AI-Powered Compliant Content Creation at Scale: How Far Are We From Using Modular Content and Technology to Radically Transform Content Marketing in Pharma

Bozidar: Hello and welcome to the new episode of the Pharma Launch Secrets Podcast. I am joined today by Diana Lee, Co-Founder and CEO of Constellation Software, Inc., a marketing technology company focused on healthcare compliant modular content. And before founding Constellation, one of the things she did in her career is leading Pfizer's Global Social Media Campaigns, leading the team to design a strategy to launch a global social media awareness campaign on medical cocculant disease. Welcome Diana, really excited to dive deep into all the topics, all things content.

Diana: Thank you so much for having me.

Bozidar: All right, great. So let's first start with just high level definition of modular content. What is modular content? What are the examples of modular content and what are the benefits of modular content?

Diana: So modular content is pre-approved content that's pre-compliant for the pharmaceutical industries to launch. Many of the content that's actually out there right now is just an image and it's very difficult because most times with the brand teams they actually have to put a word doc together that has messaging, header, the description of the content, where it's going to be placed, the format it's going to be going to, the disclosures, and then basically the images that they want to use. All that information currently is being done manually and sent to an agency of record for the creative to actually manually make that content for all the digital channels, TV, video, and every other asset. So for Facebook, for Google, for Health Endemics, for every single piece of materials that they need out, they're basically making it all manual. Modular creative content allows you to have pre-approved content that is already pre-set up so that it can launch right away. And so instead of it taking four to eight months to launch all the derivatives and permutations of that content, it could now be done with less than four weeks, which includes the entire compliance component of getting that ad unit approved.

Bozidar: All right, thank you. So the main benefit from my stand is time and I would assume that there is also a cost efficiency benefit linked to that. So it's time and money. And I guess with technology, I would dive deeper into that quality, basically less mistakes because you can spot them faster than relying on human being. Now Word documents. It's really interesting to talk about Word documents. So is it still that if you are producing a video and audio, you actually have to submit the PDF? Is it still how it works?

Diana: You have to submit everything. Everything has to get submitted to the compliance tram. And the problem is once the compliance team marks up all of those changes, it needs to be redone and then resubmit it to all of the compliance teams again. The problem that I'm seeing is the back and forth is being done all manual as well. So a person is taking a screenshot of all of the ads and creative, the imaging and the messaging, and then laying it out, usually on a PowerPoint or on a deck, and then sending it to a compliance team. And then the compliance team looks at it and then basically marks up the changes that they want. So A, B, C is approved, D, E, F is not. I want these changes. Send it back and then they basically have to re-edit, redo it all manual, place it on a PowerPoint again, send it back, and that back and forth is happening constantly in trying to getting the creative approved for all the different formats.

Bozidar: Well, it sounds like something I call it content choke. It sounds like something that would create basically content choke. You just give up, which combined with not maybe having enough competence around content operations can lead to giving up on creating content except for when you have to have it, like visual aids. I'm saying this because a lot of time pharma companies, when you say content, they think visual aids. And I'm like, we're in 2023, we should be thinking a little bit beyond visual aids if we want to engage doctors.

Diana: You are taking the original vis-aid and then making derivatives of the vis-aid and then being able to submit it automatically.

Bozidar: So how does your company do it and how does your technology solve a problem or part of the problem?

Diana: So what our company does is we take the original vis-aide and we're able to make all the derivatives of that vis-aide within seconds time. So I know this sounds weird, but artists no longer actually have to be involved. With true automation, you could pre-template all of the different designs that go into a library that's already pre-vetted. In other words, all the logos, the imaging, the background images, the header, the messaging can all be pre-loaded into a machine. And then MLR could actually review the process before it actually goes inside of an ad unit. So think of a way where basically the compliance team can actually visualize what the template could be so marketers don't actually make the template unless the review already happened with the compliance team. And then afterwards, we have an automated system that puts all of the images into a PDF automatically. And then it shows every variable of messaging, header, disclosure, images, and it lays it out and all the variation that it can run for every format, health endemic formats, Google, Meta, Programmatics. And it's able to lay it all out and then submit it. So we automated that entire process.

Bozidar: Oh, that's amazing. Sounds like music to a lot of MLR and Brand Directors, ears and AORs. Now, more technical question to dive deeper into that, what about, say, video content? I think even the FDA, when they provide guidance on that, they say you have to start with the script first. So for videos, start with that, and then you layer other pieces.

Diana: Now we're talking about generative models and even Predictive Modeling, right? So we have now the capability that background images can now be made in a second's time just through text. So if you reverse that exact same thing, you actually have the capabilities to automate that entire process end to end, instead of it going to all the different hands and manually doing each piece of work by a human. And so this is what I'm saying a lot of times, humans make mistakes. And the reason I invented this is because I had creative artists that said, I didn't go to Pratt and Parsons and the best art schools of the world in order to do permutations and derivatives of ads. They said, I quit. And so nobody wants to do this work. If you have an employee that's taking screenshots of all the derivative ads, putting it into a PowerPoint, submitting it to a compliance team, and then the compliance team says, A, B, C is approved, D is not, and then that person says, but D was approved last week. Well, I changed my mind because the word proven and demonstrated and shown are three separate words and I've changed my mind. And the reason they say this is because there is no self-contained area in the world that has all the rules based on whether MLR or anybody is going to pre-approve that ad. There is nothing. So every human is making a subjective decision on whether they're actually approving ads or not. So I am married to the general counsel of Mate Beijing, but he was also the GC of North America for Santa Fe. And he was also a lawyer from Merck and also from Pfizer as well. The reason I invented it is because I am the marketer that couldn't get those ads approved. Going to him and trying to get these things done in a speedy way and getting it all approved and realizing the entire process was subjective. And that is what I fight for every single day from a marketing perspective. If I went to any compliance team and I say, why would you market this way? And I go to a different person on the compliance team, they would have two different markups every time.

Bozidar: And that's a really just a comment and it's a sad way to spend life. Like someone went to design school, this and that, and now they're doing this. So it's sad. A few things, MLR being subjective. Yesterday I had a conversation with one of partners, a big consulting company, and he had an idea for something else, but I use something similar actually where I write emails, where it gives me a score on the side. So if I'm a marketeer brand director, I'm just like typing this and then on the side, there is like, hey, you're 80, 90% likely to be approved based on general knowledge and data sources of how MDA thinks about should good MLR be. But then there was a subjective piece there. So is there something like that, like a model that is fed with at least general knowledge of MLR. So I check grammar, check whether it's all label or off label, check whether you say and or or, things like that.

Diana: Logo sizes, fonts, colors, branding, that could be all objective, right? And this is where I get frustrated. It doesn't matter if you hire the best artists in the world. From an agency model, you can have 20 artists, and all 20 is going to make advertising slightly different because humans can't duplicate exactly what the branding would be. No matter the best artists in the world, it will always have slight variations of branding differences. And what I'm saying is a machine doesn't make those mistakes. A machine is always going to get the font size right, the font colors right, the background images right, the branding right. So all of those things, you can prelock. And what I basically tell a lot of the pharmaceutical brands out there, listen, you have to start somewhere. And they said, well, we know the entire process is 100% subjective. Okay, but can we actually make a model that 80% of what we're doing right now is objective and 20% could be subjective? And that is what I find to do.

Bozidar: Yes, yes, yeah. It's human being, possible machine instead of just a machine. So, then you like, if you got 80% objectivity, you save, you know, a lot of time. And then the second thing, if you improve the workflow as you mentioned, so brand control, brand elements, guidelines and all that is already there and you don't go back and forth, back and forth, there is a beautiful easy workflow. And also, you know, I use one of the system when I was at Sanofi for something else and for the reviewer side, it did have help for the reviewer because they also don't want to spend their life fixing groundwork mistakes.

Diana: Exactly. So my head of Pharma, Melissa Gunn, is actually a lawyer. So she was one of the top lawyers at Pfizer. And so she saw it, and this is how we recruited her. She said, I used to lose sleep when I was the attorney looking at this stuff because she'd wake up and she's like, wait a minute, did I get all the disclosures correct for social? Maybe I missed this one, right? So there is no self-contained area. So can you imagine how great it would be if there was a self-contained area where disclosures has to be scrolling versus one click away versus basically one swipe? That's what I'm saying is that if it is rule-based, why can't it be all self-contained in the area for the MLR team, but also for the brand team, making it an objective decision for many of those rules? And then the last part I would add to this is we talk about Predictive Modeling. We're talking about generative AI. There is none of that if basically there is no self-contained area that has those rules. You have to start somewhere. If you don't have the data, there is no way to predict anything. There is no way to say this is going to be the best content to serve up because we have between the last five years, we've noticed that this is the best way to actually present that advertising. There is no self-contained area to give you that.

Bozidar: Yeah. So this is a really like a perfect area. My first reaction, technology can solve a massive, massive, massive problem. So both sides are suffering. That I understand. If I'm a brand director now, what do I start with? Do I start with the overall strategy and message architecture first? And then from there, I say, well, within those core strategic messages, then I would like to further develop these pieces of content. I look, I write them, have them as text, and then I can take that text and turn it into audio, video, visual, aiding for graphic, this and that. So what do I start with?

Diana: Yes, so it really starts with, instead of it basically being on a Word doc, it needs to be on an Excel doc. It needs to be in a way that we can extract the data and then form rule-based decisions. So basically, what are all the different formats that you need it for? What are all the sizes, those formats that you need it for? So do you need it for Doximity or Medscape? Do you need it for social? What are you going to use it for? Is it for websites? So all those questions in terms of what you need the formats for and the sizes of those formats is what we need up front. And then overall, what does the branding look like? What is the messaging on the brand? Do you want the disclosures to be scrolling one click away? Like, all those decisions can be made ahead of time. And then the final decisions are really understanding messaging. What can you mix and match messaging based on header, image, disclosure? And that's what we extract out.

Bozidar: So let's talk about the future that may happen pretty soon. We live in an era of AI, Large Language Models. We use it every day also, our team internally in Evermed, and I'm starting to get a feeling that the progress of this thing is so big, so fast, that many things will look dramatically different within a year. Three years I can't even imagine. Example, text to video. Now it's like, you cannot use it in farm. Give it a year, give it two years. It's going to create magic. So with those kind of tools, I start to think about magical scenarios. So magical scenario or a brand director, like no constraint scenario is, okay, so I have my drug approved or becoming approved. I have my strategy and my market research. I have a Death Message Architecture, Strategic Imperative. So I know what I need to educate the market on, guidelines, gaps in the guidelines, disease burden, and then my product, efficacy, safety, dosing administration, and patient support. So I know I need to talk about that need. Maybe there is a diagnostic something and then talk about my product. Okay, so that's relatively standard with different variations. So let's say I know all that. I know my label and I can feed it into a system. How can I now magically, and it is the right way to think, magically say, well, I know the gaps in knowledge of doctors and I know the segments of doctors. I know what are the things we need to push in the markets for each of these segments. Can I push a button and have creatives, text-based, audio, short form video, using avatars of KOLs that I already have contracts with, created in front of me version one? In a way, that is 90% MLR approval. Is that crazy scenario that I'm asking for?

Diana: No, and also I would add more things to this. Most of the big pharma out there, they're looking for localized content. At the end of the day, they usually have a content system. So they're either using a Primo or Viva or they're using Adobe. Adobe is a very big one as well. So there's a content dam. The problem is the content dam is just an image and it doesn't localize any of it. So everything that they actually have to do is manual to the local markets. So if you actually have a modular content assembler, you can actually integrate it to your creative dam. You already have a creative dam, you can integrate it. With the integration, what it will allow you to do is based on the localized markets that you're launching in, you can now have pre-approved MLR content that has language in it based on that demographic area. And then the other part, you just mentioned it, localization and segmentation is key. So most sales teams, and this is where I get frustrated so you'll hear my passion on this, the sales teams and the marketing teams don't talk at all. Even though they may all report into the chief commercial officer, they don't usually have a lot of contact with each other. And that is where it all starts falling apart because you can never do omni-channel marketing if both sides of your platform don't talk to each other. Meaning, the sales team always works out of a CRM and typically a Viva as the CRM. And so that Viva CRM has all of the information in terms of who the customers may be, the HCPs may be. The problem is when a marketer is launching on all the different types of channels, they're not actually pulling the information from the CRM system. But that CRM system is very valuable to do audience segmentation. With modular creative content, not only can you have an integration with your creative dam, which could be a Primo, Viva or Adobe, but also integrated to your CRM as well. And what is that going to do? With the CRM/CDP Customer Data Platform, they have one. If you integrate it, now you can do Audience segmentation with all of the people within your CRM, bring that content over because it's already pre-approved through MLR, bring it over to the CMP, ours is Creative Management Platform. We already have the integrations to Meta, to Google, to Performance Max. We already have it. I think we're the only ones that actually made those integrations first. And it was because we came outside of health care. So our business unit launched originally from auto. And because of that, we became Ad-Tech Partners with Meta and all of those types of companies. We already have that integration. So we take that modular creative content, assemble it based on the audiences that's in your CRM, then bring it over to the Paid Media side. And now you can target the Paid Media side from your CRM, the modular content that's going to go out to them as well. That is omni-channel marketing.

Bozidar: I loved it. Nothing to add. Yeah. And I think that the order of everything you said, you know, I'm thinking always like first audience and segments of the audience, which CDPs are really good at, to segment of audience, coupled with that market research and strategic documents that pharma has. Because once you have an audience, you're like, okay, they have the specific barriers to adopting this new product. Barrier one, barrier two, barrier three, four segments, that's 12 barriers. So, if I have an ad segment, and I also have MPI numbers within each of the segments, I would actually have, oh, I have 5,000 cardiologists, segment one, segment two, segment three. And I know now that I have these three barriers to overcome. So, I have content all around that. So, content is very targeted. So, you have all your segments, barriers to it, barriers to adopt, plus an MPI level number. That puts me in a really powerful situation to think like, okay, what should my content be? How do I now unleash my creativity in a way that gives me high chances of getting it MLR approved within a month and not within six months? And then the final piece of the workflow that you mentioned, for the content, it needs to be pushed to third-party places, whether it's an ad somewhere like Google, Facebook, Social Media, or it's a Medical News Journal website, or for third-party, you could additionally, it's helpful that you can push it directly. So, it's end to end. And if you use a first-party content hub, like we power some of the pharma's first parties, portal's content which already need better content, then you can also push it there. So, then you have this beautiful end to end, I want to understand the process where, did I get that right?

Diana: That's right. And the only part that's missing out of all of that is pharma data as well. So, I always look at it as you got the content down, you got the CRM, and then you also have the pharma data. Now, that pharma data, if it's Viva, it could be Cross Saxs. But almost every pharmaceutical company, if they're big enough, they have their own pharma analytics data. It could be there's some form of IQVIA in there, some form of Cross Saxs data. They could use other data sets that they're already using because their data analytics team is putting that all together. If you can mix and marry the pharma data, the CRM, pulling the Content Library, and then basically assembling the modular creative content and then serving it to the Paid Media side, the metrics from the paid media side can be pulled out, matched and married to the pharmaceutical data that's also coming in. So, we do Predictive Modeling, right? There's a couple of companies that do it, ZS, Oktana, they also do Predictive Modeling as well. What they're not doing is tying in the Paid Media side of the marketing that actually goes in in order to predict that information. Mixing and matching modeling has everything to do with having as much data sets as possible to inform that decision. And that is why I keep saying there is no omni-channel marketing if you actually exclude the mixing and matching of that data. It is basically because sales teams are operating right off the CRM and marketing teams are all operating on the paid side. But there is no mixing and marrying that data and there is no omnichannel marketing with that mixing and marrying.

Bozidar: So, that's how a CDP with mixes and matches first-party, third-party data.

Diana: That's right. And then bringing it into the modular content system that can assemble all the pieces that you need and then it can actually go back into the CRM in form of a video with an avatar that you're talking about and that basically tells the doctor exactly what they want to communicate back to the CRM but they can also be pushed into the Paid Media side and then basically targeted to that doctor as well.

Bozidar: Yeah. Side thought to this, not related to Pharma. I think we're going to see massive exponential application of what you just described during the next year's US election campaign. Because you know what happened eight years ago when there was massive fake news polarizing people, all those farms and all that. Now with GPT, it's very hard to imagine for me they're not like already assembled teams from different intelligence agencies that are ready for the next 10 months to start dropping fake videos. I've also heard some of the top AI leaders talk about it a few times, like assembling videos.

Diana: It's already there. We already have it. It's like literally I only need to record one video with one person and from that video I can use the likeness of the voice, the face and everything and then spit out any messaging I want. That is like done now. It's instantaneous that we can do that now.

Bozidar: Do you think that KOLs would do what musicians start to do, which is get paid, say, this is my avatar and I reviewed all the content? So me as a doctor saying, wow, this is a fake thing. But then I say, oh, this doctor says that they reviewed all the content. So why do I care if it's a real doctor, it's an avatar?

Diana: I definitely think that that is the wave of the future. If you can basically get actually the speaker of the video to actually pre-approve all the. And then you also have the pharmaceutical companies that agree on the messages of that content that's already pre-approved. That's easy. Now you can actually generate content instantaneously. And the thing that I notice a lot of times, even on the D2C side, a lot of times that user generated content is one of the best, most powerful, effective ways to market people. But what I notice is it's so tiring to make that user generated content because it's all being done through a phone, your camera, right? And you just constantly have to record, record, record, record, right? And some of the content is horrible. So basically there's a lot of garbage content out there as well. Now imagine a world where all the content no longer has to be pre-recorded. It could actually be done fully automated. And now the content can be spit it out within a second's time. And you actually have the analytics behind that content to be able to track the doctors and your deciles to know which ones would actually react to the best messages based on the format that they're on. So what we notice is that social media works. It really does. In pharma it does. But it works with female doctors from the ages of 27 to 34, the best. The male doctors from the ages of 47 to 54, they are not on social media. They're never going to see it on Instagram. It's not going to happen. Whereas all to the doctors when I did all the analysis that go to conferences, many of them are older men, over 60 that actually go to those conferences. So what I'm saying is pharma marketing is not segmenting enough. It's a waste of money to launch a social media advertising ad for all the HCPs in your database from the ages of 27 to 60 because three quarters of that audience will never see it in social. And that is what I'm basically saying. You need to launch three sets of advertising, one for social for the ages of 27 and 34, female sampling works great for doctors, male from the ages of 47 to 54. And then obviously for speaker events, if you don't have your KOLs that are actually the same demographics of the people that are actually going to attend, they're not coming. And so these are the things that I see in marketing. Don't do broad-based marketing and think that's going to work. 

Bozidar: Yeah. So which are the segments? So when you use ads, you said social, you go from which age to which age, which gender.

Diana: It could be sampling, it could be lunch and learns, it could be all the different speaker events. But what I'm saying is there's a certain profile of HCPs that do attend those events or want communication that way. What pharma marketing does is they basically make the same message, they don't care about the segment, and they launch it to everybody in the same exact way. And the only way it works is micro-targeting. Micro-targeting is going to work, but the problem is you cannot do it without modular creative content. Because how are you going to micro-target?

Bozidar: Yeah, it's hyper personalized targeting hyper personalized content. So that's where we are going with all this. So the Doctor Smith gets a message for Doctor Smith at a time of the day using the channel that they want. So that right thing, right content, right time, right person at the hyper like 101 level. That's a dream, I guess and ability to create it quickly before Large Language Models and talking about this will be like, yeah, maybe in 20 years now that you see the power of this what we're discussing right now, the hyper hyper hyper. I don't think it's too far.

Diana: It's already happening outside of pharma, right? So we are doing it very much so outside of pharma. We're doing it in the regulated industries. So to me, people on pharma have to catch up.

Bozidar: Yeah. And then true on the omni-channel. I loved it. How you explain it. There was a news yesterday, a QVR prevented from buying Deep Intent because of FDA or FTC. That's seen as monopoly. What was your comment on things like that?

Diana: It is a monopoly. It's a monopoly. So I think that there are a couple players in the pharmaceutical industry and they do business with all the companies in pharma. And I think it is a monopoly for many reasons, right? It's not just the fact that they do business with all of the pharmaceutical companies. It's because it's very hard for innovation to happen when there's so many rules and regulations and overall, you can't innovate when basically you don't allow smaller companies to actually bring that innovation to the table. And this is where I will champion all the different types of innovative smaller companies that are going into the pharmaceutical space. If we basically get squashed down by monopolies that are out there because of the fact that they have mostly all the power, then there will be no innovation. Because think about it, from a bigger company perspective, if you got 30,000 employees or even 10,000, it's going to take an army of people and time in order to change one process. And I see it all the time, even in my own company, the bigger we get, the more decision makers there are and the more layers there are to get an approval. So let's say a new product comes in and let's say it's machine learning. In order for everybody to agree that's a decision maker to pass that for a bigger company, it could take years. For a smaller one, they can actually turn that around, make that decision happen within a week, launch a prototype within two weeks, and then now you're headed into market. And so what I'm basically saying is, with bigger companies out there that do have monopolies, the time that they take to innovate is going to be a lot longer than a smaller company that has a decision-making power to prototype something and learn about all of the different ways that it could work to be much simpler and much faster. And so these are the things that I fight every day is innovation and innovation can only happen if you can make fast decisions, if you're nimble. And I know Mark Zuckerberg used to say, you break things, and I'm not saying you break things, but what I am saying is what he was probably interpreting is that you have to take the risk to a failure and just get it done. And so that's what usually smaller companies are willing to do very quickly.

Bozidar: Yeah, because you know, I understand I've been in big corporate and I'm an entrepreneur, a co-founder. The failure is baked into getting to the result because you know that 80% of things you touch if you're innovating will not work. But those 80% are basically part of the journey. So it's not even a failure. I mean, it took me also when I switched from corporate to being more an entrepreneur within the corporations, but then being a co-founder to really not think about failure in the same way. So now I'm excited and understand the concept of failing fast because it's really about getting answers to whether this is the right direction to go or not quickly. So failing fast and go back and then change the direction. So I think that that whole notion, of course, smaller companies will do always better. There are no layers of decision making. There's no politics. There is no how many people get the promotion for when. So it's like a very different story. Great. This has been a great conversation. We'll have another episode very soon. I want to ask you a few questions about you. So listeners get to know you better. What's your favorite industry buzzword of the year 2023?

Diana: Omni-channel marketing. And I have one that's what? Micro-targeting.

Bozidar: Microtargeting, micro learning as well. I hear that very often. We used to talk. What's the best book you've read in the past year or two?

Diana: I love books on entrepreneurs because I'm an entrepreneur. And so I love books like Shoe Dog, which is the Nike story. And the reason is because I feel like as an entrepreneur, you die a thousand deaths before you actually succeed. And so to hear all the trials and tribulations of what an entrepreneur has to go through in order to grow their company, it's very inspirational. So another thing about Constellation is we are seven years old exactly. We were number 65 fastest growing company in the US. We won EY Entrepreneur of the Year in 2021. We are a minority and diverse company. We were number 10 fastest woman led company in the country as well. But you could never predict COVID, inventories, supplies shortage, and the stock market crashing last year. So it's been a lot of ups and downs the last seven years. And so I love to hear about other entrepreneurs that have had those issues and stories. And one entrepreneur that just really recently sold his company for, I think it was close to a billion dollars. He said, I have a taste for shoelaces. I said, what do you mean? He goes, as an entrepreneur, I was kicked down so many times, kicked in the face that I would fall and I'd have a taste for their shoelaces. And I was like, oh my gosh, I so understand what you're saying. You hear a thousand no's before you get the one yes.

Bozidar: Yeah. And just one comment there. So I'm also part of different founder communities and groups and software groups and so on. And there are patterns, right? There are patterns and it seems whenever I speak to founders and those patterns emerge, I'm like, we're in a video game. There's no way whoever designed it says, if you're an entrepreneur in this part of the video game, you will go through this suffering and you will find some also pleasure in it because you're thinking you're overcoming yourself as you're going. Because also relationships, entrepreneurships and parenting to me are the three ways for exponential personal growth because they bend you like in so anyways. Yeah, I feel like we could talk about these for hours. And then what type of music do you listen when you need some inspiration?

Diana: It's funny because I'll listen to some of the craziest songs like Eminem to like collapse. 

Bozidar: Oh, okay. 

Diana: Started at the bottom, music like that, because again, it's inspirational, right? It's like finding like people that have gone through difficult times, but also have not only survived, but thrived in those journeys. So I always like to listen to songs like that because again, it's all relatable content to me.

Bozidar: Good thing. Yeah, I like how you approach together and full circle to content. And where can people find you online?

Diana: Obviously we have www.helloconstellation.com as well as we also have our Twitter, Facebook, and our Instagram channels as well. So it's Hello Constellation.

Bozidar: Hello Constellation. Well, this has been a great conversation, highly relevant topic right now, so thank you.

Diana: Thank you.