 | ||||||||
| ||||||||
| | ||||||||
 Hey Marketing Bestie, I just spent the weekend trying to plan a family vacation for the week between Christmas and New Years. Apparently I should have thought about that 3 years ago? Everything is booked? 4 months notice was better than I usually do but I have regrets. Where are you going? Room for 3 more? lol Was this email forwarded to you? Sponsored by Gradial I've seen it: when campaigns take weeks instead of days, you start dreaming smaller. How to Use AI to Find the Marketing Insights Hiding in Customer FeedbackQuick gut check: how many customer reviews has your team ACTUALLY read this quarter? Not skimmed. Not glanced at the star rating like it's a horoscope and moved on. Read. Like, with your eyes, all the way to the end. Here's the thing your CMO doesn't want said out loud: your best Marketing insights aren't locked in some 6-figure research study with a 40-slide readout. They're sitting in your support inbox, your sales call recordings, and your G2 reviews, quietly composting, because reading 400 reviews by hand is the kind of task you'd assign to a summer intern. AI fixes the "there aren't enough hours in the day" problem. It does NOT fix the "we don't trust our own data" problem. Let's talk about both, and let's not pretend this is more complicated than it is. Stop asking AI what customers think. Ask it what they actually said.Most customer personas get built the same way horoscopes do: a room full of confident people, a whiteboard, some vibes, and zero verifiable evidence. "Sarah, 34, values authenticity." Cool, Sarah, so does literally every brand deck since 2015. Meanwhile the actual answers are sitting in a spreadsheet nobody's opened since the Slack message that said "great data, will dig in later!" (We were all lying. We know we were lying.) Research on AI-driven customer insights found that companies generate roughly four times more unstructured data than structured data: open-text survey responses, support tickets, call transcripts, social mentions, review comments (Prospeo, 2026). That's not a filing backlog. That's a 2nd company's worth of intel, sitting in the walls like a secret room you forgot you built. So instead of vibes, try this: "Group these 100 customer reviews, support tickets, and/or survey responses into the 10 most common problems, desired outcomes, and objections." What used to take 6+ hours of manual tagging now takes minutes, and it's sourced. Now when a stakeholder asks "what do customers actually want," you hand them receipts. Get the exact words. Not your polite, focus-grouped paraphrase of the words.Here's the rule the best conversion copywriters never shut up about: don't write the copy, MINE it. The highest-converting words on your site were never going to come from a brainstorm. They come from a customer, mid-rant, in a review or a support ticket, saying the thing so specifically you couldn't have invented it if you tried. Marketers have a real gift for translating perfectly good human language into something that sounds like it was written by a committee that hates joy. Your customer doesn't say "streamline cross-functional workflows." Your customer says, verbatim, "I was so sick of copy-pasting this into 3 different tools I almost cried at my desk." Guess which 1 converts. Try: "Find exact phrases customers use to describe this problem." AI is genuinely excellent at surfacing raw phrasing across hundreds of reviews and tickets, the stuff that should end up word-for-word in your headlines, your ads, your landing page hero copy. Not because AI is a genius. Because your customer already wrote your best headline for you, mid-rant, at 11pm, furious about a broken checkout flow. AI just finally had time to go dig it out of the pile. Turn sales call chaos into your next FAQ pageYour sales team answers the same 12 objections on a loop, every single week, and none of it gets written down anywhere useful. Meanwhile marketing is over here guessing at content ideas, throwing darts blindfolded and hoping one lands. Drop a stack of call notes into AI and ask: "Turn these sales-call notes into a list of questions prospects ask before buying." Congratulations, you just built your FAQ page, your sales enablement doc, and your next round of objection-handling ad copy, in 1 prompt, before your coffee's even cold. If sales is fielding a question live every week, marketing should be answering it in public.Find the cracks in your competitor's story Here's the spiciest: everyone points AI at their OWN reviews. Almost no 1 points it at the gap between what a competitor promises in their ads and what that competitor's own customers say. Ask AI: "Identify claims competitors make that customers do not find convincing." That gap is your positioning. A competitor claims "effortless setup," and their reviews say "took our team three weeks just to onboard"? That's not competitive intel anymore. That's your next headline, and their own customers wrote it for you. The part where AI is not allowed to freelanceI said what I said: AI is a phenomenal sorter and a genuinely dangerous narrator the second you stop watching it. Ask it to summarize customer sentiment and it will hand you a beautifully worded insight that's 20% real pattern and 80% confident fiction. Skip data hygiene, jump straight to predictive modeling, and you get confident-sounding insights built on garbage, presented in a beautiful chart. Garbage in, garbage out, except now the garbage has better typography. So every "customers say X" claim needs a receipt. If you can't point to the actual review, ticket, or transcript it came from, it's not an insight. What AI actually gave youYou can't manufacture desire out of thin air. You can only channel what's already loose in the market. AI didn't hand marketers new customer data out of nowhere. It handed marketers TIME to finally go channel the desire, the frustration, the "why doesn't this just work" energy that was sitting there the whole time, ignored, in 342 browser tabs you swore you'd get back to. The teams pulling ahead aren't the ones with the fanciest tool stack. They're the ones who actually act on what the data tells them instead of just admiring it. Go read your reviews. They've been waiting. Possibly since 2019, back when we all still thought we'd "get to it next sprint." P.S. If your last customer research project involved a survey 3 people filled out under duress, this is your sign to just read the reviews instead MEME OF THE WEEK  🎙 TUNE IN Podcast to listen to this week: What you will learn: 🎙: Why hiring a "Marketing engineer" might be the smartest move you make this year JUST FOR FUN 🤩: Brand of the week: Sprite Can't wait to bring Ari to her 7th choice destination. Wish me luck! Your friend, | ||||||||
|
