The exact six-lens teardown I run in the first thirty days of every engagement, and the order that makes it work.
Most founders who call me have two sets of numbers that disagree. The ad platforms say marketing is working. The bank account says it isn't.
Somewhere between those two reports is the real business, and nobody can see it.
So every engagement I take starts the same way: thirty days, no new spend, no new channels. Just an audit. The logic is simple. If the system is broken, more money makes it break faster.
This is the exact teardown I run. What I pull, what I look for, and the warning signs that show up again and again.
None of it is secret. If you want to run it yourself, this is the map. If you'd rather hire someone to run it, you should know exactly what you're paying for.
The six lenses run in order on purpose. Each one depends on the one before it.
Lens 1: Tracking and attribution
Attribution means figuring out which marketing deserves credit for each sale. This lens checks whether your version of it can be trusted.
What I pull: every conversion event your ad platforms optimize toward, where each event fires (the visitor's browser or your server), and a comparison of what the platforms report against a source of truth, usually billing data.
What I'm looking for: one thing above all. Add up the revenue every platform claims credit for. Is it more than the revenue that actually exists? It almost always is.
Then I check which event the money is optimizing toward. Ad platforms deliver exactly the event you ask for. Ask for signups and you get signups. Signups aren't customers.
Red flags:
- The same event means different things in different tools.
- Conversions fire only from the browser, which blocks more tracking every year.
- Finance keeps its own private spreadsheet because it stopped trusting marketing's numbers.
That last one is the flag I take most seriously. At FX Replay, attribution had been broken for two to three years, and finance had stopped believing any marketing number.
The rebuild took about one quarter: four clearly defined events, sent from the server to every platform at once. Everything else in that engagement, including the growth that took annual recurring revenue to 2.6x, was built on that fix.
Tracking comes first because bad data poisons every other lens.
Lens 2: Unit economics
Unit economics is a fancy phrase for a plain question: what is one customer worth, and what does one customer cost?
What I pull: how long customers stick around, customer value split by plan and billing period, gross margin, and payback period by channel. Payback period means how long until a customer earns back what you paid to get them.
What I'm looking for: whether anyone in the company can answer the question, and whether they all give the same answer.
Averages are where this goes wrong. One blended number can look healthy while most of your customers sit on the shortest, churniest plan you sell.
Red flags:
- Customer value was calculated once, years ago, from averages.
- Nobody separates monthly subscribers from annual ones.
- Payback period is a guess.
At FX Replay, this lens produced the most valuable finding of the whole engagement. Over 90% of subscribers paid month to month, which quietly capped what any customer could ever be worth.
The fix: make annual plans the default offer and point paid campaigns at the Subscribe event instead of Signup. Trial-to-paid conversion went from 16% to 21%. The ratio of customer value to customer cost landed between 4:1 and 15:1, depending on the channel.
Economics comes second because it defines what "good" means for everything after it.
Lens 3: Funnel and conversion
Your funnel is the path from first visit to paid customer. Every traffic source pours into it.
What I pull: conversion rates at every step, session recordings, form analytics, page speed, and the mobile experience specifically.
What I'm looking for: the step where people quit, and whether anyone can explain why.
This lens sits before the channel audit on purpose. A channel's results depend on two things: the quality of its traffic and the quality of the funnel that traffic lands in. Until the funnel is graded, you can't tell those apart. A "bad channel" is often good traffic hitting a bad step.
And since every channel shares the funnel, one fix here lifts all of them at once.
Red flags:
- Forms that open with the most sensitive question.
- Paid traffic landing on generic pages.
- One funnel for every visitor, on every device.
YouMail shows what happens when channels get judged first. Paid acquisition looked like it had stopped working, and the channel was taking the blame.
The real defect was the funnel. Paid traffic went straight to the app stores, where 76% of it bounced. A web-first funnel cut bounce to 25%, raised conversion from 9% to 22%, and cut the cost of winning a customer by 77%.
Before the fix, paid looked dead. After it, paid grew from nothing to 15% of the business.
Rocket Mortgage shows the size of the prize. Reordering one form so the easy questions came first lifted completion 4% and closing rates 12%. That was worth $1.5B in loan volume with zero extra ad spend.
Same traffic. Different funnel.
Lens 4: Paid accounts
What I pull: full account access, read-only. Campaign structure, where spend concentrates, which events campaigns optimize toward, audience overlap, search terms, and what each campaign actually returned, measured against the numbers from the first three lenses.
What I'm looking for: channel quality, which is finally measurable at this point. With trustworthy events, real economics, and a working funnel, each channel can be judged on how its traffic moves through the same machine.
Most accounts I open have never been checked against real revenue. The platform grades its own homework, the agency forwards the grade, and everyone moves on.
Red flags:
- Campaigns optimizing toward signups or other early events.
- Budgets that haven't moved in months, no matter the results.
- Agency reports built only on platform-reported numbers.
One note on order: basic account hygiene doesn't wait for the funnel work. A wrong optimization event or a runaway budget wastes money no matter what page the click lands on.
One agency-run campaign I audited spent over $15,000 and produced fewer than five subscribers. It ran for months. The tracking above it was broken, nobody could see the return on anything, so nobody caught it.
Lens 5: Creative
Creative means the ads themselves: the images, videos, and words people actually see.
What I pull: every ad from the past six months, how winners get decided, and how often new concepts ship.
What I'm looking for: whether creative runs as a system. Platforms have automated most of bidding and targeting. The ads themselves are the biggest lever you still fully control, and most accounts treat them as an afterthought.
Red flags:
- Three ads running for six months.
- Winners chosen by feel, with no scoring.
- No record of past tests, so old losers get rebuilt every couple of quarters.
The working version looks like FX Replay's: over 1,000 ad variants, each scored against the same revenue-connected events from lens one. Winners got more budget and losers got shut off without a debate.
Lens 6: Lifecycle and owned channels
Lifecycle marketing is everything you send to people you already have: email, SMS, and programs like affiliates and referrals.
What I pull: email volume, automated flows versus one-off sends, what a customer receives in their first 30 days, and any affiliate or referral program.
What I'm looking for: whether the company markets to its own list. Winning new customers gets all the attention because it's expensive and visible. The cheapest revenue in almost every audit turns out to be in the existing customer base.
Red flags:
- Email only sends receipts and password resets.
- Nothing happens when a trial is about to expire.
- An affiliate program that is really a forgotten signup page.
FX Replay's email program was a help desk when I arrived. Eighteen months later it sent 5.9 million emails a month at a 26% open rate. The affiliate program passed 4,000 active partners, returning about $5 in revenue for every $1 paid in commission.
Neither required a dollar of ad spend.
What comes out the other side
The deliverable is a 90-day roadmap, ranked and sequenced. Every finding gets scored by expected impact against effort. Fixes that unblock other fixes go first. Each item has an owner.
The order follows the audit. Measurement gets fixed before spend scales. Economics gets settled before channels get judged. The funnel gets repaired before traffic gets graded. Then winners get more budget.
That order is also why results arrive in a predictable sequence. Tracking and conversion fixes show up first, usually within the first quarter. The compounding gains come after that, once experiments run on numbers everyone trusts.
The roadmap is written so your team can run it without me. If the right next step is a retainer, I'll say so. If it isn't, you have the plan either way.
The 20-minute version
Six questions, one per lens. Honest answers will tell you most of what a full audit would find.
- 1. Add up the revenue each ad platform claims. Is it more than the revenue that actually exists?
- 2. Can anyone on your team say what a customer is worth, by plan and billing period, without building a spreadsheet first?
- 3. Where do most visitors quit your funnel? Can you name the step without looking?
- 4. Which event are your campaigns optimizing toward? Is it the one that correlates with money?
- 5. How old is your best-performing ad? Do you know which one that is?
- 6. What does a new customer get from you by email in their first 30 days, besides receipts?
Two or more uncomfortable answers means the audit will pay for itself. That has been true in almost every account I've opened.
If your dashboard and your bank account disagree, that's the tell. The audit finds out why. Let's talk, or email me at karran@karrangupta.com.