GCM Enterprise LLC · Marketing Verdict

Your ads grade their own homework. Your books don't.

Every month I line up what Google and Meta claim, the orders that followed an ad click, and what your books show was sold and kept. You get a signed verdict: which ads to grow, which to hold and test, and what to fix. I don't run your ads or take a commission. First verdict within 15 days of read-only access, on your latest closed month.

Book a 30-minute call See a specimen verdict

30 minutes · your break-even, and which of your records can tie to your close

Northcove Supply Co.Specimen Marketing Verdict · September 2026 · illustrative Return on ad spend (ROAS): sales per $1 of ads Meta's dashboard5.0x Meta, on the books (clicks only)2.0x Google's dashboard4.0x Google, on the books (clicks only)3.0x Break-even for this store2.5x On the dashboards, Meta ranked first. On the books, counting clicks only, it fell below break-even and Google carried the month. $25,000 of ads; break-even 2.5x from a 40% margin. Read the full specimen

The Problem

The dashboards look great. The P&L doesn't.

Each ad platform reports its own results, and two can claim the same order. Nobody checks those claims against the books, so three people get stuck.

01 · Finance

"I don't trust marketing's numbers."

Every platform grades its own homework, so finance discounts every ad report, the winners too.

You getROAS built on sales that tie to your month-end close.

02 · Marketing

"I know the ads work. I can't win budget."

Without numbers finance can check, strong channels get cut alongside weak ones, and tracking gaps get blamed on the team.

You getevidence finance can check, so channels that clear break-even can earn more budget.

03 · Owner

"I pay an agency and still can't tell if the ads make money."

The agency's report starts from the platforms' numbers, and nobody checks it against the books.

You getyour agency's results, checked against your books. Your agency keeps the work.

One signed verdict a month, in three steps.Built for companies doing $1M–$100M a year that sell or take leads online, whether a team or an agency runs the ads.

The Method

Three steps, every month.

See what's counted. Check the books. Make the call. Then keep score on what changed.

1 · See what's counted

Is every sale counted once, and only once?

I start with tracking you can see failing. Tags, the small pieces of code that tell Google or Meta a sale happened, break after a theme update, a checkout change or a new cookie banner. Platforms also take view-through credit: a sale credited to an ad the buyer saw but never clicked. And two platforms can claim the same order. That's default platform counting, not your team's mistake.

2 · Check the books

What was sold and kept, not what was claimed.

Each store order or CRM lead is credited once, to the ad clicked before it, then tied to your month-end close. Orders claimed twice, view-through credit, and sales tax and refunds each come out as a separate step. In the specimen below: $110,000 claimed, $80,000 after an ad click, $65,000 sold and kept.

3 · Make the call

Grow what clears break-even. Hold and test the rest.

Break-even = $1 ÷ your margin. Margin is what $1 of kept sales leaves after product, shipping, payment fees and return costs; finance calls it contribution margin.

50% 2.0x · 40% 2.5x · 25% 4.0x

Clear it on your books and the channel grows, in steps. Miss it and it's tested before any cut, never cut on a dashboard number alone.

Then · Keep score

Did last month's changes happen, and did they pay?

Every change gets an owner, a date and a rule for whether it worked, agreed before the money moves. Next month's verdict checks each one.

The Deliverable

This is what lands on your desk.

One month's verdict for an illustrative online store (the same one as in my LinkedIn example), with $25,000 of ads. The company, names and figures are invented. The structure and the depth are what a client receives.

Specimen · fictional company, illustrative figures, not a forecast · client verdicts are confidential

GCM Enterprise

Marketing Verdict

Northcove Supply Co.

Period: September 2026 | Google Ads, Meta, Shopify, GA4 and the general ledger, tied to the September close | Prepared 9 October 2026 | Signed: Graham Miller

Verdict

Together, the ads just cleared break-even, and Google carried Meta. $25,000 of ads brought $65,000 of sales that followed a click, matched to the books and kept: 2.6x against a 2.5x break-even, +$1,000 of contribution after ads (what's left after product, shipping, fees, returns and the ads themselves). The dashboards showed 4.4x because they count $15,000 of orders twice and $15,000 of Meta view-through credit, and count $15,000 of sales tax and refunds as sales. That's default platform counting, not the team's mistake.

Google clears break-even at 3.0x (+$3,000) and lost 22% of eligible searches to budget: grow it in $1,500 steps, keeping each step only if it adds at least $2.50 of kept sales per added $1 of spend (about $3,750 for a full month at the higher budget). Meta ranked first on the dashboards at 5.0x but shows 2.0x on clicks alone (−$2,000), as in August; if its view-through credit is earned and shared orders are split evenly, it is nearer 3.4x. Break-even sits inside that range, so hold Meta at $10,000 through the holiday peak and let a six-week regional holdout from 11 January decide. First, restore the Meta checkout pixel a theme update removed on 29 September: it has cut the purchases Meta can see by about a fifth, and Meta's automated bidding learns only from the purchases it can see.

Net sales (books)$180,000ties to the September close

Ad spend$25,00013.9% of net sales

Ad sales, matched and kept$65,0002.6x · of $110,000 claimed

Break-even return2.5x$1 ÷ 40% margin

Truth · from claimed to kept

The platforms claimed $110,000. $65,000 was sold and kept.

What the platforms claim (Google $60,000 | Meta $50,000)$110,000

Counted twice: orders both platforms claimed−$15,000

View-through: saw a Meta ad, never clicked−$15,000

Orders that followed an ad click$80,000

Sales tax and refunds−$15,000

Sold and kept, matched to the books (Google $45,000 | Meta $20,000)$65,000

How each step is caught: counted twice, by crediting each store order once, to the last ad clicked before it (from the click ID on the order); view-through, from each platform's own split of click and view credit, set aside until a test shows what it caused; sales tax and refunds, from the books. After a click: Google $55,000, Meta $25,000. Store net sales tie to the ledger's $180,000 for September before any return is quoted.

Profit · by channel

Judged by your books, the ranking flips.

Channel	Spend	Dashboard	Your books (clicks only)	vs 2.5x	Verdict
Google	$15,000	4.0x	3.0x	above	Keep + grow+$3,000 contribution after ads. Grow in $1,500 steps; keep each only if it adds $2.50 of kept sales per added $1.
Meta	$10,000	5.0x	2.0x	below	Hold + test−$2,000 on clicks alone; nearer 3.4x if its view-through is earned and shared orders are split evenly. Hold at $10,000 through the peak; a regional holdout from 11 Jan decides.
Both	$25,000	4.4x	2.6x	above	+$1,000 contribution after ads, on clicks alone.

"Dashboard" is the return each platform reports. "Your books (clicks only)" counts sales that followed a click on that platform's ad, credited once to the last ad clicked, after sales tax and refunds. Last-click credit favours the last touch, usually search: Google's 3.0x includes brand searches by people who may have bought anyway, and Meta's 2.0x leaves out its view-through credit. That's why each Google step is judged on what it adds, and Meta gets a test, not a cut. Break-even = $1 ÷ margin; margin here is 40% (contribution margin, not gross margin), so each $1 of ads needs $2.50 of kept sales. Contribution after ads = kept sales × 40% − ad spend. Only a holdout test (ads off in some regions, compare sales) proves cause.

Tracking · what each platform can see

A theme update on 29 September removed Meta's checkout pixel. Meta now sees 74% of store orders, down from 92%.

70%85%100%

Google tag (via GA4)94% 93%

Meta pixel + server feed92% 74%

1–28 Sep29 Sep – 7 Oct29 Sep – 7 Oct, pixel missing

Purchases each platform's tag reported, as a share of store orders, compared day by day. Google's tag runs through a separate app and was unaffected. Meta's server feed still reports the purchases it can match. Meta's spend and the store's daily orders held steady over the same days, so the drop is the pixel, not demand. The break touched only the last two days of September; October's verdict will carry all of it.

Execution · moves for October

ID	Move	Owner · by	What it decides	How we'll know
OCT-01	Restore the Meta checkout pixel removed by the theme update	Web developer · 12 Oct	Measurement	Meta sees 90% or more of store orders again
OCT-02	Google: raise budget in $1,500 steps, first to $16,500	Agency · from 14 Oct	Growth	Each step stays only if Google's kept sales rise by at least $2.50 per added $1 of spend at the next close (about $3,750 for a full month at the higher budget); otherwise roll back. Directional; only a test proves cause.
OCT-03	Meta: hold at $10,000 through the holiday peak; from 11 Jan, ads off in a quarter of regions for six weeks, same budget per remaining region (Meta runs at about $7,500 a month while the test is on, down from $10,000)	Agency · 11 Jan – 21 Feb	Decides $10,000 a month	Against each region's sales in the six weeks before: if regions with Meta out-sell the rest by $2.50 or more of kept sales per $1 of Meta ads, its view-through counts as earned and Meta gets a grow move; below that, Meta comes down in steps.
	This month			+$1,000 contribution after ads, on clicks alone. OCT-02 and OCT-03 decide the next step. Illustrative; no outcome is assumed.

Sources: Google Ads and Meta (spend, claimed sales, click and view credit); Shopify orders with their click IDs; GA4 sessions and purchase events; the general ledger (net sales, sales tax, refunds) and the margin inputs you confirm (product, shipping, payment fees, return costs). Every figure is saved with its inputs and its formula.

Confidential · GCM Enterprise

That's one illustrative month. Want your own break-even?

Book a 30-minute call

You leave with your break-even and which of your records can tie to your month-end close. After that, access is read-only: about 15 minutes per account, nothing can be changed, and your agency keeps the work.

How It Works

Exactly what happens, step by step.

Here is every step, who or what does it, and what it touches. Nothing in your accounts is ever changed.

1
Days 1–3 · You
Give read-only access
About fifteen minutes per account, with step-by-step instructions: Google Ads Read only, Meta View performance, GA4 Viewer, a store staff login limited to viewing orders and reports (or a read-only CRM user for leads), and your accounting file as Reports only (QuickBooks) or Read only (Xero). None of these roles can change a campaign, a budget or a setting. If your agency administers the ad accounts, it adds the user; nothing changes in how it works.
2
Days 4–10 · Software
Pull, match and calculate
Software pulls spend, clicks and claimed sales from each ad platform, every order or lead from your store or CRM, and the closed month from your books. It ties store net sales to your books, after sales tax and refunds, before any return is quoted. It credits each order once, to the last ad clicked before it (from the click ID on the order), so an order Google and Meta both claim is counted once. Then it calculates each channel's kept sales, contribution after ads, and your break-even. Every figure is saved with its inputs and its formula, and the same code runs every month.
3
Days 10–14 · AI drafts, I decide
Draft, check and decide
An AI model reads the calculated figures and last month's record and drafts the findings. It doesn't do the arithmetic, doesn't see your logins, and doesn't decide. I check every finding against the numbers and correct or remove what doesn't hold. Whoever runs your ads sees the draft findings and can answer them, with data, before I sign; their answer goes in the verdict next to mine.
4
Day 15, then monthly · With you
Your signed verdict and the meeting
The first verdict covers your latest closed month: what clears break-even and can grow, what to hold and test, what's broken, and the changes for next month, each with an owner and a date. After each month's close, a new verdict scores last month's changes.

Software

Does: pulls the data, matches orders and calculates every figure, the same way each month.

Never: makes up a figure. Every figure traces to its inputs, and where matching is uncertain, the verdict says so.

AI model

Does: reads the calculated figures and drafts findings and questions.

Never: does arithmetic, logs into your accounts, or makes the call.

Graham

Does: checks, corrects and decides, then meets with you.

Always: signs the verdict and answers for every number in it.

Book a 30-minute call See the specimen ↑

Execution

A verdict is only worth the change it causes.

Every recommended change goes on a numbered list with an owner, a date, the result expected and how it will be checked, agreed with you before anything changes. Your agency or your team makes the change; I have read-only access. Next month I check that it happened and whether it paid, and the list keeps the score, wins included.

Move register · specimenRe-checked every month. Every test has a budget, a deadline and an exit, agreed before the money moves.

ID	Move	Raised	Status	Result
OCT-01	Restore the Meta checkout pixel	Oct	In progress	Due 12 Oct. Done when Meta sees 90% or more of store orders; escalates if still under 85% on 19 Oct.
OCT-02	Grow Google in $1,500 steps	Oct	Agreed	Each step kept only if it adds $2.50 of kept sales per added $1 at the month's close; directional.
OCT-03	Meta regional holdout	Oct	Agreed	Starts 11 Jan, after the holiday peak; reads 22 Feb. Decision rule fixed now.
AUG-03	Show Meta's view-through credit in its own column	Aug	Done	The agency report splits click and view credit; Meta's test set for after the peak.
AUG-02	Raise Google from $12,000 to $15,000 in two steps	Aug	Grew	Each step added more than $3,750 of kept sales at the August and September closes; the higher budget stays.
JUN-02	TikTok test, $3,000 over 6 weeks	Jun	Stopped	Below break-even on the books at the exit date. Stopped as agreed.

Your agency stays

I don't run your ads or take a commission. I check what they earned.

Your agency or in-house team keeps the work. They get specific, numbered requests, not opinions, a scoreboard they can check themselves, and evidence to win budget for what clears break-even.

Questions

What finance, marketing and owners ask first.

What do I get from the 30-minute call?

Your break-even, worked out from your own margin, and a plain answer on which of your records (orders or CRM leads, ad accounts, books) can tie to your month-end close. If you're in marketing, bring finance: read-only access to the books needs their OK, and you'll both leave with the same break-even. If it's a fit, your first signed verdict covers your latest closed month and arrives within 15 days of read-only access.

Will this be used to cut my budget?

Sometimes, but only after a test, never on a dashboard number alone. It's built to show what has earned more budget too: a channel that clears break-even on your books gets a grow move. View-through credit is left out of the books figure because the books can't confirm it, but that isn't treated as proof it earned nothing, which is why a channel near break-even is held and tested. And whoever runs your ads sees the draft findings, and can answer them with data, before I sign.

Where does break-even come from?

$1 ÷ your margin. Margin is what $1 of kept sales leaves after product, shipping, payment fees and return costs; finance calls it contribution margin. At 40%, each $1 of ads needs $2.50 of kept sales. It isn't gross margin, which leaves out shipping, fees and returns and makes ads look better than they are.

How do you know the ads caused the sales?

Matching shows which sales followed an ad click. It can over- or under-count any platform, so on its own it doesn't prove cause. Only a holdout test (ads off for a test group, compare sales) shows what the ads caused. Where a decision is big enough, the verdict proposes one, with the decision rule fixed in advance and the dates kept clear of your peak season.

Do you replace my agency, or its report?

No. Your agency or team keeps running the campaigns; if it administers the ad accounts, it only needs to add a read-only user. Its report starts from what the platforms record; the verdict starts from your books, what was sold and kept, and checks the platforms against that. Good agencies welcome a scoreboard they didn't write themselves. Nor is this a fractional CMO: a fractional CMO runs your marketing; I tell you what it earned.

What if my tracking is broken?

It usually is, somewhere. That's typically the first finding: which tag is broken, since when, how many sales it's hiding, and the exact fix, written for whoever runs your website or your ad accounts.

What systems do you work with?

Google, Meta, TikTok, LinkedIn or Microsoft Ads; Shopify, WooCommerce or your CRM for leads; QuickBooks, Xero or your accounting team's monthly close. Each gets a read-only role; How it works, step 1 lists the common ones, and I send step-by-step instructions for the rest. If you also sell through Amazon, wholesale or stores, online sales tie to their own revenue line in your books; sales elsewhere aren't credited to ads unless a holdout test measures the lift.

Who sees our books, and what happens to the data?

Me and the software, plus the AI provider that drafts findings from calculated figures (never your logins or ledger). Whoever runs your ads sees the draft findings, not your ledger. Every role is read-only and you can remove it at any time.

Is this an audit?

No. It's management information for your own decisions. It isn't an audit, review or attestation, and it gives no assurance on your financial statements.

What does AI do here, exactly?

It runs the analysis. Each month the system rebuilds the channel economics, reconciles platform numbers against your books, and names every risk and opportunity it identifies — including the ones that contradict the dashboard. It doesn't log into your accounts, doesn't move anything, and doesn't decide. I review every finding, correct what's wrong, and sign the verdict. Your data isn't used to train AI models: the model runs under commercial terms that exclude training on your data.

What does it cost, and how long is the commitment?

A fixed monthly fee, typically $1,000 to $3,000 a month, quoted after the call once I know your channels and systems, and never a share of your ad spend. The first verdict stands on its own. After that it's month to month.

Book a 30-minute call 30 minutes · month to month

Let's Talk

Find your break-even.

Thirty minutes with me, Graham Miller, in Virginia Beach: a bank SVP background in data and analytics, ALCO and credit risk, sharpened by running my own companies. You leave with your break-even, worked out from your margin, and which of your records can tie to your month-end close. If it's a fit, your first signed verdict covers your latest closed month and arrives within 15 days of read-only access. I don't run your ads or take a commission.

Book a 30-minute call graham@gcmenterprise.com

Thirty minutes · No obligation · Typically $1,000–$3,000 a month, fixed, never a share of your ad spend