Agency deck
$1.08M
September revenue claimed against the paid program in the monthly deck
Stan Consulting · Marketing Atlas · Case File · Agency Burn
The Agency Report That Looked Good Until the Bank Account Disagreed.
case_type: composite cluster: agency-burn published: 2026-05-07
01
A Series-A funded DTC apparel brand. Six-point-four million annualized. Two-hundred-twenty-five thousand monthly paid spend through an outside agency: one-hundred-thirty in Google Ads, seventy-five in Meta, twenty in display. The agency's monthly deck shows a four-point-eight-x blended ROAS, a thirty-day rolling CAC of forty-two against a target of fifty-five, and a YoY click-through-rate up eighteen percent. The CFO opens the bank statement and the deposits run one-hundred-eighty thousand below the agency-claimed revenue. The marketing director cannot reconcile.
That is the composite. The names change. The shape does not.
The brand had been on the agency for eleven months at the point the audit was scoped. The relationship had been signed during the post-Series-A growth push as the founder team scaled out. The agency was on a sixteen-thousand-monthly retainer covering two channels (Google and Meta) and a separate twelve-thousand fee for the display program. The retainer included a campaign-management scope, a creative-refresh cadence, a weekly check-in, and the monthly board deck. The agency had two named operators on the account, a reporting analyst, and a creative producer.
The reporting stack was the agency's. A Looker Studio dashboard pulling Google Ads conversions, Meta pixel conversions, GA4 sessions, and a hand-keyed Shopify revenue total. The deck was assembled by the analyst against the dashboard each month, walked through on a video call with the CMO and the marketing director, and sent to the CFO and the board observer as a PDF. The cadence had not changed since the agency took the account. The deck had not been audited since it had been built.
The CFO had been quietly skeptical for two months. Cash deposits had been running flat against the agency's rising revenue claims. The CFO opened the September deck against the September bank statement and saw a one-hundred-eighty-thousand-dollar gap. The CFO took the gap to the marketing director. The marketing director took it to the agency. The agency answered with a sixteen-page response document arguing that platform-attributed revenue and bank-deposited revenue were different things and that the deck was reading correctly. The CFO read the document and decided to scope an outside audit.
The audit landed on the desk on a Thursday. Seventy-two-hour written verdict. The brief from the CFO was one line: tell us whether the agency deck is reading honestly, and if not, name the conventions inflating the picture and what the honest read would look like.
- Stage
- Series-A funded · DTC apparel
- Annualized revenue
- $6.4M
- Monthly paid spend
- $225K ($130K Google · $75K Meta · $20K display)
- Agency retainer
- $28K monthly across the two scopes
- Agency tenure
- 11 months on the account
- Reported September ROAS
- 4.8x blended
- Reported September CAC
- $42 against a $55 target
- Bank vs deck spread
- $180K below agency-claimed revenue
- Engagement
- Conversion Second Opinion · written verdict
02
Six numbers from the September deck and the September bank statement. The deck told one story. The bank told a different one. The marketing director read both for two weeks before scoping the audit.
Bank deposits
$900K
Net Shopify deposits to the operating account for September
Agency deck
4.8x
Blended ROAS as reported on slide three of the September deck
Bank-honest read
3.0x
Bank-deposited revenue against the same paid spend in September
Agency deck
$42
30-day rolling blended CAC reported against a $55 target
Bank-honest read
$71
CAC against bank-confirmed new customers, the read the CFO ran
The deck and the bank disagreed on every line that mattered. The agency-claimed revenue was twenty percent above the deposits. The blended ROAS in the deck was sixty percent above the bank-honest read. The CAC in the deck was forty-one percent below the bank-honest read. None of the deck numbers were random; each was produced by a defensible-on-its-own counting convention. Every one of those conventions ran in the agency's favor.
The agency response document had argued that platform-attributed revenue and bank-deposited revenue measure different things and that the deck was correct on its own terms. The argument is technically true and operationally hollow. The deck was the only revenue read in front of the board. The fact that the deck would never reconcile to the bank without an inflation factor of twenty percent was the whole problem.
03
Four explanations were on the table when the audit started. Each one was almost-right and pointed away from the layer that actually mattered.
Wrong reason 01
"Platform-reported revenue is not the same as deposited revenue; the deck is correct on its own terms." The agency's defense. The argument is that paid platforms report attributed revenue against their own conversion windows and the bank reports deposits net of refunds, declines, and timing lag. The argument is half-true; the half it leaves out is that the agency had folded all four counting conventions in the deck's favor without ever calling them out. The deck did not say "this is platform-attributed." The deck said "September revenue: one-point-zero-eight million." That is a representation about the business, not a footnote about a convention.
Wrong reason 02
"The disagreement is a tracking issue; install server-side and it closes." The technical-engineering read. The argument is that the platform-attributed numbers and the Shopify-confirmed numbers diverge because of pixel loss, cookie blocking, and iOS-fourteen attribution gaps. The argument fails because the spread runs in the opposite direction. Tracking loss makes platforms under-report against Shopify. The agency deck was over-reporting against Shopify by twenty percent. The defects in this account were not making the deck conservative; the conventions in this account were making the deck inflated. Server-side install would not narrow this gap; it would slightly widen the platform-claimed numbers, not the deposits.
Wrong reason 03
"The CAC target was wrong; the agency was hitting an old target." The internal-blame read. The argument is that the fifty-five-dollar CAC target had been set when CACs across the category were higher and that the deck's forty-two-dollar number was a real improvement. The argument fails because the forty-two number was not the bank-honest CAC. The bank-honest CAC was seventy-one dollars, against the same fifty-five target. The deck was not beating an old target. The deck was missing the current target by about thirty percent and reporting a beat anyway.
Wrong reason 04
"The CFO does not understand performance marketing; the deck is fine." The agency-side cultural read. The argument is that finance teams read marketing reports without context and that the spread will close if the agency walks the CFO through how attribution works. The argument fails because the CFO understands attribution adequately and the agency does not have a defense against the bank deposits. The CFO's job is to defend cash. The agency's job is to produce a deck that survives a CFO's read against cash. When the deck does not survive, the deck loses.
All four explanations let the team defer the decision the audit was scoped to force. The structural defect was upstream of the platforms and upstream of the agency's defense. None of the explanations went there.
04
Four reporting conventions, all defensible on their own, all pointed in the same inflationary direction, and never named in the deck. The deck did not say what it was reading. The agency was reading well within its own definitions. The operator had no read at all.
The audit decomposed the September spread into four named conventions. None of them was the cause on its own; the cause was that all four were stacked silently, with no footnote, no caveat, and no separate read against the bank.
Convention one. Platform-reported conversions ungrouped from actual revenue. The deck's revenue line was the sum of Google-Ads-claimed revenue and Meta-Ads-Manager-claimed revenue, not the sum of Shopify orders attributed to paid traffic. Google reports conversion-value against attributed conversions; Meta reports purchase-value against pixel-attributed purchases. Each platform is measuring its own attributed share of the funnel. Adding them produces a number larger than the funnel itself, because users counted by both platforms get counted twice. The deck's one-point-zero-eight million was Google-claimed plus Meta-claimed. The bank's nine-hundred thousand was the funnel itself. The gap was double-counting at the platform level, dressed up as a revenue line.
Convention two. View-through conversions counted as last-click. Meta runs a one-day-view window by default. Roughly thirty-five percent of Meta-claimed conversions in September were view-through: users who saw an ad, did nothing, and converted within twenty-four hours through some other surface. The deck reported these as "Meta conversions" with no distinction from clicks. View-through is a defensible diagnostic; presented in a board-pack revenue line as if it were a click is not. The September deck's Meta-claimed revenue included roughly seventy-thousand of view-through revenue presented as click-driven. The CFO had no way of knowing that.
Convention three. Branded-search conversions credited to "search expansion" campaigns. The Google Ads account included a brand campaign and three "search expansion" campaigns targeting category and competitor terms. The brand campaign was assigned a separate Google account ID for billing reasons; the search-expansion campaigns ran under the main account. The deck's CAC line was calculated against new-customer revenue divided by spend on the main account only. New customers acquired through the brand campaign were counted in the numerator (because they were new) and not in the denominator (because the brand spend was billed separately). The arithmetic was producing a CAC roughly twenty-five percent below the truth.
Convention four. ROAS calculated on gross attributed not gross deposited. The blended ROAS line read four-point-eight-x. The numerator was the sum of platform-claimed conversion value across Google plus Meta plus display. The denominator was paid spend net of platform fees but not net of refunds, declines, payment-processor fees, or shipping discounts. The bank-deposited number was net of all of those and net of timing lag. ROAS calculated on gross attributed produces a structurally inflated ratio. ROAS calculated on gross deposited produces a CFO-defensible ratio. The deck used the former and labelled it "blended ROAS." The label did not say "platform-claimed ROAS." The label said "ROAS." That is the convention that was running below the surface and producing the headline ratio.
Four conventions, each defensible in isolation, stacked in the deck without a single footnote. The agency was honest within each convention. The deck was inflated against the funnel. The CFO read the deck against the funnel. The mismatch was structural; it would survive any analyst handover, any pixel install, any conversation with the agency's response document.
05
The decomposition reads in three layers. The data, the report, and the operator's read of the report. The agency was operating cleanly at layer one. The deck was assembled cleanly at layer two. The third layer, the operator's read, was missing. Without it, the team had no way to ask whether the deck reconciled to the bank.
L1
Captured data
Technical layer
The data layer was largely intact. Google enhanced conversions firing. Meta pixel and Conversions API matching at ninety-one percent. GA4 enhanced ecommerce in place. Shopify orders flowing into the financial stack against the bank reconciliation each Friday. The agency had inherited a working layer-one stack and had not made it worse.
The technical layer was not the source of the spread. The spread was twenty percent. Two-percent technical defects do not produce twenty-percent spreads. The spread was running through the layer above.
- Google enhanced conversions · firing, validated
- Meta Conversions API · 91% match against pixel
- GA4 enhanced ecommerce · configured, mapping intact
- Shopify orders · reconciling cleanly to the bank weekly
- Pixel duplicate rate · under 1% at confirmation
L2
Reported deck
Reporting layer
The reporting layer is where the four conventions lived. The agency was assembling the deck against the platform-claimed numbers without naming the conventions in the deck or against the bank-deposited numbers. The deck was a defensible artifact within the agency's chosen conventions. It was an indefensible artifact against the operating reality of the business.
This is where most operators stop, conclude the deck is "agency reporting" and that the agency must know what it is doing, and sign the next quarter's budget anyway. The reporting layer is not broken. It is unmediated. The fix is not a new agency; the fix is a written rule about what the deck is allowed to read against and how the conventions get footnoted.
- Revenue line · sum of platform-claimed, not Shopify-confirmed
- View-through · folded into the conversion line silently
- Brand-spend · excluded from CAC denominator
- ROAS denominator · gross attributed, not gross deposited
- Convention footnotes · absent across the deck
L3
Operator's read
Judgment layer
The judgment layer was empty. The marketing director read the deck without a written rule about which numbers map to the bank and which do not. The CMO read the deck the same way. The CFO read the deck against the bank and saw the spread; nobody on the marketing side had a structured response, because nobody on the marketing side was operating against a written read.
Without the document, every read defaulted to the agency's. The agency had a conventionally-honest deck and a financially-inflated representation. The marketing team had no counterweight. The CFO had no source-of-truth read against which to challenge the deck. The agency was not lying; the agency was operating in the absence of a written operator-side read. That is the structural condition that produces agency burn.
- Source-of-truth document · never written
- Bank-honest revenue read · never assembled
- Convention-footnote rule · not enforced
- Deck-vs-bank reconciliation cadence · not assigned
- Variance threshold for CFO escalation · not set
06
The audit's written verdict named the install order. Order matters. Renegotiating the agency before installing the operator-side read is a fight without a foundation. Installing the read first gives the marketing team something to negotiate from.
The audit drove into the Conversion Second Opinion engagement format and from there into a thirty-day install. The honest-reporting framework below is what was installed.
-
Day one · Declare Shopify-net the source of truth for revenue and CAC
The decision is written and signed. Shopify-net (gross orders less refunds, less discounts, less shipping discounts, less payment-processor fees) is the revenue read for board purposes. Bank-deposited is the financial-reconciliation read. Platform-claimed is a diagnostic read footnoted in the deck. The agency's deck reads against Shopify-net for the headline revenue line; platform-claimed appears in the diagnostic appendix. The decision takes one document and one signature.
-
Week one · Rebuild the revenue line against Shopify-confirmed paid orders
The deck's headline revenue line is rebuilt as Shopify orders with a paid-traffic UTM on the originating session, summed at gross-net. Google and Meta no longer add to a single revenue total in the deck. The agency continues to optimize against its own platform reads internally; the deck's revenue line is now the operator's read. View-through is reported as a separate diagnostic line, footnoted, with the convention named.
-
Week two · Rebuild the CAC line against bank-confirmed new customers
The CAC denominator is rebuilt as total paid spend across all accounts (including the brand-search line item that had been excluded). The CAC numerator is bank-confirmed new customers in the period, not platform-attributed conversions. The CAC reported in the September deck would have read at seventy-one against the rebuilt definition; the new convention surfaces the gap to the fifty-five target instead of hiding it. The agency is held to the rebuilt CAC definition going forward.
-
Week three · Add the deck-vs-bank reconciliation chart
One chart added to the deck's first page: monthly Shopify-net revenue, monthly bank-deposited revenue, monthly platform-claimed revenue, all on the same axis with variances expressed against Shopify-net. Variance from Shopify-net exceeds eight percent: investigated before the next budget decision. The chart is the operating contract for the engagement and the artifact the CFO reviews each Monday. The agency owns producing it; the marketing director owns reading it; the CFO owns escalating against it.
-
Week four · Renegotiate the agency scope against the rebuilt deck
The agency retainer is renegotiated. The scope retains campaign management, creative refresh, weekly check-in. The reporting scope is rebuilt so the deck reads against Shopify-net by default and platform-claimed only in the appendix. The retainer is held flat for ninety days while the rebuilt deck runs against the bank. The agency either holds the bank-honest read or the relationship rewinds. The position is the read; without it, the renegotiation has nothing to land on.
-
Month two onward · The board read
The board pack now leads with one slide: Shopify-net revenue and bank deposits side-by-side, paid spend net of all fees, blended ROAS calculated against deposits, CAC against bank-confirmed new customers. Platform-claimed numbers and view-through diagnostics live in the appendix with named conventions. The CFO has the answer she did not have in September. The marketing director has a read she can defend. The agency has a contract she can renew or unwind against a read both sides agree on.
07
The deck was honest within its conventions and dishonest against the bank. The agency was not lying. The operator was not reading. Both things were true. That is the lesson the marketing director kept turning over for the week after the install.
The CFO's September question was not really a tracking question. It was a reporting-conventions question. Which numbers in the deck correspond to numbers in the bank, and which numbers in the deck do not. The agency had assembled a deck that answered the first half and elided the second half. The CFO read the deck against the bank and saw the elision. The fix was not a tracking install. The fix was a sentence: Shopify-net is the headline read; everything else is footnoted. The sentence took ten minutes to write and a month to install.
The lesson is that any DTC operator running paid spend above one-hundred-thousand monthly through an outside agency needs the bank-honest source-of-truth document before the next monthly deck. Without it, the operator is reading whatever the agency assembles, in whichever convention favors the agency's work, with no counterweight. The default reporting stack does not include the document. The default is the failure mode. The cost of writing the document is one signature. The cost of not writing it is a one-hundred-eighty-thousand-dollar gap the CFO finds anyway.
Five Cents · Stan's note
Five Cents
The part of this case file that I keep coming back to is the September call where the marketing director walked the CFO through the agency deck for the third time and the CFO finally said: I do not need you to defend the deck, I need you to explain why the bank does not match. That sentence is the whole job. The marketing director did not have an answer because she had been reading the deck the agency was assembling, in the conventions the agency was choosing, with no separate read of her own. There was no operator-side read. The agency was filling the vacuum, not maliciously, just operationally, the way agencies always fill the vacuum when the operator does not have one.
What I want operators to take from this is that an agency monthly deck is not an operator-side read. It cannot be one. Even an honest agency, even a well-intentioned one, is going to assemble a deck inside the conventions that favor its work, because the conventions that favor its work are the ones that make the relationship renewable. That is not a moral failure on the agency's part. That is the retainer model doing what the retainer model does. The fix is not a better agency. The fix is the operator-side read that the agency's deck is benchmarked against.
What this case file is for: if your agency deck shows healthy ROAS and your bank deposits do not, and you cannot reconcile the two on a phone call with the agency, you have this case file. The Conversion Second Opinion delivers the verdict in seventy-two hours. The next move is the read; the read is what the engagement produces, and the read is what the renegotiation hangs on.
Stan Tscherenkow · Marketing Atlas · 2026-05-07
09
Each link below points at a related Atlas page that handles a piece of the case file in more depth. Reference pages give the definition. Position pages give the firm's defended doctrine. The hub gives the map.
If this is the pattern in your account
Read the deck against the bank. Then renegotiate.
If the case file maps to your account — agency-claimed revenue running ahead of bank deposits, ROAS reading better than cash, CAC reading lower than reality, the CFO question still open — the engagement that runs this diagnostic is the Conversion Second Opinion. A written verdict against the four-convention reconciliation framework, delivered in seventy-two hours. If the verdict says install, the Sprint engagement runs the rebuild. If the verdict says hold, you keep the read and act on it yourself.