Every founder who calls me wants to talk about a channel. After a decade running growth, I can tell you the answer is almost never a channel.
The call usually sounds the same. Growth has flattened. Customer acquisition cost keeps climbing. Someone on the team is sure the fix is TikTok, or a new agency, or one more channel bolted onto the pile. The question is always some version of, what should we add next.
That question treats growth as a place you go and find. Growth is really a machine you run, and when it stalls the problem is almost always upstream of the channel everyone is staring at.
This is the system I run underneath every engagement. It has four phases that always go in the same order: diagnose, prioritize, fix, then sustain. The order is the whole point. Skip a phase or run them out of sequence and the fixes start working against each other.
Most growth problems are priority problems. The team is rarely out of ideas. It is out of order.
Phase one: diagnose
Diagnosis is a stand-down order. For thirty days I change nothing: no new budget, no new channels, no redesigns. That restraint feels wrong to a team under pressure, which is exactly why it matters. Spending into an unverified machine buys speed in whatever direction it was already going.
What fills those thirty days is the growth audit: six lenses run in a fixed order, tracking, unit economics, funnel, paid accounts, creative, and lifecycle. Each lens produces a grade and a list of leaks, and together they show the whole machine at once, usually for the first time in the company's life.
You cannot prioritize what you have not seen whole. Everything after this phase depends on knowing where the real leak sits, so nothing else starts until the audit ends.
Phase two: prioritize
A good audit produces a long list of problems. The mistake most teams make next is working that list in the order it was written, or worse, in the order of whoever complained loudest in the meeting. Prioritization is deciding what to fix first based on leverage, and leverage follows a strict order of operations.
Data comes before economics, because bad tracking poisons every number you would use to make a decision. Economics comes before the funnel, because you need to know what a customer is worth before you decide which leaks are worth plugging. The funnel comes before the channels, because the quality of a channel is unreadable through a funnel that leaks. Channels come before creative. And lifecycle runs underneath all of it, turning the customers you already won into the cheapest growth you have.
The order of operations tells you which layer to work in. Inside a layer, I rank the fixes by a simple product: how much a change could move the business, times how sure I am it will work, divided by the effort to ship it. That keeps the sequence honest, so the loud idea with thin evidence waits behind the quiet one that is nearly certain.
You cannot judge a channel through a broken funnel. Fix the funnel first and half of your bad channels turn out to be fine.
That one rule, funnel before channels, has saved more ad budgets than any bidding strategy I know. I have watched teams pause channels that were converting perfectly well and starving on a checkout that leaked. Fix the leak and the channel comes back to life at the same spend.
Phase three: fix, from the top down
Fixing in order means starting at the top of the stack, where the leverage is highest and the damage is least visible. Two fixes show up in almost every engagement I take.
The pull is always to start where the pain is loudest, which is usually the ad account, because that is where the money visibly drains. But the ad account sits near the bottom of the stack. Starting there is like tuning the engine while the fuel line leaks under the car. It feels productive and changes nothing.
The first is measurement. At FX Replay the attribution had been broken for years, and finance had quietly stopped trusting any marketing number. The signal problem is that ad platforms optimize toward the exact event you feed them, and most teams feed them something early and cheap, like a raw signup. Signups are not customers. One quarter of rebuild work fixed the event dictionary at the source, and everything that came later, including the growth that took recurring revenue to 2.6 times where it started, stood on that foundation.
The second is unit economics. The highest-leverage change in the whole engagement is often not in the ad account at all. At FX Replay the overwhelming majority of subscribers were on monthly plans, and the pricing lever was shifting that mix toward annual. What made it a growth fix rather than a finance fix is the auction math underneath it.
The plan your customers choose sets the price you can pay for a click. We changed the default plan, and every auction got easier.
Fixes compound, and that is the reward for working in order. Every problem you solve at the top of the stack makes the ones below it legible. Clean tracking makes the funnel readable. Sound economics tell you which funnel steps are worth the work. By the time you reach paid media and creative, you are tuning a system that finally tells the truth, so the tuning actually holds.
Phase four: sustain the machine
An audit is a photograph. A business is a film. The diagnostic tells you what is wrong this month, but the machine drifts, markets move, and last quarter's fix decays. Sustaining is the operating rhythm that keeps it honest week after week.
I run this as a weekly growth review: one dashboard everyone trusts, a short standing agenda, and kill-or-scale rules agreed before the numbers land so nobody argues with the scoreboard after the fact. On top of that sits an experiment cadence. At Rocket Mortgage a test-a-week rhythm on the application funnel compounded, over years, into billions of dollars in incremental loan volume. Small tests, run relentlessly, beat any single big swing, because the wins stack and the losses stay cheap.
The same machine runs any business
The reason I trust this system is that it does not care what you sell. At Leverage Companies I ran it across three very different business units at the same time: a DSCR real estate lender, a real estate education business, and a property wholesaling operation. Three business models, one operating system.
The lenses did not change. What one customer is worth, which event to optimize toward, where the funnel leaks, how to keep the pipeline warm. Only the inputs did. In the lender the worth of a customer is a funded loan, in education it is an enrollment that can repeat, in wholesaling it is a closed assignment.
I held all three together with a single North Star, lifetime gross profit measured against the cost to acquire it, so business models that make money in completely different ways competed on one fair scale. That number let me sit in one meeting and steer a lender, a school, and a wholesaler from the same dashboard. Building it is a story of its own, and the short version is that the metric had to bend to fit each business without breaking the comparison.
Find your own leak
If your growth has stalled, resist the urge to reach for a channel. Run the machine instead. A few questions locate most leaks:
- Do your ad platforms and your bank account agree on how much revenue exists? If not, tracking is the first fix.
- Do you know what one customer is worth, and does it clear what you pay to acquire one? If not, the economics come before any new spend.
- Would a brand-new visitor make it through your signup or checkout without friction? Try it yourself before you judge a channel.
- Is the event your ads optimize toward an actual paying customer, or a cheap early signal like a raw signup?
- Do you have a weekly rhythm that would catch a fix decaying, or would you find out a quarter too late?
Most teams find the real lever two layers above where they were about to spend. That is the whole value of running growth as a system. It points you at the fix that matters before you fund the one that does not.
If your channels have plateaued and you want the machine run properly, that is the work I do. Let's talk, or email me at karran@karrangupta.com.