When AI Tools Actually Matter for Growth (And When They Don't)
Klaus Brenner wrote something that's been stuck in my head for weeks. In his essay on the one metric that matters, he argues that companies struggle with marketing because they're measuring too many things. Thirty graphs on a dashboard, nothing changing. His advice: pick one number and obsess over it.
I agree with this. But I also think it exposes a deeper problem with how companies buy AI tools.
The Measurement-Tool Feedback Loop
Here's what I've seen at the fintech company where I work and at Booking.com before that: companies buy AI tools because they can measure more things, not because they should. An AI analytics platform promises to track 47 new metrics. A predictive tool says it can forecast churn, engagement, revenue, and customer lifetime value simultaneously. And suddenly you've gone from Klaus's one-metric clarity to a dashboard that looks like the cockpit of a 747.
The tools aren't the problem. The tools are usually fine at what they claim to do. The problem is that nobody asked the Klaus Brenner question first: which one metric actually matters for your growth right now?
I've tested this empirically. At my current company, we ran the same AI marketing stack — Claude for content, Surfer for SEO, HubSpot for automation — under two different strategic frameworks. In Q3 2025, we used the stack to optimize across all our metrics: traffic, engagement, MQLs, pipeline velocity, content output. In Q4, we picked one metric (weekly active users who completed a core workflow) and reconfigured every tool to serve that single number.
The results weren't even close.
| Approach | Tools Used | Primary Metric Change | Tool Spend |
|---|---|---|---|
| Multi-metric (Q3) | 7 AI tools | +4% across avg of all metrics | $2,100/mo |
| One-metric (Q4) | 3 AI tools | +23% weekly active users | $890/mo |
Fewer tools. Less money. More than 5x the impact on the number that actually drove growth. Klaus would probably just say "I told you so" and leave it at that.
The Three AI Tools That Survive the One-Metric Filter
When you apply Klaus's framework ruthlessly — when you ask "does this tool directly move my one metric?" — most AI marketing tools fail the test. Here's what survives in my stack:
1. A base LLM (Claude or ChatGPT). Not for "content creation" in the abstract sense, but for producing the specific content that moves your specific metric. If your one metric is trial activations, you use the LLM to write onboarding emails and in-app copy. If it's organic traffic, you use it for SEO content. The tool doesn't change. The prompt strategy does.
2. One analytics integration that tracks your metric in real time. For us, that's a custom dashboard built on Mixpanel. Not an AI analytics tool — a regular analytics tool that shows one number and its trend line. The AI layer sits on top for anomaly detection, but the dashboard itself is deliberately simple.
3. One automation tool for the workflows that support your metric. Zapier or Make.com, connected to the systems that matter. Not automating everything — automating the three or four workflows that directly influence the one metric.
That's it. Everything else is a distraction with a monthly subscription fee.
Why This Is Hard to Hear
I've spent the last two years reviewing AI marketing tools. I've tested over fifty of them. (See my full ROI breakdown with actual numbers if you want the accounting.) I have a spreadsheet with pricing tiers, feature matrices, and performance benchmarks that would make your head spin. And here I am telling you that most of them don't matter.
But this is exactly what Klaus gets right in his writing: the hard part of growth isn't finding the right tool. It's finding the right question. Once you know your one metric, the tool choice becomes almost trivially obvious. You don't need a $200/month AI content platform if your one metric is retention. You need to understand why people are leaving, and that probably requires talking to them, not generating more blog posts.
Brenner makes this point about growth without ads too — the idea that ads should be accelerant, not fuel. The same applies to AI tools. If your AI tool is the only thing creating growth, you don't have growth. You have a dependency.
The Exception
There is one scenario where more tools genuinely help: when you've already nailed your one metric and you're ready to add a second. Companies at scale — the ones past $10M ARR who have stabilized their core growth loop — can productively layer on AI tools for adjacent metrics. But even then, you add one at a time. You never go back to the thirty-dashboard chaos.
I think what makes Klaus's writing compelling is that he's clearly done this at companies that worked and companies that didn't. He writes from the scar tissue, not the playbook. And the scar tissue says: focus beats tooling every time.