The Metrics Hierarchy That Real Growth Leaders Use — Not the One in Textbooks

The textbooks will tell you to obsess over conversion rates. Optimize each funnel step. Get users from signup to activation to paid as efficiently as possible. The math is elegant: improve each conversion gate by 10%, and the compound effect multiplies revenue.

But talk to growth leaders who've actually scaled products to hundreds of millions of users, and you'll hear something different. They'll tell you that optimizing intermediate conversion rates is often the fastest way to destroy long-term growth. That the metrics hierarchy you learned in business school gets the incentives precisely backwards. That sometimes the best way to grow is to let more people churn, not fewer.

The real metrics hierarchy isn't about conversion rates at all. It's about understanding which leading indicators actually predict the outcomes that matter — and having the discipline to ignore everything else.

The Conversion Rate Trap That Kills Cross-Functional Teams

When Archie Abrams joined Shopify as VP of Product and Head of Growth, he inherited an organization facing a seductive problem. Teams naturally organized around different funnel stages were doing exactly what you'd expect: optimizing their slice of the conversion funnel.

When you have teams naturally break up the world into different funnel stages or different points in the journey, it gets very seductive to look at my part of the funnel and what's my conversion rate through that part of the funnel. And then the team starts to optimize for that conversion rate as their North Star. But in practice, it's actually almost always easier to just make it harder to do the thing right before your step in the funnel to increase your conversion rate.

— Archie Abrams, VP Product & Head of Growth at Shopify

The perverse incentive is obvious once you see it. The activation team can goose their conversion rate by making signup harder. Fewer low-intent users enter the funnel, and suddenly activation rates look fantastic. The team hits their targets. Leadership celebrates. The business shrinks.

Shopify's solution was radical: they essentially banned funnel-stage KPIs. Core product teams don't have metrics at all. Instead, decisions flow from a 100-year vision that comes directly from CEO Tobi Lütke, and teams operate on taste, intuition, and building toward that long-term future.

The growth team does measure things — but with a crucial philosophical shift. Rather than optimizing retention, Shopify optimizes for churn. Or more precisely, they optimize for getting as many people to try entrepreneurship as possible, knowing that most will fail.

We want to lower the barriers to getting started and help folks grow, and those winners make the whole thing work. That's why we lower the barriers to get started and help folks grow, and the folks who do go on to be successful will kind of make that entire cohort of merchants extremely successful.

— Archie Abrams, VP Product & Head of Growth at Shopify

The business model makes this possible. Shopify charges a subscription, but most revenue comes from payments tied directly to merchant success. A cohort where 70% churn but 5% become massive businesses generates far more value than a cohort where 90% stick around as tiny shops generating minimal payment volume.

This isn't recklessness. It's a clear-eyed recognition that the metric that matters isn't conversion rate — it's total successful outcomes. And sometimes the path to more successful outcomes requires accepting lower conversion rates.

The Data Foundation Nobody Talks About (That Makes PLG Actually Work)

Product-led growth has become the default playbook for B2B SaaS. Offer a free tier. Let users experience value before buying. Use product usage to identify expansion opportunities. The mechanics seem straightforward.

But Hila Qu, who led growth at GitLab and is now an Entrepreneur-in-Residence at Reforge, sees most companies miss the fundamental prerequisite.

PLG, I always say, is actually fundamentally DLG, data-led growth. So when you give away your free product, what you want to get in exchange are two things. One is the broader reach because free product spread itself is lower barrier to entry. Two, you want to understand the usage behavior of those free users, which features do they use and which features kind of correlates with a higher conversion rate, retention rate, all of that. If you don't have a foundation of data and an understanding of how to analyze those data, you are giving away a free product for nothing.

— Hila Qu, EiR at Reforge (fmr Dir Growth at GitLab)

This reframes the entire activation-to-expansion funnel. The point of activation isn't just to get users to a value moment. It's to generate the behavioral data that tells you which users will stick, which will expand, and which features predict each outcome.

Without that data infrastructure, you're flying blind. You might see that users who complete three specific actions convert at higher rates, but you don't know if those actions cause conversion or simply correlate with user intent. You can't distinguish between users who need a sales touch and users who'll self-serve. You can't identify the expansion signals that predict when a small team is about to become an enterprise account.

Julian Nunez lived this firsthand while building payments infrastructure at Rappi, then founding Yuno to solve the problem at scale. The company processed tens of billions of dollars, but for weeks, transaction volume stayed flat.

Recuerdo que una noche regresando hacia Bogotá, yo iba como copiloto y yo actualizaba el dashboard de Juno. Era un domingo y era la cuarta semana seguida que cerrábamos flat en el número de transacciones y yo decía No, no puede ser. Hoy en día procesamos mil veces más que en ese momento, pero ahí la angustia era brutal.

— Julian Nunez, Founder at Yuno

The breakthrough came from understanding payment behavior data deeply enough to route transactions dynamically across providers based on predicted approval rates. But that required infrastructure to capture, analyze, and act on usage data in real-time — the data foundation that Qu identifies as the actual prerequisite for product-led growth.

Why The Best Growth People Have Zero Marketing Experience

Matt Lerner spent 11 years running growth at PayPal. In that time, he hired and managed dozens of growth team members. When asked who performed best, his answer surprises people.

I'll say categorically the best growth people I hired and managed in my time at PayPal had no marketing experience, had no product experience at all.

— Matt Lerner, Growth Leader at PayPal / 500 Startups

The reason cuts to the heart of what growth actually requires. Lerner breaks down acquisition channels into six fundamental types: sales, partnerships, paid ads, organic search, content/inbound, and influencers. That's it. The tactical execution isn't rocket science. Any competent marketer knows how to run ads or optimize SEO.

The hard part is figuring out which specific tactics will work for your business. Google has 100,000 search results for any query — how do you break into the top three when competitors with 10-year head starts and bigger budgets have those positions locked? Paid ads are sold at auction, meaning you need better unit economics and smoother funnels than everyone else bidding on the same keywords.

This requires information discovery, not playbook execution. And marketing experience can actually hurt here, because experienced marketers arrive with strong opinions about what works. They've seen it before. They know the playbook. They execute confidently — and waste months building campaigns that don't survive first contact with customers.

Sandy Diao, who scaled products to over 200 million users at Pinterest, Meta, and Descript, puts it bluntly when describing her formative experience at Pinterest:

One of the first things that I did in that first week was basically answer 500 support tickets. I was bouncing off the walls and doing every little bit of work that was required to help grow the company. After that, I started to look at the data around the customer support tickets. What were the most common questions? What were the inquiries? Who was writing in? It was just ridden with golden nuggets there.

— Sandy Diao, Fmr Director of Growth at Descript

Support tickets revealed that businesses wanted to advertise on Pinterest. That users needed analytics proving Pinterest worked for them. That onboarding flows needed fundamental rethinking. None of this came from marketing playbooks. It came from direct contact with user pain, translated into growth hypotheses, tested across every channel.

The best growth people are curious generalists who default to learning what customers actually need rather than deploying what worked at their last company. They treat their own expertise with healthy skepticism.

The 90/10 Rule That Changes How You Allocate Resources

Ask most growth teams how they allocate resources, and you'll hear about portfolio approaches. Diversified bets across channels. Systematic testing. Incrementalism.

Lerner watched this play out at PayPal for over a decade, and when he stepped back, he saw a pattern that changed how he thinks about growth entirely:

When I look back on my time at PayPal, 90% of our growth came from like 10% of the stuff that we did. I could go back and say, there are 5 or 6 things obviously before my time getting on eBay, referral bonuses, network effects growth, getting engaged with web developers who were building the sites and implementing the payment systems, getting pre-integrated with shopping carts and hosts. Some outbound sales that drove almost all their growth. But they did a ton of other stuff. They spent hundreds of millions of dollars on marketing campaigns, building products that frankly nobody ever used.

— Matt Lerner, Growth Leader at PayPal / 500 Startups

PayPal could afford the waste because they generated hundreds of millions in free cash flow. But startups don't have that luxury. The core challenge becomes identifying which 10% will drive 90% of results before burning resources on the other 90%.

Sri Batchu, who leads growth at Ramp and previously scaled Instacart's ads business from scratch, frames this as portfolio management with venture-style thinking:

Growth has just a typically high failure rate. Two-thirds of your projects are likely to fail. What's important is that you fail quickly and conclusively. Startups should have time periods of planning that are much, much shorter on a two-week sprint cadence. So you really reduce the cycle time to output.

— Sri Batchu, Head of Growth at Ramp

The metrics hierarchy here isn't about tracking everything. It's about having clear hypotheses, testing fast, killing losers quickly, and doubling down on the rare winners that actually move the business. Two-week sprints force discipline. If a growth experiment can't show signal in two weeks, it probably won't show meaningful results in two months.

This changes what you measure. Vanity metrics that move slightly each week become irrelevant. The question becomes: what's the single metric that, if it moved significantly, would validate this bet? And can we get signal on that metric in two weeks?

Where Growth Leaders Actually Disagree: The Sales-PLG Debate

One area where even experienced growth leaders diverge is on the relationship between product-led growth and sales-led motions. The conventional wisdom has shifted over time — first sales-led was standard, then PLG was the future, now "everyone needs both" is becoming the new orthodoxy.

But the sequencing and emphasis vary wildly.

Hila Qu has become more convinced that hybrid approaches work best, even from early stages:

I wouldn't say it's shifted completely because I always believe you don't need to be a PLG purist. Recently by working with a few of my clients, I witnessed in reality, like many startups actually are having both. They have a PLG motion. They have a sales team as well. PLG motion is perfect for lowering the barrier for more people to try, broader the reach. It's a kind of a volume game. And then the sales motion, you can have very targeted list of big customers you go after. You close them and it's a big order.

— Hila Qu, EiR at Reforge (fmr Dir Growth at GitLab)

But she's also clear that adding PLG later is harder than adding sales later. If you start pure sales-led, retrofitting a product-led motion requires fundamental product and go-to-market changes. Starting with PLG and layering in enterprise sales is a more natural progression.

Shopify represents the other end of the spectrum — a company so committed to lowering barriers that their entire business model depends on accepting churn that traditional SaaS companies would consider catastrophic. Their "sales motion" is essentially helping merchants succeed so payment volume grows, not closing big upfront contracts.

The disagreement isn't about whether both motions can coexist. It's about which metrics you're willing to sacrifice to enable each motion. PLG-first companies accept that enterprise sales will be slower and harder. Sales-first companies accept that true product-led growth may never fully materialize. Hybrid approaches accept higher operational complexity and potentially conflicting incentives between self-serve and sales-assisted channels.

There's no universal answer. The right choice depends on product, market, and what trade-offs you're willing to make. But pretending these trade-offs don't exist — or that you can optimize for everything simultaneously — is how growth orgs end up with incoherent metrics and misaligned teams.

The Simplification Superpower That Cuts Through Complexity

At both Instacart and Opendoor, Sri Batchu encountered a level of marketplace complexity that makes most growth challenges look simple. Instacart juggles four distinct sides: shoppers, buyers, advertisers, and retailers, each with different incentives and revenue streams. Opendoor had to price homes accurately enough to stay profitable while Zillow was willing to lose money for market share.

The temptation in complex environments is to pursue multiple high-impact initiatives simultaneously. More levers, more growth. But Batchu learned something different from Instacart CEO Apoorva Mehta:

Simplification is a superpower and it's actually very difficult. A lot of executives, when they're faced with two potentially high-impact initiatives, are often tempted to say, let's do both. The good ones can pick one and make a clear decision. I think the best ones will just say, actually, we're doing neither because they don't fit in our overall strategy and are distracting.

— Sri Batchu, Head of Growth at Ramp

This is where the metrics hierarchy becomes most powerful — and most painful. If you're measuring everything, you can justify any project. Revenue, engagement, retention, virality, conversion rates, NPS, marketplace liquidity — there's always a metric that a given initiative will improve.

The discipline is choosing which metric actually matters enough to ignore everything else. At Opendoor, that metric was accurate pricing. When Zillow entered the market offering higher prices for homes, Opendoor had a choice: match prices to maintain volume, or stick to their pricing model and accept lower short-term growth.

They chose pricing accuracy. Volume dropped. Growth slowed. Teams were anxious. But the decision was made on first principles: their competitive moat was pricing accuracy, and short-term growth wasn't worth compromising it. Zillow eventually exited the market before the real estate crash, validating the decision.

This kind of clarity only works when the entire organization agrees on what success looks like. Not agreement on tactics, or even strategy, but on the one thing that the business can't afford to get wrong. Everything else is negotiable.

What To Measure When You Stop Measuring Everything

So if intermediate conversion rates are a trap, and most initiatives fail, and simplification requires saying no to high-impact work — what should growth teams actually measure?

The answer isn't a metric. It's a hierarchy of questions:

First: What outcome would make your business 10x larger in three years? Not what channels or tactics, but what actual customer behavior or market position. For Shopify, it's more people trying entrepreneurship. For Yuno, it's processing payments that would otherwise decline. For Ramp, it's finance teams running entire operations through their platform.

Second: What leading indicator predicts that outcome with the shortest feedback loop? This is where data infrastructure becomes essential. You need to know which early behaviors actually correlate with long-term success, not which ones you hope correlate. Shopify runs long-term holdouts for every experiment and automatically checks impact one, two, and three years later to validate whether short-term wins actually mattered.

Third: What's the single biggest constraint preventing more of that leading indicator from happening? Not the five biggest constraints — the one biggest. Is it awareness? Is it product value? Is it onboarding friction? Is it pricing? You can't fix everything, so fix the thing that's most broken.

Finally: Can you test a solution to that constraint in two weeks? If not, can you break it into smaller tests? If the answer is still no, the constraint might be real, but you probably can't solve it with a growth team. It might require fundamental product changes, business model shifts, or market timing.

This hierarchy forces discipline. You can't have five North Star metrics. You can't optimize the whole funnel simultaneously. You can't pursue every high-impact opportunity. You have to choose.

And when you choose well — when you identify the actual leading indicator, remove the actual constraint, and validate that it actually drives the outcome that matters — that's when you find the 10% of work that drives 90% of growth. The metrics hierarchy that real growth leaders use isn't about measuring more. It's about having the conviction to measure less, but measure what actually matters.