AI, but make it good

What I wish more brands understood about AI adoption

Klarna was the first to fumble this: last year they replaced their 700 customer service agents with AI chatbots, and CEO Sebastian Siemiatkowski even bragged about cutting their workforce nearly in half by shifting to AI. Now they’ve been hiring back human customer support agents. More recently, Duolingo announced it would continue to outsource all its contractors’ work to AI. And Shopify said it would add headcount only when AI won’t suffice.

At every single one of these announcements, I cringed. It’s not because I’m anti-tech. I’m anti-brands who frame their operational efficiencies as consumer benefits. My friend Marisa Lather said to me at last week’s ContentCon conference, “I expect companies to keep up with technology. I’m not surprised to learn when a company is leveraging it to improve their offerings.” And I agree. I’m not shocked when teams use AI to streamline communications or debug code — much like I don’t even blink an eye to learn about writers using spellcheck and Grammarly before they publish their work.

I don’t know why I chose this photo, but it’s cute! // Photo by Eric Krull on Unsplash

As a tech worker, I use ChatGPT. And I’m good at it. I’ll use deep research to surface the articles I knew existed but didn’t know how to Google myself. I’ll have ChatGPT restructure an outline, brainstorm blog post headlines with me, and clean up a final draft before I publish. These are things that help me move faster, but I wouldn’t dare say that makes me an innovative marketer. Because efficiency ≠ innovation.

I had a feeling I wasn’t the only marketer with the AI ick. So I asked Gurdeep Dhillon, CMO at Contentstack, what he thinks when he hears of a company going “AI-first.” He readily answered, “Skeptical. As a customer, I don’t want to know if AI is involved.” Nobody wants to see the sausage get made, and right now, this is especially true of AI. As the tech continues to evolve and adoption becomes more widespread, consumers will come to expect some sort of AI integration. Maybe it’ll come in the form of hyper-personalized shopping guides, or the AI-built vacation itinerary that actually works. The goal then, for companies, shouldn’t be to dazzle us with their “AI-first” experiences. It should be to dazzle us with customer-first experiences… that happen to be powered by AI — done well and at scale.

So as of now, what does “good AI” look like? To me, it looks like…well… nothing.

At SparkToro (where I work as VP of Marketing), AI is part of our toolkit, and we’re using it to serve Conversational Queries to our users. ChatGPT offers the framework that allows our customers to describe their audience in plain English. But it’s our SparkToro (clickstream + search + social) data that powers the insights. We’re always going to be a marketer-first company and any way that we evolve our tool is always going to serve marketers, or our customers, first.

The way I see it, any company using AI has a new competitor — and that’s the consumer’s personal ChatGPT account. You know, the agent that’s trained on that individual’s knowledge and preferences. If you can deliver an AI experience that’s better than the ChatGPT I trained — the one that knows my hard-to-find clothing size, style, ideal price point, favorite brands, and who helps me decide whether a new clothing item is worthy of being added to my closet — I’ll bite.

After commiserating with Gurdeep over the healthy skepticism over AI, I was half-surprised to learn how Contentstack uses it. Turns out, 90% of the company’s prospecting is AI-powered, but fully orchestrated, vetted, and led by humans. It’s a business efficiency, yes, but not one that’s positioned as a value-add to Contentstack customers. It’s an efficiency that helps the team do their jobs more effectively.

One prediction that Ben Goldstein, Director of Content at Contentstack, has is the increase in strategic use of AI from efficiency gains to personalized experiences. He writes in the Digital 2030 report, “We're already seeing a consumer backlash against the use of AI chatbots that don't understand what site visitors actually want. At a certain point, marketers will be forced to ask ourselves: What would be good for the customer?”

Good AI doesn’t announce itself. It integrates. It uplifts the user experience without demanding attention for the tools behind it. For brands, that means using AI not as a gimmick or a headline, but as a quiet infrastructure shift that helps you serve your people better. The future isn't “AI-first.” It's customer-first — powered, quietly and competently, by AI.

Maybe this is the marketer in me talking, but I think every team adopting AI should pause and ask: “Is this genuinely better for the user? Or is it just better for us?” Because if the answer is the latter, it’s not an innovation — it’s a shortcut dressed up in buzzwords. And the people on the other end? They’re going to notice.

---

🧁 Petits Fours

A newsletter called “the Menu” wouldn’t be complete without dessert! Here are 4 things on my mind + worth your time:

🎯 The problem with one-off “best practices”: They don’t get guarantee you’ll get the same benefit. Your customers, product, and user journey are all different. Worse, you can’t apply it everywhere else in your marketing. That’s where my friend Talia' Wolf’s new book, “Emotional Targeting” comes in. She puts the structure behind emotional marketing, offering a repeatable system to boost conversions.

📆 The conference you really ought to come to: Surprise, surprise, it’s my conference! My team and I are hosting the second in-person SparkTogether conference in Seattle this fall. Unlike other conferences, our speakers don’t teach best practices and how-to’s. They share the hard-won lessons through their biggest business successes — and failures. My friends Jay Acunzo, Wil Reynolds, Scott Heimendinger, Asia Orangio, and others will be there. Plus, there will be a whiskey tasting. :)

🔎 Do this before your next marketing campaign: Audience research. Powered by real data and some AI magic. My boss Rand Fishkin shows you how. (Yeah, yeah, you should have a SparkToro account. But relax, you can just make a free one.)

🎧 Please listen to my marketing podcast: My friend Sonia Baschez and I break down the logic behind what goes viral… and what doesn’t. In this week’s episode we talk about athlete (and creator) marketing, Uber Eats’ TikTok-style reels, whether Deloitte’s recent Instagram video missed the mark, and wtf is a “wedding content creator” and is it worth it??

.