Tejas Manohar was at Segment when it sold to Twilio for $3B. Now he’s co-CEO at Hightouch, watching marketing teams make the same mistake over and over.

They’re using AI to make their old playbook faster. Wrong move.

“You can’t assume that people are going to interact with your website the same way they interacted with your website a few years ago,” Tejas said. “Consumers are going to be viewing their travel planning in ChatGPT. If you’re just optimizing your website, that’s not going to work.”

The marketing calendar worked when you could only ship 10 campaigns a quarter and customers had three alternatives. Now AI can generate thousands of variants and customers are researching in ChatGPT. The constraint was production capacity. That constraint is gone.

March 3, 2020

Before Hightouch, Tejas built a travel company. March 3, 2020, every metric went to zero. He shut it down.

“What do you want to be doing with your time? Do you want to wait out a pandemic or do you want to do something? We just had to make the decision to ignore sunk cost fallacy.”

Most marketing teams are in this moment right now. Your playbook worked. You know how to execute it. You have dashboards that measure it.

But every signal says it’s broken, and waiting won’t fix it.

Thanks for reading Playbook Broken! I’d love it if you would share, re-stack or comment!

Why Personalization Actually Failed

We’ve been lying about one-to-one marketing for 30 years.

CRM was going to deliver it. CDPs were going to deliver it. Neither did.

“We’ve had data for a really long time. What’s really different now is understanding each customer.” Before, you dictated the rules. This segment gets the credit card offer, that segment gets subscriptions. “You didn’t have a technology that could actually look at every user and think about the individual journey for them.”

AI checks a hundred times more variables than any analyst. It experiments with offer sequences, timing, amounts, ordering. It figures out whether $10 or $20 works better, and whether that varies by purchase history.

The catch: “It’s a big change to go from a world where you’re releasing five, 10 marketing campaigns a week to a world where you have this huge database of content and this AI system is helping decide who to send them to.”

Entirely new workflow. Entirely new org chart problems.

The Conversation Nobody Wants to Have

Technology is ready. Your organization isn’t. Think about Martec’s Law which I seem to invoke all the time lately. Sure, the tech is cool, but can your people and teams change fast enough to keep up?

“Just having new tools that do the old process a new way doesn’t cut it. You need to get one that has a new mindset of how we can use AI in a different way.”

Merchandising wants their new product pushed to everyone. Brand spent three months on messaging. The CEO has quarterly goals that require specific numbers. All of that runs into an AI that says 75% of your customers aren’t ready yet.

“Your merchandising leader who is only responsible for one product, they’re not going to be happy when you’re not sending it out to all their customers. It’s a hard conversation.”

This is where it dies. Not in the tech selection. In the hallway where someone tells the merchandising VP the AI overruled them.

Like the Playbook Broken vibe? Why not subscribe?

What’s Ready, What’s Not

Tejas runs a company selling this stuff, so factor that in. But his read is right.

“AI is actually really good at analysis of data and getting insights from data. That’s the functionality that is really reliable today.”

Content generation? Not (quite) ready. Fully automated creative? Not (quite) there yet. But close and getting closer.

But AI figuring out which customers want which offers, when to send them, how to sequence them? According to Tejas, that works now.

“You shouldn’t be manually running all these queries to figure out what drives your customer to high LTV. You should be asking a system that uses AI to tell you.”

And the downside is lower than you think. “Let’s be real. The current state is that you’re sending a lot of things that are likely not relevant to your customers.”

20% open rate, 1% clicks. You’re already wrong 99% of the time.

It’s the Brain, Not the Interface

With every new model release the picture gets clearer. These are not just “upgraded” chatbots but pieces of systems that break down problems, reach for tools, and delegate work.

“It’s not just having a better chatbot interface. You need an interface where you can say a problem and it can help you break it down. It’s not the interface, it’s the brain.”

The calendar is dead. The question isn’t whether AI will replace your workflows. It’s whether you’ll replace them before your competitors do.

Your March 2020 moment has already happened, did you notice?

This episode of Playbook Broken was previously published on MarTech.org as “The CDP fantasy is over.”

Around the Town

* I spent the weekend in Las Vegas and now engineering and innovation is on my mind. We did a half day at the Hoover Dam, an experience not to be missed. The engineering and scale of that project is mind-blowing. From there, we had our first “Sphere” experience, seeing Zac Brown Band absolutely crush it. The Sphere is a modern marketing marvel, truly astounding to see it in person.

* Our AI for GTM even on Dec 3 was a blast! We had nearly 70 folks hang out with us from 9-5 to learn practical AI skills. Super fun. Excited to do another event.

* I’m nearly done reading Neal Stephenson’s 880 page Fall; or, Dodge in Hell: A Novel which, as with all of his books, is mind-bending in all the best ways. How about a virtual world where we go to become immortal and recreate meatspace? Yea, he wrote that. Let’s see if he can stick the landing. I’m 75% of the way home!

Thank you for watching, listening to and reading Playbook Broken!. If your playbook is shaking, we’re here to help rethink and rebuild it with you.

—Marc

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit marcsirkin.substack.com