The Customer Acquisition Audit
Find out what it actually costs to win a customer, whether the customers each channel wins were worth having, and where the next dollar should go. This is the complete system in one place: the checklist, the worksheets, the finding format, and the exact prompts, so you can run it on your own acquisition spend this week. One example runs through the whole page, a home-services company with four acquisition channels, so every blank has a filled-in version next to it.
You need a spend export covering your acquisition channels, a CRM or billing export you can follow forward at least 90 days per customer, and the willingness to let a channel’s reputation lose an argument with its own numbers.
What you are looking for
“Customer acquisition cost” means several different numbers depending on who’s asked. This audit forces one answer for each, and then goes further than cost alone.
The output is a small set of channel-level decisions backed by your own cohort data. It is not a single CAC number, and it is never a claim that your LTV:CAC needs to clear 3x to be healthy — that figure is a venture-operator heuristic, not a settled standard, and the defensible comparison is always your own channels against each other.
Pick the scope
Name every channel you’re actually reviewing and the decision this is for, before pulling a single export.
Evidence checklist
Gather it before you interpret any of it. Building the fully loaded number while you’re already forming a view on which channel is best is how the exercise finds what you expected it to.
Cost side
- Media spend by channel and month, from each platform’s own reporting
- Sales and marketing payroll, including variable comp, for anyone whose time goes toward acquisition
- Tools, agencies, content production, and events tied to acquisition
- The P&L or accounting export those costs should reconcile to
Customer side
- New customers by channel and month, from the CRM or billing system
- Revenue and gross margin per customer, followed forward at least 90 days
- Activation, refund, and early-cancellation flags
- Reactivations, marked separately from new logos
Getting your evidence into an assistant
The step nobody writes down, and the one that stops most people before they start.
Getting the files out
- Most systems have an Export or Download button. Your ad platforms, CRM, and accounting or billing system will each have an export or reporting tab.
- Email: copy the thread, or forward it to yourself.
- Can’t export it? Screenshot it — these assistants read images.
- Nothing exports? Write down dates, counts, amounts, who did what.
Which assistant, how many chats
- ChatGPT, Claude, Gemini, or Copilot — pick one that takes file uploads.
- One conversation per process. Upload the evidence first, then the prompts, in order.
- New process, month, or account? Start a fresh chat.
- Upload failing or replies drifting? Send less at a time.
Load the real cost
Full method in System 01. Build both the blended, fully loaded number and the per-channel breakdown — never present one as a substitute for the other.
Judge the customers you won
Full patterns and evidence ratings in System 02. Follow each channel’s customers forward before ranking the channels at all.
Check per channel
- Gross margin at 30, 90, 180, and 365 days
- Activation and early-cancellation rate
- Whether win-backs are mixed into the new-logo number
- Whether anyone else bids on your brand terms, if branded search is in scope
Don’t do this
- Rank channels by acquisition-day CAC alone
- Cite a vendor’s precise “X% of leads are junk” figure as fact
- Judge a cohort younger than its natural evaluation window
Decide with your own cohorts
Full method in System 03. Build cohort-based payback, then give every channel one of five decisions.
Add it up without lying to yourself
What the finished brief contains
If it’s missing any of these, it isn’t finished.
In it
- The channels, the scope, and the decision this was run to inform
- Fully loaded CAC, blended and by channel
- Cohort margin and retention per channel, out to at least 90 days
- Realized, cohort-based payback per channel
- A decision for every channel in scope, with what would reverse it
- The cohorts too young to judge yet, named as such
Not in it
- A single CAC number standing in for the channel breakdown
- Revenue used in place of gross margin
- A borrowed LTV:CAC or payback threshold presented as a pass/fail bar
- A vendor’s precise lead-quality or SaaS-waste percentage cited as fact
- An incrementality claim for branded search made without running the test
The master prompt
Paste this into any AI assistant along with your spend and CRM exports. It runs the whole method in order and will not hand you a single CAC number or an unearned pass/fail verdict.
I want you to help me work out what it actually costs to acquire a customer through each of my channels, whether those customers were worth acquiring, and where my next dollar of acquisition budget should go. Your job is not to hand me one CAC number or tell me whether I've cleared some external benchmark. Your job is to build the real, fully loaded numbers and compare my channels to each other. I will give you spend data by channel, and a CRM or billing export I can follow forward for at least 90 days per customer. Work through these stages, in order, and do not skip ahead: 1. SET THE FRAME Read back to me, in one sentence each: the channels in scope and the decision I said this is for. If I haven't told you one of these, ask before continuing. 2. LOAD THE REAL COST For each channel, build a fully loaded CAC: direct spend, plus allocated labor and shared costs using a rule you state explicitly. Also build the company-wide blended, fully loaded CAC. Never let one substitute for the other in anything you write. 3. FOLLOW THE CUSTOMERS FORWARD For each channel, calculate gross margin per acquired customer at 30, 90, 180, and 365 days where the data supports it. Flag any cohort too young to have reached these windows rather than estimating its value. 4. FLAG THE MISCALCULATIONS Check for: free trials counted as paying customers, quick churns or refunds counted as normal wins, organic or referral customers blended into a paid channel's numbers, and reactivated customers counted as new logos. Report each separately from the clean numbers. 5. BUILD REALIZED PAYBACK For each channel, calculate payback from actual cohort revenue and margin month by month, not a static ARPA-over-churn formula. Report it in months. 6. COMPARE, DON'T GRADE Rank channels against each other by realized 90- and 180-day margin, not by acquisition cost alone. Do not compare any channel's LTV:CAC or payback against an external published threshold as if it were a pass/fail test — say explicitly that those figures are heuristics, not proven standards. 7. DECIDE, PER CHANNEL For each channel, recommend one of: grow it, hold it, fix the qualification, test it before trusting it, or cut it. For anything where the underlying conversions might not be incremental — branded search chief among them — recommend a test rather than a verdict, and say what evidence would settle it. 8. WRITE THE BRIEF The channels, scope, and decision this was run for. Fully loaded CAC, blended and by channel. Cohort margin and payback per channel. A decision for every channel, with what would reverse it. The cohorts too young to judge. And the claims you are refusing to make.
The refusals matter as much as the findings. An assistant asked to judge acquisition spend will happily produce a ranked list and a verdict, because that’s what the request sounds like it wants. Most of the stages above exist to make it say which numbers are heuristics and which cohorts are too young to judge, which is the part you’d otherwise have to catch yourself.
Run it once and you’ll have a real cost per channel, a margin-based ranking that may look nothing like the CAC ranking, and a decision for each one. Run it again next quarter on the same channels and you’re checking whether those decisions held, which is a shorter exercise than the first pass.