Published July 27
Depending on who you ask, AI is either marketing’s savior or its most overhyped technology. The reality, as always, is more nuanced.
As Hightouch co-CEO Tejas Manohar tells Zac Wang, the technology isn’t infallible. It needs context and high-quality data to Distinguish between what’s right and what’s wrong.
As AI-generated ads flood the market, execution along won’t be enough. Taste will be the differentiator for brands that want to stand out.
What did you see in the market changing that inspired you to build Hightouch?
When we started Hightouch six years ago, I saw a clear pattern across all the largest brands in the world. Some of our customers today are DoorDash, Warner Music Group and Domino’s. When I spoke to all of these companies, the pattern I saw was they were all saying they want to do one-to-one personalization and market at the scale of their customers and the speed of the market.
But when I looked at the platforms they were actually using to run their marketing, there was a big challenge. Those platforms didn’t have all the data about their customers and their business.
What we discovered at Hightouch was a way to connect marketers to all the data that companies are putting into Snowflake or Databricks — these data analytics platforms — and help marketing teams use that data to run highly targeted, personalized one-to-one marketing.
We’ve had the trend of walled gardens centralizing the internet. Do you see the internet going even more toward centralization? Does that worry you?
I think centralization has always been a concern for any brand out there. And ultimately, it’s not centralization, it’s being disintermediated by these walled gardens and AI interfaces getting between you and the customer.
This is a dynamic we’ve had for years, even though everyone right now is wondering if there will be too much power centralized in Google or OpenAI or Anthropic.
I’d argue this is a trend we’ve seen for many years with search, with social media, with walled gardens.
Centralization has always happened. You have to figure out how to best utilize those platforms and, second, how to really gain that direct relationship with your customer and own your data to do so.
Has there been a piece of feedback, something about Hightouch, the product or the industry that changed how you do business?
The biggest topic for any CMO or large consumer brand right now with AI is how do you make sure AI is truly on brand?
What I’ve learned is that on brand isn’t just colors and styles, it’s the content. What are you saying? Is it compliant? Does it follow the legal guidelines of each market?
When a product is represented, is it actually the product, or is it the common issue that we see with AI models, where it’s a little bit off? There’s a little bit of hallucination and it’s regenerating the product every single time.
Does it actually follow the creative systems of a company? This is the top problem in using AI for marketing — especially in creative, content and campaigns right now. We’ve heard this firsthand from customers, and we’re pouring a lot of fuel into solving it.
I’ve seen this joke online “make no mistakes” when you prompt the AI. Do you think we’ll get to the point where it doesn’t make any mistakes?
I love that meme “make no mistakes.” It’s so funny.
Making AI outputs on brand for marketing isn’t about telling the AI to make no mistakes. It’s about telling the AI what is a mistake and what is correct. That’s the whole problem of context. Context engineering is a really hot topic in AI right now. The models will get better and better and better.
When you tell the models to do something, they’ll do it properly. But how do you actually tell the models what the right thing is to do? That’s the context that often lives in the head of a designer, a creative team, a marketer.
Companies have slide decks on their brand guidelines, but there’s a lot that’s not specified there.
There’s a lot of talk about an explosion of ads because it’s easier and faster to create them. Do you think that may actually make it harder, because there’s more content to train on?
In a world where I can create 100 ads in a day, or 1,000 ads in a day, what’s going to differentiate me from other brands? Taste is the differentiator at the end of the day.
It’s knowing which ad to create next. That’s a discipline. That idea of taste and actually having perspective is something we’ve found AI to be surprisingly weak at.
AI can really scale execution, but knowing what’s right when there’s no quantitative evidence, which is often the case in marketing, is what’s going to truly differentiate a world-class brand from one that starts to fall behind.
That was true before AI, and it’s true after AI. Now something on top of that: If you have taste but you don’t execute at scale, don’t modernize and don’t engage customers on all these new channels, you’ll fall behind. What we’re trying to help with is that transformation.