What’s up folks, today we have a super fun conversation with Pratik Desai, Founder and Chief Architect at 1to1.
- Pratik’s a Rocket Scientist turned Martech personalization expert
- He’s armed with a bachelor’s from Rutgers in Aerospace and Mechanical Engineering
- He got his start at Accenture in Technology Consulting and later J&J in consumer apps as a digital product manager
- He later took a deep dive into Martech when he became Lead product manager at PVH focused on Salesforce Marketing products
- This led him to spend 3 years at Salesforce where he worked his way up to Personalization Practice Lead (Head of Delivery Services for Personalization)
- Most recently, Pratik started his own agency called 1to1 to focus on personalization strategy and implementation
- He also runs a weekly AI Discussion Group to help folks keep up with the fast changing landscape of Curation and Generative AI
- He’s a well traveled, trivia loving full stack developer
Pratik, pumped to chat with you today, thanks for your time!
From Aerospace and Sci-fi to martech and personalization
Pratik, you have a degree in Aerospace and Mechanical Engineering as well as your pilot license, is this all a backup plan for AI takeover and you naturally shift to space exploration and interplanetary marketing? 😆
Pratik’s answer:
- Aerospace industry wasn’t as mainstream when I graduated and the lucrative Aerospace jobs were in defense. I struggled to see myself going down that route and…
- Accenture does a damn good job of recruiting engineers out of Rutgers
- Luck is taking the opportunities as they present themselves….which really just set the tone for my career for the next 10 years
- The pilot's license came after! After a few years of working in technology, I started to miss the thrills of aviation and decided to get a private pilot's license. The feeling of freedom you get when you start traveling is exponentially increased when you actually fly yourself there!
What does aerospace and martech have in common?
In preparation to transition to my next question we asked ChatGPT what martech and aerospace have in common, it said.
- Data-driven decision making: In both cases, the ability to collect, analyze, and make decisions based on data is critical.
- Technological advancements and innovation, specifically use of simulation and modeling tools: Both fields need to stay at the cutting edge of technology to be effective.
- Problem-solving and customer-centric approach: Both fields involve solving complex problems while keeping the user in mind.
- Integration: Whether it’s engines, avionics, control systems or landing gear or if it’s CRMs, CMSs, CDPs and MAPs… Both aerospace and martech involve the integration of multiple systems and components.
Which one would you pick?
Pratik’s answer:
- Studying engineering definitely sets you up for success in so many different industries. The problem-solving coupled with the data-centric decision making puts you on a path that really helps you excel
- But the biggest parallel to getting things right would be integrations. In Aerospace Engineering, there are SO many systems that have to work together and if they don’t, the outcomes could be catastrophic.
- I can’t tell you the amount of MarTech implementations that I’ve been apart of where integrations don’t get enough love, for various reasons:
- The source or destination system is owned by a team that wasn’t informed of the transformation
- The IT team has conflicting priorities
- ETL transfers are easier, so we’ll start there - and it just never becomes priority to make things real-time
- Etc
Science fiction
We’re huge fans of science fiction on the podcast, so I’d be remiss not to take a short turn here. I made a big assumption here but based on your field of study I guessed that you are a sci-fi fan… I’d love to get your list of favorite science fiction books or movies but more importantly I’d love your take on the speculative future of personalization and what that looks like according to Pratik?
I recently read All Our Wrong Todays by Elan Mastai and in one of his alternate future timelines he describes a world where advertising isn’t just 1to1, it’s also tailored based on your mood that day, what you had for breakfast, events on your calendar next week. The protagonist’s big idea is to offer consumers a flat fee to opt out of ads completely, but it’s a big flop. In that world, consumers actually wanted hyper tailored ads.
Pratik’s answer:
First, I absolutely LOVE time travel stories. I think we, as a society, have learned so much about how the physical world works - it’s fascinating to see how movies/books start to build out the rules for things we don’t understand. With time travel, time dilation is a starting point - but then you’re really free to start building your own rules. Are we in a multiverse? A fixed timeline? A dynamic timeline?
Some of my favorites in how they build out the rules and create logical consistencies: Primer, Interstellar
That being said, I TRULY agree that consumers crave hyper-personalization down to minute by minute desires - in my mind, the BIG question is whether or not you’re personalizing to remove friction and promote tailored discovery OR are you personalizing with the intent to misinform, and consequently influence outcomes. It’s SUCH a fine line and intent is crucial.
My network and I have spent a lot of time thinking about this - to the point where we even had Y Combinators attention for a bit on a universal preference center. The rules of engagement are the problem because they’re so ambiguous:
- Where and when does personalization begin?
- What opt-in and opt-out ability exists without adding additional friction?
- How do you balance guiding and promoting discovery with the desire to change behaviors?
- How does a customer's willingness to accept personalization change from e-Commerce to media companies?
- What control, if any, does a customer have over their personalized journeys with you?
- At what point should you trigger customer awareness that personalization is or isn’t happening?
The road to building and launching 1to1
The human side of launching a martech agency
You founded 1to1, your agency in Oct 2022, at the time of recording you’re about 9 months into the journey. You’ve already surpassed what? 35 personalization implementations!
Talk to us about the human side of this journey so far, how have you managed all the sleepless nights, the mistakes and all the contract negotiations?
Pratik’s answer:
- 1to1, to date, has managed 35 MCP implementations ranging from eCommerce to financial services to the streaming industry
- The sleepless nights are real. In the beginning…. I was staying up worrying about where our work is coming from next. Now, I stay up worrying about how to fulfill the amount of work we’re partnering on. I’ve evolved to better sleepless nights
- I’ve made so many mistakes and I plan on making so many more. I think the beauty of working in MarTech is uniquely understanding the power of experimentation.
- I’ve learned SO much from all the mistakes I’ve made and that's allowed me to improve my processes, our codebase, our pitches, etc.
Starting over
If you were starting 1to1 today, what’s one of the biggest things you would do differently?
Pratik’s answer:
Definitely starting sooner - and taking more risks. I have a lot of conversations with aspiring entrepreneurs - and I’d say the biggest roadblock of folks starting is their fear of failure. The irony to that, is failure is such an important part of of success - I see a lot of great ideas and great teams walk away from what they’re doing because of paralysis before they even start.
I probably would have started sooner - and I was definitely a little trigger shy on certain conversations and felt I couldn't figure out things like taxes, insurance, etc.
That being said, I kind of feel things might’ve worked out the way they were supposed to - If I started earlier, I think I would’ve hired folks into roles that Generative AI companies would have replaced shortly after. Because I started when I did, by the time I needed those roles - the solutions existed to solve for them
- Project Manager - Spinach.IO + Butler
- ChatGpt - Junior Developer, Proofreader, Contract Writer
Data and prereqs to personalization
Data excellence
I think you’ll agree that one of the precursors to doing anything related to personalization is data. It’s the main thing holding most companies back, not the actual tech, that can come later. We had a full episode (
#66) on data models and pipelines and tried to introduce that to marketers.
What’s your advice to marketers to help them bridge that gap and can you walk us through a few real life examples of data going wrong?
Pratik’s answer:
I always say that Data Excellence and Operational Excellence are the foundational elements for a omnichannel personalization program
First, the data you have is:
- As important as the data you don’t have
- Only as valuable as how real-time it is
Example story: We recently had a major win of a digital campaign where we ended up moving a product that was marked promotable with a certain amount of inventory by 10x its normal velocity. Well, our inventory was delayed and the product was not actually in stock. According to our marketing dashboard, we had a huge revenue uptick - but the merchandising team and the customer support team spent the entire weekend figuring out how to rush order new product and/or send appeasement coupons for canceled orders. 1 amazing marketing campaign, cost the company a lot of money because of incorrect data signals
Second, the data you have, needs to be structured in a way that meets the needs of your business:
Example story: We’ve had several customers who bought personalization engines (both Salesforce and not Salesforce) who then rushed to load their data and created undesirable consequences.
- End-customers who were recommended products without proper style context
- End-customers who were provided recommendations that they did not qualify for
- Etc.
Those are awesome examples, it’s important to partner closely with your data team, try to get to eye level when it comes to understanding data flows and data integrity and test everything first. I like how you coined it data and operational excellence.
Personalization roadmap
You built a pretty awesome roadmap to help folks see potential use cases and charted those for complexity and business value and help illustrate where folks might start and what might be less complex and more effective
Can you walk us through it?
Pratik’s answer:
Appreciate that! We’re trying to do this per vertical
We found a lot of marketers entering the world of omnichannel personalization focusing on 1 or 2 use cases that they’re hearing and seeing across their industry.
The problem with this, is that what works for 1 company - may not work for another….everyone has different pain points across their funnel - so they fail to realize ROI and then abandon the omnichannel strategy altogether.
The roadmap is, from our experiences across all of our collective careers, a take on how impactful these omni channel use cases can perform for the retail/ecommerce vertical.
It was our attempt to show that there are use cases that are talked about alot - but here is the average business value you’d expect OR here is a use case that don’t get alot of discussion, because of the complexity - but here is the potential ROI you could see.
The slide is generally an entry slide into a workshop where we then individualize this for a client through an intensive data-driven workshop. I don’t think we have time to necessarily go through what that workshop is, but here is an example of a recent customer.
Example Story: We had a client who was stuck on the idea of Abandon Cart and Abandon Browse - it works so well for other customers, it makes sense that is what they would want to go after first. However, after truly analyzing their data - we found that they had a major onboarding problem. Yes, the abandonment problem was there - but over 60% of customers who visited their site, never even looked for a product to abandon. Let’s say they had a 2% conversion rate on the 40% that end up looking at something. The question is, if I run abandon journeys and up the 2% on the 40% to 5% on the 40% - is that a greater use of my time versus pushing 20% of those 60% bouncers into the 40% that view and convert 2% of the time?
Long story short, we went after the onboarding problem - pushed more folks further down the funnel AND THEN focused on the abandon problem to uplift a larger population. Focusing on abandon first and then the onboarding problem would’ve gotten us to the same place - but we would’ve captured a lot less revenue in the process of getting there.
Can AI replace everything a marketer does?
You wrote about integrating OpenAI to MCP. You’ve clearly been on the cutting edge of AI and marketing applications.
What do you think are the challenges that AI has to replace everything a marketer does today?
Pratik’s answer:
1) AI will take time to replace what marketers do - even when it has the technical feasibility to do so. I work with organizations today who have set up sophisticated AI engines - but still apply marketer control over them. Why? Because 1) There is fear that AI won't get it right - so there is a heavy need to test into things and 2) for industries outside of eCommerce - getting things wrong sometimes isn't an option. In financial services and healthcare - you can't accidentally provide the wrong information with the expectation of converting the other 85%
2) But - when it does get good enough, I think we can break down the work into 2 categories. Curation and Generation. Generation AI will generate Copy and Content for the Curation AI to target the right users, at the right time, with the right AI generated content. Where does this leave the marketer? I think, within industries of low regulation - marketers become critical prompt thinkers and in industries with high regulation, they also add the job title of "AI regulator"
*runs to add AI regulator to his LinkedIn headline 😆
Warehouse native martech apps
This is where I wanted to tie your ETL comment to at the start of the episode. There’s a trend of Martech tools that are called connected apps or warehouse-first, basically sitting on top of your data warehouse vs creating yet another copy of your user db. This is a big shift from needing to have different databases of your users and connecting everything via individual API integrations. What are your thoughts here and how long until other tools catch up?
Pratik’s answer:
I think of Martech products as a tool in your toolbox. The right tool ultimately depends on the problem you’re trying to solve.
Connected apps are another tool within a toolbox and they’re fantastic for the problem they’re solving, but we - as Martech practitioners should be clear on what that means - I think a lot of the conversations right now have been muddied by - ironically - product marketing strategies.
Let’s talk about the elephant in the room - Composable CDPs versus “Traditional CDPs”. I think this argument is stemming from the desire to package Reverse ETL tools into a way that sells to Marketers - but that then causes a lot of confusion.
If a Reverse ETL can replace my CDP - then how does it solve identity resolution? Since it doesn’t and that would need to be done by the owner of the data warehouse - immediately we’re not talking apples to apples.
So an organization that has their Marketers in a position to work with their data warehouse owners and get their data structured in a way that works for them - Reverse ETLs are a great tool - but for those Marketers that are still fighting for a seat at the data table, they still have operational work to do before replacing their CDP make sense.
Let’s not forget that different databases of your users were solving a problem - not accidentally creating one. Long story short, the Modern Data Stack makes sense - assuming the organization has taken the time to create a composable organization - assuming that the organization is ready to look at problems differently and break old patterns.
That being said, I don’t want this to be confused with me saying that composable products AREN’T of value. There are plenty of organizations that could benefit from these types of technologies today and plenty of organizations that might even use these types of tech as an impetus to change their ways towards a true single view of their customer - which is the ultimate goal.
That being said, I guess the question around SFMC catching up depends on if they need to? Do we believe all potential organizations will always have the same problems that can only be solved by composable architectures? It’s tough, I do believe we’re moving more and more into bringing customer data into a single database - but I’m also on the ground seeing organizations really struggling with getting their siloed teams around this type of way of working. The larger the company, the harder it is to do it.
Our take:
Love this topic, tons to unpack here. I don’t actually think rETL tools are trying to convince marketers that they can replace packaged CDPs, rather they are simply a player in a long list of components that can make up a custom composable CDP, which would include an additional solution for ID resolution and a way to collect 1st party data. Like you said, different tools solve different problems and sometimes a solution requires more flexibility than what a legacy CDP could offer.
Marketers fighting for a seat at the data table. Love the way you put that. I think that’s where both marketers and data teams need to learn more about their respective counterparts. Ultimately the end goal is growing the business and both teams should work together to hit those goals.
Happiness question
You’re an agency founder, investor, husband, dog dad, sports enthusiast, avid backpacker and armchair space explorer… you have a lot going on.. One question we ask all our guests is how do you remain happy and successful in your career? How do you find balance between all the things you’re working on while staying happy?Pratik’s answer:
It’s funny, a lot of folks have asked me how it feels to build your own business and somewhere in the response I always say “and if I fail, I can always get a job”. My wife is always quick to remind me “you have a job”.....but it really doesn’t feel that way. I’ve been able to capture my entrepreneurial spirit, with my drive to solve problems, along with my passion for AI and honestly, it really doesn’t feel like work.
That being said, recharging is always super important ….I do that by spending time/Traveling with my wife and my dog.
I think I’ve always had a really hard time celebrating the small wins and acknowledging accomplishments - my wife forces me to slow down and recognize how far I’ve come and take time to switch things off
Oh and….I stay off of social media!!! I found myself competing with folks on social media alot and that got really toxic really quickly. Instead,
I’ve started to focus on being better than I was yesterday. Ultimately, that made all the difference.