Happy Sunday! If you're reading this, I hope you're hydrated, you got a good sweat in this weekend, and you’re starting to get prepped for Q4. In that vain, I wanted to write today’s newsletter… hopefully something that helps diagnose why your ads may not be working, or how to think about trying it differently. It’s a fun read, and I hope you walk away with one or two nuggets to improve your work this week.

A few years ago, I wrote a newsletter called "4 Reasons Your Ads Aren't Working." The short version was that if you think your ads aren't working, you're probably diagnosing the wrong problem. Back then, my answer came down to awareness, messaging, benefits, and price.

That idea came back to me this week on a call with one of my portfolio companies. This business does roughly $90 million a year through DTC and another $30 million through eRetai (Amazon, Walmart, etc.) and physical retail. This is not a small brand hunting for product-market fit. It has real scale and more customer data than most brands will ever collect.

The team had roughly doubled their creative output over the past few weeks. Performance, however, barely moved.

You know the instinct in that moment... Make even more ads. Swap the creators. Question the media buyer. Wait for the algorithm to figure it out. But the better question was sitting right there... did the company test more net-new ideas, or did it just produce more variations of previous ones? The customer, the angle, the offer, and the path to purchase hadn't diversified at anywhere near the rate the creative output had. More ads had just created the appearance of more testing.

That's what today is about. "Our ads aren't working" is not a diagnosis. It's a signal to look at the entire system that creates demand, converts it, measures it, and learns from it. So I'm giving you the 7 questions I'd run with any brand before making another batch of creative or cutting another dollar of spend.

But, before we get into that...

Vendor of the Week

RetailClub — where AI stops being a pilot and starts changing the P&L.

Every retail leadership team has an AI roadmap now. Search and discovery. Creative. Customer service. Supply chain. Internal tools. Half the projects are pilots, the other half are slides, and nobody agrees on which ones should actually scale first.

That’s the problem RetailClub AI Festival is built to force into the open. It’s a 3 day gathering for operators ready to move past surface-level “AI in retail” talk and confront what full-scale transformation actually demands.

The speaker lineup is basically the group chat you’d want for that problem: senior leaders from Target, e.l.f. Beauty, Pacsun, Kendra Scott, Albertsons, Bayer, Steve Madden, Sephora, ServiceNow, and Salesforce.

The agenda gets specific too: agentic commerce, the future of search and discovery, intelligent supply chains, enterprise AI strategy, and future-proofing the tech stack. There are also Level Up sessions, expert-led groups, and live demos built to sharpen your own AI skills, not just hand you another deck.

RetailClub runs September 22-24 in Huntington Beach. Qualifying retailers and brands can get a free ticket plus $1,250 toward travel and hotel through the Hosted Program. The deadline to register is August 28.

Apply for your free RetailClub ticket by August 28.

Question 1: Are you testing more ads, or more ideas?

Back to the portfolio company. The team's read was simple... we doubled output, nothing happened, so creative volume must not be the answer. My response was that Meta doesn't just want more ads. It wants more of the full pipe: avatar, angle, funnel, and offer. Twelve variations of one idea can still register, commercially, as one idea.

Here's the thing. An ad file is not automatically a new test. If twelve ads share the same customer, the same problem, the same promise, the same offer, and the same destination, you've given the platform twelve executions and given your business exactly one commercial hypothesis.

And I'm not arguing against Meta here, by the way. I'm agreeing with them. Meta's own definition of creative diversification is a wide range of creatives with different themes, messages, and visuals for different audience segments. Their whole point with Andromeda is that AI-driven delivery moved the leverage away from niche targeting and toward creative diversity. Different themes and messages... not more volume for its own sake.

Here's how I break it down. Four inputs:

• Avatar: Which specific customer is this for?

• Angle: Which problem, desire, belief, or moment makes that customer care?

• Funnel: Which argument and sequence should continue after the click?

• Offer: What product, bundle, price, commitment, or incentive are we asking them to accept?

A different creator holding the product is not a new avatar. A new hook pointing at the same broad benefit and the same offer isn't a new angle either. And that fresh edit of last month's winner? That's an iteration. Iterations are fine, but label them as iterations... don't count them as tests.

Your move: export your last 20 ads and group them by customer, problem, promise, offer, and destination. If most of them collapse into one or two groups, you did not run 20 tests. You ran two. Count the hypotheses, not the files.

And before you approve the next round of production, make the brief answer one question: which of the four inputs is genuinely new here? If the answer is none, you're about to pay for another iteration and call it a test.

Even a genuinely diverse creative system has a ceiling, though. You can't keep selling to a market that isn't growing.

Question 2: Are enough new potential customers discovering you?

This was the first point in the original newsletter, and it's still the one most brands skip past. People cannot buy something they don't know exists. Advertising cannot capture demand the company never created.

Here's why this one is sneaky... everything looks healthy on the surface. Retargeting is efficient, branded search converts great, and the conversion campaigns hit their CPA every month. Meanwhile the brand is quietly exhausting the same pool of people, frequency is creeping up, and the ad account slowly gets worse at a job it was never designed to do alone... educating a category.

Split your acquisition into two buckets. Demand creation: organic content, creators, partnerships, PR, podcasts, events, sampling, retail presence, broad prospecting. Demand capture: retargeting, branded search, marketplaces, shopping ads, conversion campaigns. Then ask the uncomfortable question... has demand creation grown at the same rate as your revenue target? For most brands the honest answer is no. The revenue goal went up 40% and the demand creation engine stayed exactly the same size.

The evidence is easy to pull. New-customer reach and frequency. Branded search trend. Direct and organic traffic trend. Creator and referral traffic. The percentage of spend going to prospecting versus retargeting/remarketing. And new-customer volume, not just blended acquisition cost... blended numbers love to hide a shrinking top of funnel.

Meta cannot harvest demand the rest of the company never created. You may not have an ad problem. You may have an awareness ceiling. Once enough people are discovering you, the next question is whether the message gives a specific person a reason to care.

Question 3: Is the message specific to a real customer moment?

Remember, features are what the product does, benefits are what changes for the customer. The angle is different. An angle takes that benefit and connects it to a specific person in a specific moment.

I've used this example before… a rice cake brand can say "low calorie, light, easy to prepare." Those are features. "An on-the-go snack" is a benefit. "The after-school snack a mom can actually feel good about" is an angle. Same product. Only one of those makes a specific person feel recognized.

Most scaled ad accounts are stuck at the benefit layer. "Better sleep." "More energy." "Clean ingredients." "Backed by science." All true, and all category-level claims that every competitor can also make. That's the trap... the copy isn't wrong, it's just not owned.

Take your top five ads, strip out the logo and product shots, and read the copy. Could a direct competitor run it word for word? If yes, you're running category language, not customer insight.

Now rewrite one ad around one customer, one moment, one failed alternative, one desired change, and one proof point. The raw material already exists in your reviews, your comments, your support tickets, your cancellation reasons, and the Reddit threads about your category. Your customers have already written your best angles. Most marketers just never go read them, they’re drinking the brand’s own Kool Aid too hard.

And this pays off way beyond the headline. The more concrete the moment, the more obvious every downstream decision becomes... which creator, which visual, which proof, which page, which offer. The customer should recognize the moment before they understand the mechanism.

Of course, a specific message can win attention and still lose the sale if the customer can't quickly figure out what the purchase is worth.

Question 4: Is the value easy to understand?

"Our price is too high" is where a lot of these conversations dead-end. Sometimes the price really is wrong. But I watch operators collapse four different problems into that one complaint:

• Price: The absolute number is outside what the market will accept.

• Value: The customer doesn't understand the outcome, the quantity, or the savings.

• Commitment: You're asking them to buy too much or subscribe too early.

• Risk: They don't trust the product, the guarantee, the delivery, or the cancellation experience.

Those are four different diagnoses with four different fixes, and only one of them is "charge less."

Question 5: Does everything after the click continue the argument?

I went deep on this last week, so I'll keep it short. Every ad creates a mental receipt... the reason the customer clicked. The next screen either recognizes that receipt or asks them to start the sale over.

I've written before that the hero should match the source of traffic. If the ad is conversational, the page stays conversational. If the ad shows a specific flavor or color, the first screen shows the same one. Last week's issue took that further into full buying routes... one customer insight carried from the creative through the argument, proof, offer, and checkout. If you missed it, go read it.

For today, the diagnostic is simple. Take your three highest-spend ads and manually walk each path: ad, first screen, argument, proof, product, offer, checkout. At every step, ask whether it's still the same customer and the same promise, and whether the proof answers the objection that promise creates.

The ad can win the click while the next step loses the customer. And when that's happening, no amount of new creative will show up in your results.

Which brings us to the question almost nobody asks... are the results you're staring at even complete?

Question 6: Are you measuring everywhere the customer can buy?

Customers do not organize their purchases around your channel P&L. They see your ad on Meta, browse your site on their phone, check the reviews on Amazon, and buy at Target on Saturday because they were already there. Shopify records nothing and Meta gets little or no credit. But the business just acquired a customer.

An operator I trust told me recently about a brand that expanded nationally into Target. DTC declined sharply. Amazon grew. And total company revenue was reportedly up around 30% year over year, because purchases shifted to wherever buying was most convenient. On the DTC dashboard, ROAS and CAC looked worse... while the actual business was growing. He also watched a separate brand pull back web-focused media and saw Amazon revenue dip almost immediately after.

The customer may have converted. Your dashboard may have lost the checkout. And once you can finally see the whole system... somebody still has to turn what you're seeing into decisions.

Question 7: Who owns the learning loop?

This is the question that separates brands that get smarter every quarter from brands that just stay busy.

At scale, every function can hit its numbers while the company learns nothing. Creative ships assets, media buys traffic, web ships pages. CRO runs tests. Creators post. The retail team manages the marketplaces. Analytics produces another dashboard. Everyone's output is fine... and nobody owns the answer to four questions: What did we test? What happened? Why do we think it happened? What should change next?

I had this exact conversation with a brand deciding between hiring an agency for a channel or building it internally. My take was direct: outsource execution if it helps, it'll probably even be faster. But you want the learnings in house. If the knowledge of what works for your customer lives entirely inside an external team, you didn't buy growth. You rented it.

The portfolio company from the top of this email is the proof. Multiple creative sources, media ownership, attribution tools, dashboards, a fractional growth lead... all the pieces. And the question of whether doubled output represented real strategic diversity still required someone to connect creative, media, funnel, offer, and measurement into one diagnosis. The tools were there. The owner wasn't.

The fix doesn't need software. Run a weekly learning review with a single page per test: hypothesis, customer, angle, offer, destination, spend and duration, primary result, guardrail results, interpretation, and next decision. Assign one owner to keep the record, and force every new test to cite what the last one taught you.

"Our ads aren't working" is not a useful conclusion. It's the beginning of an investigation. Before you brief another batch of creative or cut another dollar of spend, figure out which part of the system is actually failing. The fastest way to fix an ad account is often to stop looking only at the ads.

Better ads matter. Better questions tell you which ads to make.

That's all for this week

This one was a refresh of something I wrote years ago, and honestly, the fact that a $120 million brand and a $20 million brand can have the exact same complaint tells you how evergreen this problem is. Don't make another ad until you know what the next ad is supposed to teach you.

It's Sunday night, so I hope you plan to get a full 9 hours of sleep, stay hydrated, and get ready for a monster of an upcoming week. I am going to be getting things prepped as we approach the upcoming Q4 Summit on September 24th in New York City. If you’d like to join, reply to this email or apply here.

I'll see you next Sunday. Same time, same place. Have an amazing upcoming week!

Nik