Quantitative analytics for Amazon and Shopify brands · $1M–$20M/yr

We find the profit your Amazon and Shopify numbers are hiding.

We run the math on your Amazon account, your Shopify store, or both — and execute the profit fixes for you: pricing, ads, inventory, and the reimbursements Amazon owes you. Your written Teardown lands inside 24 hours of your exports. You spend fifteen minutes once, then three minutes every two weeks, and keep the margin. First month free, no card — at day 30 the ledger decides, not us: if we didn't find more than we cost, no invoice exists.

Get your free Profit Teardown

$0 · first month free · four answers start the call · Teardown on video inside 24 hours · no card, ever

The margin problem, in four numbers

Selling more while keeping less.

~50%
Amazon's effective take of seller revenue
67%
Sellers who raised prices in 2024
~60%
Of those ended up less profitable
5.4
Tools the average $1M+ seller runs — none decide

Sources: Marketplace Pulse · SmartScout seller survey 2025 · Fortune · industry operations research

Every number up there is a math problem — and math problems have answers. Whether the fee line says Amazon, Shopify Payments, Meta or UPS, the difference is whether anyone actually runs the math on your catalog. That's the entire service.

What you actually get

Three things land on your desk.

Not a dashboard subscription. Specific documents and decisions, on a monthly rhythm.

Sample · demo data

The Profit Teardown

MARGIN.DECOMPlatest period
WIDGET-BLUE +40%
WIDGET-RED +26%
GADGET-PRO −7%
Platform feesCOGS AdsNet

Every SKU's true net margin — every Amazon or Shopify fee, landed cost, and ad spend honestly allocated — plus stockout probabilities, elasticity reads, and your ad break-even. Written report and a recorded voice-over-screen walkthrough.

Sample · demo data

Executed Corrections

CORRECTIONexecuted
Advertising

Negative-matched 11 search terms that spent with zero attributed sales — $1,940 of pure bleed, stopped.

Executed Aug 24 expected $1,940 · now being measured

3–5 corrections a cycle, each with its expected dollar impact — executed for you under a standing mandate with a veto window. Every two weeks, a three-minute brief: what we found, what we changed, what it's expected to earn. A year of this is 40–60 measured, compounding moves your competitors make on gut — without you pushing a single button.

Sample · demo data

The Decision Ledger

DECISION.LEDGERcumulative
CorrectionExpectedMeasured
Cut bleed terms$1,940$2,105
Price test +4%$860$1,240
Trim campaign spend$1,200$890
Measured to date$4,235

Every correction's expected and measured dollar impact, on one ledger — including the calls that came in under. Agencies get fired because nobody can prove what they earned. The Ledger is that proof, walked through together on your review calls — and you can export it the day you leave.

See your version — get the Teardown
How it works

One seat. The models run. Teardown back in 24 hours.

Day 0

Apply

A two-minute application — where you sell, revenue, model, catalog size — then you book the 20-minute call on the spot. Your answers ride along on the booking, so the call starts from your numbers. Every brand that applies gets a seat.

2 minutes

One seat

Add one email to your Seller Central, or approve one collaborator request in your Shopify admin — a permissions-scoped seat covering reports, pricing, and ads. That seat is how the corrections get made in your account. For the numbers themselves, about fifteen minutes of exports through a secure upload page, once — the list is on the page and every step is spelled out.

Banking, settings, and everything else stay untouchable — and you revoke the seat in one click, any day. Prefer to send files instead? A secure upload page exists for exactly that — Seller Central reports or Shopify exports, your choice.

Within 24 hours

Teardown delivered

The models start the moment your seat goes live — ten thousand simulations per SKU. Written report plus a recorded walkthrough of your numbers, back in 24 hours.

Your call

Decide

Say yes and your first month runs free — full service, corrections executed, ledger running. At day 30 we compare what we found against what we cost, and if it did not clear the fee there is no invoice. If the Teardown finds nothing worth fixing, we tell you so — and you keep the full report.

Every 2 weeks

We execute

Models refresh, corrections get executed for you under a standing mandate — you hold the veto. A three-minute brief tells you what changed and what it earned. Price tests watched against Buy Box share daily.

Your total cost in time: one 20-minute call, two minutes granting a seat, then about three minutes a cycle watching your brief. We do the rest — including the clicking.

Get your free Profit Teardown
Method 01 / Formula 05, running live

Know your stockout risk before it happens.

We simulate the next 90 days of demand against your supplier lead times — ten thousand times per SKU. You get a reorder date, a quantity, and the exact probability of stocking out if you wait. A stockout costs you rank you'll buy back with ads; overstock is trapped capital plus storage fees. Both sides of that trade are priced in the model — no Amazon or Shopify-native tool does probabilistic inventory math.

See this run on your catalog
MONTE_CARLO.RUNn = 10,000 · horizon 90d
P50 units on hand P5–P95 envelope Safety stock
Method 02 / Formulas 02–03, running live

Know how far a price can move before it costs you.

For every SKU we measure how demand responds to price — with a 95% confidence range, not a hunch. Two-thirds of sellers raised prices in 2024; most of them lost money doing it, and sellers report losing ~80% of Buy Boxes on the way up. We tell you which SKUs can carry an increase, then watch your Buy Box share on Amazon or your conversion rate on Shopify daily, so a step too far is caught in days, not quarters.

See this run on your catalog
ELASTICITY.MAPSKU B0-7841 · ε = −3.68
Net profit index Current baseline = 100
Method 03 / Formula 04, running live

Know where your next ad dollar stops paying.

Ad spend has a break-even point. We estimate your account's response curve and mark the spend level where one more dollar returns less than a dollar — and flag the 20–40% of budget that typically leaks to search terms that never convert, in dollars, by term. No ad tool, Amazon or Meta, derives what your target ACoS or ROAS should be — they optimize toward whatever number you typed in.

See this run on your catalog
SPEND.RESPONSEad spend response · $k / mo
Incremental revenue Break-even (ROAS 1.0) ▲ Bleed zone — cut
The problems you already know you have

You know the symptom. Here's the equation that ends it.

Every leak below is one you've already felt in a payout. Each one has a closed-form fix. These are the working formulas the Teardown runs on your catalog in 24 hours — the actual models, not illustrations. They run the same on an Amazon account and a Shopify store; only the fee lines change.

  1. 01

    “Sales are up. The payout isn't.”

    Fees, landed cost and ad spend live in three different reports. No single number tells you what a SKU actually keeps.

    MARGIN.DECOMPModule 04 · True net margin
    NetSKU = Revenue − Σ |platform fees| − Units × clanded − Ads × RevSKU / Revtotal

    clanded = unit cost + inbound freight + packaging + other. Fees summed across every line the platform charges — Amazon's referral, FBA and storage fees, or Shopify's payment processing, shipping labels and apps; ad spend allocated by revenue share within the period.

    What it hands you

    Every SKU's true net margin — and the SKUs that go negative once ads are honestly allocated. Those get fixed or cut first.

  2. 02

    “I raised prices and made less.”

    The price moved without anyone knowing how demand responds. Two-thirds of sellers raised in 2024; most of them lost money doing it.

    ELASTICITY.FITModule 02 · Price elasticity
    ln Q = α + ε · ln P (+ γ · ln Sessions) ε ± 1.96 · SE → 95% range

    Log-log OLS per SKU. Guardrails run first: at least 5 periods and a price that actually moved (CV ≥ 2%) — otherwise the answer is “not enough data”, never a number that looks like a finding.

    What it hands you

    ε per SKU with its confidence interval. Below −1, a cut grows profit. Between −1 and 0, a careful increase converts almost straight into margin.

  3. 03

    “Fine — what price, exactly?”

    Elasticity gives you the direction. Profit calculus gives you the destination, to the cent.

    PRICE.OPTIMUMPrice-testing program
    Π(P) = Q₀ · (P / P₀)ε × (P · (1 − f) − c) dΠ/dP = 0 ⇒ P* = [c / (1 − f)] · ε / (1 + ε)

    Valid for elastic SKUs (ε < −1). f = the channel's fee rate, c = landed unit cost. Every move is capped at ±5% per cycle and Buy Box share (Amazon) or conversion rate (Shopify) is watched daily — walk toward the optimum, re-measure, never teleport.

    What it hands you

    An exact new price per SKU, the expected profit delta per period, and its 95% range — then the next step once the result is measured.

  4. 04

    “I can't tell where my ad spend stops paying.”

    Your PPC tool or your Meta ads manager optimizes toward the ACoS or ROAS you typed in. Nobody derived that number from your margin.

    SPEND.RESPONSEModule 03 · Ad efficiency
    Sales(s) = a · sh / (kh + sh) Break-even: dSales/ds = 1 / m Bleed = terms: spend > $10, sales = 0

    Hill saturation fit per campaign, log-curve fallback. m = contribution margin before ads, so ads break even on profit, not on revenue. The 20–40% of spend leaking to search terms or ad sets that never convert is listed by term, in dollars — Sponsored Products, Google, or Meta.

    What it hands you

    The spend level where the next dollar returns less than it costs, and every leaking search term — negative-matched on the first cycle.

  5. 05

    “I stock out. Then I overstock.”

    A reorder sized on average velocity ignores that lead time and demand are both random. One of them always surprises you.

    MONTE_CARLO.RUNModule 01 · Probabilistic inventory
    L ~ LogNormal(lead time, σ = 0.2) r ~ N(μ, σr) · D ~ Poisson(r · L) P(stockout) = P(D > on hand + inbound) ROP = μd μL + 1.645 · √(μL σd² + μd² σL²)

    Ten thousand simulated lead-time demand paths per SKU. The closed-form reorder point (95% service, Z = 1.645) runs alongside as a cross-check — the two must agree.

    What it hands you

    A stockout probability per SKU, a reorder date and a quantity — from the whole distribution of outcomes, not one average.

  6. 06

    “How much safety stock is actually worth carrying?”

    A flat 95% service target treats a 40%-margin SKU and a 6%-margin SKU the same. The money doesn't.

    NEWSVENDORInventory economics · fee cliffs
    q* = Cu / (Cu + Co) Cu = unit margin + low-inventory fee Co = storage + capital + obsolescence Order-up-to = q*-quantile of D(L+7d)

    The critical fractile: the in-stock probability the money justifies, SKU by SKU. Capital at 12%/yr, obsolescence 2% of cost per cycle, Amazon's aged-inventory and peak-season storage cliffs priced in; on Shopify, your 3PL's rate or nothing at all.

    What it hands you

    The service level each SKU earns — 99% for a fat margin, 85% for a thin one — the order quantity, the wire amount, and a hold-vs-liquidate call on excess.

  7. 07

    “Amazon owes me money and I never file.”

    Amazon accounts only: 1–3% of FBA revenue is recoverable every year, and about 40% of it dies of neglect because the claim windows are short. A Shopify store has no warehouse losing units on its behalf.

    RECOVERY.EVReimbursement recovery
    EV = P(approved) × units × landed cost Windows: 30→60 d · 60→120 d · 0→60 d

    Warehouse loss and damage, refunds with no return, and damaged returns — reconciled from your own exports. P(approved) 0.85 / 0.70 / 0.60 by claim type, stated as assumptions. The Ledger books only what Amazon actually pays.

    What it hands you

    A dated, valued claim list with deadlines — filed before they close.

  8. 08

    “Fees crept up and I noticed a quarter late.”

    A $0.30-per-unit fee change or a conversion drop is invisible inside a monthly total.

    ANOMALY.SCANChange detection · every cycle
    z = (xlatest − median) / (1.4826 MAD) CUSUM: S⁺ = max(0, S⁺ + z − k) k = 0.5σ, alarm at h = 4σ

    Robust z on every fee, conversion and spend series; Page's CUSUM finds the period a level shift began. Only a crossed threshold counts as a finding, and the dollar impact is computed for the adverse direction only.

    What it hands you

    The date a fee, conversion rate or spend level shifted, and its dollar impact — caught in the cycle it happened, not the quarter after.

Cash runwayP(ruin) = paths below $0 ÷ 10,00090 simulated days, payouts every 14 on Amazon and daily on Shopify — the bridge-capital line, not a prophecy.
ConcentrationHHI = Σ si²Above 1,800 is “highly concentrated” by the DOJ/FTC rule — priced as the dollars at risk if your top listing is suppressed.
A bad monthVaR95 · CVaR95The 5th-worst month in 100, and how bad it is on average once you're there.
Forecast honestyMASE-ranked model ladderNaive → SES → damped Holt → Croston-TSB → seasonal, chosen by rolling backtest. Value added over naive reported as computed, even when it's zero.
Health ScoreΣ wi · scoreiSix sub-scores, each deduction tied to the dollar figure behind it.

Three of these run live on demo data further down — the inventory, elasticity and ad-response charts are the models, not artwork. The rest run the same way. You see all of them on your own numbers in 24 hours.

Run these on my catalog
Where we sit

Dashboards tell you what happened. Nobody tells you what to do next.

Your tool stack (avg 5.4 apps) PPC agency Fractional CFO Hubricon
Tells you what to do, with expected $Reports what happenedAds onlyBoard-level onlyEvery correction
Probabilistic inventory mathVelocity averages——10,000 simulations / SKU
Derives what your target ACoS or ROAS should beYou type the target inYou type the target in—From your margins
SKU-level true net marginOnly as good as hand-keyed COGS—P&L altitude, not SKUFees + COGS + allocated ads
IncentivesSubscription% of your ad spend · 49%/yr avg client churnRetainerFlat fee — advice never inflates our invoice
Measures each recommendation's realized impact———The Decision Ledger
Executes the changes for youYou operate themAds only—Pricing + ads, done for you — you never log in anywhere
Cost$29–$500/mo, you operate them% of spend$3,000–$12,000/mo$6,000/mo flat

Churn and pricing figures: industry benchmarks and published price lists, 2025–26

Fit check

Built for a specific kind of seller.

This is for you if

  • Your own brand on Amazon, Shopify, or both — roughly $1M–$20M/yr
  • 10+ SKUs with real pricing power
  • You own the goals; you're done owning the mental load of margin management
  • You can spend two minutes granting a seat, then three minutes a cycle watching your brief

This is not for you if

  • You want a day-to-day PPC agency — we execute the corrections our models find, not creative campaign management
  • You want another dashboard subscription
Get your free Profit Teardown Under $1M, or not private label? Book the call anyway — the Teardown is free, and we'll tell you straight on the call whether the fee maths for you.
Pricing & terms

Less than the margin you're leaking.

One flat price, anchored against what it replaces — and every term in writing below.

Step one · The Profit Teardown
$0

The same SKU-level teardown a fractional CFO would bill thousands for — free, because it's how we earn the retainer. Application-gated: the gate is fit, not slots.

  • 10,000 demand simulations per SKU — your stockout probabilities and exact reorder points
  • Every fee decomposed — true net margin per SKU across 40+ Amazon fee types or every Shopify cost line (payments, shipping, apps), COGS, and honestly allocated ads
  • Your demand curve, measured — which SKUs can carry a price increase, with a 95% confidence range
  • Your ad break-even, in dollars — the spend level where the next dollar stops paying
  • Back in 24 hours — written report + recorded walkthrough of your own numbers

If we find nothing worth fixing, we tell you so — and you keep it all.

Get your free Teardown
Step two · The Co-Pilot Retainer · first month free
$6,000/MO FLAT

Your first month is free — the full service, corrections executed, ledger running. If we don't find you more than we cost, walk away owing nothing; if we deliver, a short testimonial and your anonymized results are the price of the seat — and when a founder you send stays past their own day 30, your next month is on us.

What this replaces, at the prices those things actually sell for:

  • SKU-level finance modeling — margin, elasticity, forecastingFractional CFO
    $3,000–$12,000/mo
  • Ad break-even + bleed elimination — with the target derived, not guessedPPC agency
    % of spend, $2,000+
  • Probabilistic inventory planning — risk-priced reorders, not velocity averagesPlanning software
    $300–$500/mo
  • Elasticity-based price testing — with the math shown, Buy Box tracked dailyPricing AI
    $350–$650/mo
  • Done-for-you execution — corrections made for you under a standing mandate, veto always yoursSold nowhere
    at this depth
  • The Decision Ledger — every move's expected vs. measured dollars, cumulativeSold nowhere
  • Founder on every brief — the person who builds your models is the person who executesNot an account manager
Assembled from parts, four vendors$5,700–$15,000+/mo
One operator, one ledger$6,000/mo flat
The break-even math: sellers waste 20–40% of ad spend on search terms that never convert. On a $25k/month ad account that alone is $5,000–$10,000 — the fee covers itself before we touch pricing or inventory. And every dollar we claim, the Ledger has to prove.
First month free — walk away owing nothing Every month after: invoices never run ahead of your ledger Month-to-month, cancel anytime Invoiced by email, ACH, net-7 — no card on file Flat fee, never a % of your ad spend Your data and your ledger export free, anytime — including the day you cancel Founder on every account — no handoffs, ever
The honesty guarantee

If the Teardown finds nothing worth fixing, we tell you so — and you keep the full report. The first month is free either way — if we don't find you more than we cost, you never pay a dollar. And it never stops: our invoices can never run ahead of your ledger. Any month the value we've measured and identified has not covered what we've billed, that month's invoice is void.

Who runs your numbers

No headshot. Here's what you get instead.

Hubricon is a founder-operated practice. The person who builds your models is the person on your calls — no account managers, no outsourcing. Trust the terms, not a face:

01

The math is public.

Every chart on this page is the working method, not marketing art.

02

The results are ledgered.

Expected vs. measured, recorded even when we're wrong.

03

The incentives are flat.

No % of spend — our advice never inflates our own invoice.

04

The exit is open.

Month-to-month, free export. We re-earn the fee every 30 days.

FAQ

Zero questions left.

Do you need access to my Seller Central or Shopify admin?

One permissions-scoped seat on whichever platform you run — that's the whole setup, and it powers everything. On Amazon: Reports (view) lets the models read your numbers; pricing and Campaign Manager (view & edit) let us execute the corrections. On Shopify: a collaborator account with Orders, Analytics, Reports and Marketing to view, and Products and Discounts to edit, nothing under Settings or Payments. Banking and account settings stay untouchable on both, and you revoke the seat in one click, any day. Prefer not to grant it? You can send a handful of CSV exports through a secure upload page instead.

What data do I need to send?

About fifteen minutes of exports, once. On Amazon: the Business Report, SKU Economics, your advertising reports and an FBA inventory snapshot. On Shopify: orders, products with cost per item, payouts, and your Meta or Google ads exports. Your private upload page lists each one with the exact clicks to get it. The one thing no platform knows is your costs: a one-row-per-SKU template (unit cost, freight, fulfilment, lead times), estimates welcome. After that the seat does the rest, and we never ask you for homework again.

The Teardown is free — what's the catch?

So is your first month — and the catch is honest: if we deliver, we ask for a short testimonial and permission to publish your results with the numbers anonymized. We're building the track record in public, and your account is how we prove it. You also get a link of your own: a founder you send gets the same free month, and when they stay past their day 30, your next month is on us. No obligation, and nothing renews against your will.

How much of my time does this take?

One 20-minute application call, two minutes granting the seat, five minutes on a cost template, then roughly three minutes a cycle watching your brief. The modeling, the measurement, and the execution all happen on our side — you're buying the abdication of the mental load, not another task.

How long until I see value?

Within 24 hours of your seat going live — the models run the moment we can read your account. Your first corrections are executed inside the free month, and price tests read out in days because Buy Box share, or conversion rate on Shopify, is tracked daily.

What are the contract terms?

Your first month is free. After that: $6,000/month flat, month-to-month, cancel anytime. Invoiced by email via ACH, net-7 — no card on file, so nothing can renew against your will. Never a percentage of your ad spend. Your data and your ledger export free, anytime.

What if you find nothing worth fixing?

We tell you exactly that, and you keep the full report. A Teardown that honestly clears you is a good outcome for you and cheap marketing for us.

Why not just use Sellerboard, Helium 10, or Triple Whale?

Keep them — they're good mirrors. But they report what happened. None runs probabilistic inventory math, none derives what your target ACoS or ROAS should be, and the average $1M+ brand already runs 5.4 of them. We're the layer that turns their numbers into decisions, then measures what each decision earned.

Why not a PPC agency or a fractional CFO?

Agencies bill a percentage of your ad spend — an incentive problem — and average roughly 49% client churn a year. Fractional CFOs run $3,000–$12,000/month at P&L altitude without SKU-level modeling. We sit in the gap: SKU-level math, measured outcomes, flat fee.

Who are you?

A founder-operated quantitative practice. The person who builds your models is the one on your calls — no outsourcing, no account managers. The method is on this page; the Teardown is how you evaluate us.

Get your free Profit Teardown