little bit rough around the edges because I wasn’t writing it for publication, but I think you’ll get the points I was trying to make, and if this topic is up your alley it might be useful in some way. Hope so.
Let me start by telling you about the presentation I’m not going to deliver.
The talk I’m not going to give is about how I launched Anagram 6 months ago to solve the problem of programmatic creativity, and through a unique and proprietary methodology delivered amazing results…blah blah blah.
In that talk I’d tell you about our work with my first client, OptiShot Golf, and all the great things that we did. And, see, there are great stats… the client said nice things… it even got a little bit of press.
And these are all true things, by the way.
But honestly, I wouldn’t want to see that talk if I were you. Whenever I see a talk like that I always feel like I didn’t hear the real story, and it was really just about making the presenter look like they have all the answers.
Which I don’t.
But I don’t think anyone does yet when it comes to programmatic and creative. As far as I can tell it’s something no one was even thinking about until very recently.
But just for the record before i move on… I could totally have given that talk.
And I think you would have been really impressed :)
I decided to call this presentation a playbook for programmatic creativity.
It’s not “the” playbook. Just “a” playbook.
I think this topic — reconciling automation with creativity — is probably one of the most interesting and important discussions in our business today.
And there’s not one simple, right answer.
It’s definitely as not as straightforward as saying “do dynamic creative.” It’s a topic that impacts tools and process and culture and even the business model itself.
This talk is going to be less about answers and more about the path my company is on to try to make meaningful progress in the area of programmatic creativity.
I’m going to tell you the avenues we’re pursuing, and share some principles, some techniques, and some of the partners we’re working with.
I hope it sparks some thoughts of your own that you might be able to use.
I have a tendency to over-simplify, but this is the premise I’m working with.
I think essentially all advertising spend will be transacted programmatically. Whether that’s one year out or five I don’t know, but automated transactions across all forms of media seems to me to be an inevitability.
Next, programmatic has been a sizable part of the digital mix for 5 or 6 years, and I’d say that at best, creative has been an afterthought. Almost all of the intellectual energy in the space has been put towards targeting and bidding, which made sense as a place to start.
But at a certain point, targeting can only get so much better. And to me it seems like the path toward improving outcomes from here on out is through the application of creativity to the programmatic infrastructure that’s been built.
That’s the premise for my talk. It’s also a big part of the premise of my company, so if you disagree with me, please don’t tell me, because it’s too late for me to go back to my old job.
The reason it’s such a big opportunity is because creative is more important than media and targeting.
People see ads. Not targeting. Not algorithms.
Hard to find data that quantifies the relative value of creative and media, but a Comscore study from 2010 said that creative has 4x the impact.
Media is incredibly important. It can be strategic and creative in its own right, but ultimately, it’s just the delivery mechanism for advertising.
It’s not an end in and of itself.
The most powerful lever we have is creative, and it’s the lever that has so far has been used the least in programmatic.
I’ve spent a lot of time thinking about this and the best way I’ve come up with so far to frame the opportunity of programmatic creativity is to think about this question:
“What would you say if you knew who you were talking to?”
You wouldn’t say the same thing to everyone.
And you wouldn’t say the same thing over and over to the same person.
But of course, that’s what most advertising does. Even in digital, if we’re honest.
To live up to the challenge of the question “what would you say if you knew who you were talking to?” we have to think about versions and sequence.
In two words, this is what programmatic creativity is about: versions and sequence.
We have to make a lot more stuff, and we need to communicate in a way that evolves over time at an individual level.
I actually will tell you about one aspect of our work with OptiShot, about the versioning we did. Worked with a great creative agency on this called Detroit Branding Company.
OptiShot is a golf simulator. It’s a $500 product that appeals to avid golfers, and they’d never really done much advertising other than some DRTV product demonstration ads.
So we wanted to find out which aspects of the product were most appealing to people.
We defined four broad thematic areas, and then for each theme we had four messaging variations, so there were 16 distinct messages.
We created two visual templates, and then we did the 7 0r 8 resizes for each concept, which meant we made something like 200 discrete ad units.
And the production of those 200 units, all of that versioning and resizing, all the way through to exporting the ad tags, took about 2 and a half hours.
I did this using Paper G, which I’ve found to be an amazing platform for building variations at scale.
The point is, we’re now at a point where production cost and timeframes are absolutely not an impediment to making creative that has the potential to speak to people at an individual level.
A lot of people aren’t taking advantage of this. I’ve worked on campaigns where the same two or three ads were running for a year, which is such a simple missed opportunity when we do in fact know who we’re talking to.
So, make more stuff.
I’m proud of what we did with OptiShot but I think it’s just the tip of the iceberg. A lot of what we did was intuitive.
The really interesting thing about programmatic creativity is the idea of using data to make ads people actually want to see.
Or at least, given the rise of ad blocking, making ads people don’t want to not see.
But what kind of data?
One of the things I’m thinking a lot about is data about feelings as opposed data about facts.
At Hill Holliday I had the chance to work with really great creatives and brand strategists over the years, and the one thing the best of them all understood is that human beings are motivated by emotions, and that decision-making isn’t rational.
People do things because of unconscious motivators that none of us really understand, and then post-rationalize those decisions with facts that justify them. At Hill Holliday the team talked about this in terms of “motives” and “alibis.”
But so far the digital and programmatic world have been operating with data that is purely about facts. “You looked at this, therefore you want that.”
It’s not that it’s wrong. But it’s incomplete, and not aligned with how people behave, or how they make decisions.
This is one of the classic quotes in the business, and of course it’s completely true.
I think about how this quote could be amended for our era, and i think it’s something like ‘there’s nothing more powerful than an insight into human nature… applied at the individual level.’
That’s really what the potential of programmatic creativity represents… it’s about an understanding of human nature, but executed in a way that speaks to one person at a time.
When data operates only at the level of facts, it can’t come close to living up to this ideal about insight into human nature.
But data about feelings… about motivation… well, that could.
Here’s one human truth: people are inherently motivated by different things.
The same brand, the same product, can be motivating to many different people in many different ways.
So what if you knew, based on data, the driver of a person’s motivation? If you actually knew what made them tick, and could present your brand or product in a way that aligned with their motivations?
This is the idea of motivational fit. When something is presented to a person in a way that aligns with their motivations, good things happen. People are more receptive.
I’ve been working with a new company called Real Engagement And Loyalty (REAL.org). They’ve built a whole new taxonomy for digital targeting based on 24 character strengths from positive psychology that have been shown to be predictive of behavior.
REAL is has managed to tag the US population with these 24 character strengths, each strength having 100’s of studies behind it, so that creative can be modified to appeal to segments and individuals.
When character strengths and messaging align people have been shown to purchase — not click thru — at almost 4x.
Sometimes when a character strength in an individual matches the character strength in an ad or product, it triggers a subtle positive emotion that is based largely on the hormone and neurotransmitter oxytocin.
This starts to get at the notion of how to use data to make ads people want to see, because we’re getting at not just the facts about what they’ve done online in the past, but about the way they actually interpret the world around them and make decisions.
Next lets get into sequence.
And the way I think about this is, how do you always say the next right thing?
That is the ambition — to speak to people in a way that is informed by all of this past interactions your brand has had with them.
This is one of the reasons I really believe that programmatic will be the default setting for all marketing, because it allows you to map communications to the customer journey at an individual level.
I think it’s pretty well understood by now that the funnel is a simplification, and that in reality people move through a fluid journey, at different speeds and in different ways.
But there’s a huge difference between just making up a prototypical journey for a client presentation, and actually building a system of communications that responds in real time to where an actual person is in their actual journey.
I’ve been involved in presentations for years where the journey was the construct we used to map work to, but there was no way to actually deliver it. It was completely theoretical.
Programmatic makes the customer journey actionable.
Because at any given point in time, it’s possible to know some large set of the interactions your brand has had with them. Not every interaction, but a lot: digital ad exposure, site behavior, email, purchases, etc.
To do this you need to be using a DMP as the quarterback for your communications, and I’ve become convinced that this is the path forward, as opposed to a reliance on DCO.
DCO seems to me to be operating in its own silo, not informed by how the DSP is optimizing, whereas the DMP is simply telling you where someone is in their journey based on the most up to date data, which gives you the ability to say the next right thing.
(Note for normal readers: DMP = Data Management Platform; DSP=Demand Side Platform; DCO=Dynamic Creative Optimization)
I didn’t realize this until recently, but there’s a big change happening with DMPs.
DMPs are becoming much more fluid.
The first generation of DMPs kept people in fixed buckets.
You had to identify a segment in advance, wait for it to build, and then once it populated the users stayed there. So the segment “suburban dads who have purchased once” would not update automatically when that suburban dad purchased again. He’d just stay in the wrong bucket.
But the next generation of DMP releases are dynamic in the sense that a user can move from one segment to another based on new information
Which means you can define your journey, and then as people move through it, present them with the right creative automatically
That’s how you approach the DMP, as a game of if/then. Set up the creative rules in advance, and then as people move thru they journey they are automatically presented with the right creative.
That’s the what, version and sequence, now let’s talk about how…
Good place to start when trying to build a model for digital success is BuzzFeed.
Jonah Peretti talks about BF as a learning machine, which is an idea I’ve modeled a lot of our processes after.
And I think of a programmatic strategy as building a system that is in effect a learning machine.
Building a Learning Machine starts with having a Learning Agenda.
There’s a big difference between doing things and actually learning.
And a trap I see a lot of people falling into is confusing activity with learning.
In many cases I find that new clients really don’t know what they know or don’t know. They’ve done a bunch of things, but because there’s been no discipline to it they’re actually no farther ahead than when they stared.
We try to be very disciplined in the process of having a learning agenda.
Defining in advance what it is we’re trying to learn.
This is a real example of learning agenda for a test we’re about to do for a client, so I’ve covered up some of the detail here.
But you can see that we are testing a couple of key things, and when the test is done we hope to know something more about how to differentially message to younger and older audiences.
And no matter what happens, six months from now if someone on the team wants to know what we know about age targeting, this test will be tagged and archived and searchable. They won’t have to learn the same thing again.
The in-market process we use to manage campaigns is an adaptation of agile software development, which to me is about having a sense of humility in the face of the marketplace.
The faster you get your ideas into the real world, and get real feedback from real users, the better off you are.
You can have the best, most clever strategy in the world, but the only thing that matters is how people do or don’t respond to the work, and how fast you can get better than you were the day before.
We talk about that as being right fast vs. being right in advance.
I’ve been in digital forever, and for as long as I can remember optimization has been about reduction, about taking things out. Start with 10 sites and optimize down to 6. Start with three creatives and eliminate the under-performers.
Again, BuzzFeed as a reference point.
That’s not how BF looks at things at all. They are looking for signals of receptivity, and then make more things.
Lesson I learned from looking at what they did with siblings videos.
When BF sees something is working, they do more things like that.
This is something that in my experience very rarely happens in the agency world, but we’ve built it into our process, which looks like this when you look at it from end to end…
Big point — this means your creative team has to stay on the business. Not the way things usually work. Usually they do the work up front and move on.
Media people and analytics people can only make programs better through reduction…taking out the things that aren’t working.
But when creative people are involved in optimization they can actually make new things, responding to the feedback the market is giving them, and that’s much more powerful.
Not only do creatives need to stay involved, they need to play a much bigger role than we’ve let them in the past.
I’ve been influenced by the idea of the full stack developer, person that can do front end and back end coding, sort of a jack of all trades.
And have started thinking about the idea of full stack creativity, by which I mean a person (or small team) that comes up with ideas, makes the stuff, looks at the metrics, and then makes the optimization decisions.
From beginning to end, these people are responsible for the success of the thing real people in the real world see.
And really I have one simple point to make here: give as much autonomy and accountability as possible to the people who are making the assets.
I think there’s a reason for this that science would support. Just like we all know (and has been proven scientifically) that when we post something to Facebook or Instagram we are compelled to check on it, the same thing happens when the person who made a piece of creative has access to metrics.
Yet most of the time creatives have zero access to metrics. And if they do, it’s a report that comes weeks or months later.
Well what I learned first hand (because Anagram is a startup and I had to wear multiple hats) is that when you’re the person who made the piece of creative that’s in the world AND you’re the one looking at the analytics, the same exact thing happens.
You become incredibly motivated not just to know what’s going on, but how to get better feedback, better results. Give creatives access to data and accountability for improving results and you will see amazing things happen, right away.
Programmatic creativity can’t improve unless the most talented creative people want to be a part of it.
And I think we have to be honest with ourselves that right now, this isn’t a corner of the industry that appeals to them.
I think the answer is to give creative people a level of freedom and accountability they’ve never had before, to see their work through and take real ownership over it’s success in the market.
And in a world defined by outcomes, maybe even pay them more based on their ability to generate results.
Instead of hiding the data from them, give it to them.
And instead of assuming they don’t care about real business metrics, let them try to influence them.
I do hope this has been helpful to you in some way. I know that thinking these ideas through and getting them down onto paper was very helpful to me.
Thank you very much for your time.
