01
Paying for customers you already had.
Branded search retargets people halfway through the checkout. Email takes credit for a purchase that would have happened organically. The dashboards count it as new growth. Most of it isn't.
The problem
Your business is more fragile than the dashboard suggests.
Most DTC brands die quietly, then suddenly. The dashboard stays green the week before the cash runs out. Three things are usually breaking long before anyone notices.
02
Channels going negative, silently.
By the time a channel's weekly ROAS turns red, you have been losing money on it for a month. Diminishing returns don't announce themselves. They show up in the bank.
03
Organic carrying ads. Or the other way around.
Your top organic channel sends someone to a retargeting ad and the ad takes credit. YouTube prospecting drives a search query a week later and Google takes credit. The math never adds up because no platform has the full picture.
What we actually build
A mathematical model of your business as a system.
Most DTC operations are copy-paste. Glossier did a loyalty program, so should we. That is not strategy. It is mimicry. We replace mimicry with the math of your business.
01
Your business, as a system
Funnel states, channel inputs, conversion flows, repeat patterns. The whole journey as one connected model — not five disconnected dashboards.
02
Priors that learn from your data
Every transition has a probability with uncertainty bounds. Each week the model ingests fresh data and the priors sharpen. The model becomes more yours over time.
03
Every decision goes through it first
Launch a loyalty program. Raise prices. Cut a channel. We tell you what the model predicts, then measure what actually happened. Then the model gets sharper.
Inside the monthly report
Two questions the model answers every month.
What did your marketing actually cause? And where should the next dollar go? The charts below are mock examples, in the same style your monthly report uses.
What did your marketing cause? Each bar splits a channel's attributed revenue into what would have happened anyway and what was truly caused by the spend. Google looks big, but most of it is branded search. TikTok looks small, but almost all of it is incremental.
Where should the next dollar go? Recommended spend, channel by channel, derived from where each one sits on its saturation curve. Total budget stays constant; the shift comes from the math, not a gut call.
Additional studies
Productized, one-time studies.
Each study answers a specific founder question. Each one has a clear before and a clear after. Available alone, or layered on top of the monthly retainer.
01One-time engagement
Cash & margin diagnostic
Know what you actually earn after COGS, returns, and payment fees.
Before
A platform tells you a 3.8x ROAS. Your accountant cites last year's margin. Nobody connects the two.
After
One reconciled P&L view: every dollar of revenue traced to true contribution per SKU, channel, and month.
02One-time, refreshable
Cohort & LTV study
Know if your growth is real customers or repeat business in disguise.
Before
You track new customers per month and total revenue. Repeat vs. new is fuzzy. Channel quality is a guess.
After
Customer-cohort curves by acquisition month and channel. Payback period, projected LTV, and the channels that pay you back fastest.
03One-time engagement
Pricing-elasticity study
Know if you can raise prices without killing volume.
Before
Pricing decisions made by gut. Promo calendars built on tradition. Margin left on the table or sales sacrificed to fear.
After
Estimated price elasticity per SKU with confidence bands, plus a recommended test plan to confirm before you commit at scale.
04One-time or monthly monitor
Market & competitor health
Know if the market is moving against you. Hyperfocus, or diversify?
Before
Competitor signals scattered across screenshots, social listening, gut feel. No one is correlating their moves with your sales.
After
A category-demand diagnostic with competitor pricing patterns, correlated to your weekly outcomes, and a recommended strategic posture.
Studies can be commissioned standalone or as add-ons to the monthly retainer. Custom-priced based on scope.
How it works
Three stages. One relationship.
1
Discovery
Free 30-min call- We learn your business, your stack, and the goals that matter (new customers per week, payback, LTV, gross margin, cash runway)
- You ask anything about the methodology, the deliverable, or the engagement shape
- We decide together whether the engagement is the right fit
2
Setup engagement
Roughly four weeks- Data integration: Shopify, Klaviyo, Meta Ads, Google Ads, bank statements, plus any platform that moves your numbers
- We build the first version of your business model: channels, funnel states, transitions, priors calibrated against your history
- First intelligence report and walkthrough call delivered at the end of the setup
3
Monthly retainer
Ongoing intelligence- Weekly data ingestion, monthly report, 45-min analyst call, quarterly calibration test
- Optional add-on studies layered in as the questions come up (pricing, cohorts, market health)
- Cancel any month. The model and methodology stay yours.
Why a person, not a platform
You have the dashboards. They are not the problem.
The bottleneck isn't data access. It's someone who can read the math, separate signal from noise, and turn it into a decision you can defend.
Attribution dashboards
Triple Whale, Northbeam, Polar
Their shape
Last-click pixel tracking, broken by iOS 14 privacy. Tells you which click happened before which purchase, not what your marketing actually caused.
Vantage
Bayesian causal models on aggregate data. We answer what would have happened without each channel, not just which one got the last click.
Self-serve measurement platforms
Admetrics, Blue Alpha
Their shape
150+ KPIs and a chat box. Breadth without judgment. You still have to decide what matters and what to do, and the tool can't help you choose.
Vantage
One model, one report, one analyst conversation per month. The judgment is part of the deliverable, not the founder's homework.
Hiring a senior analyst
Full-time data scientist
Their shape
$120K+ all-in per year. Twelve to eighteen months to ramp. Often only ever sees inside one business.
Vantage
Fractional quant rigor at a fraction of that. Methodology sharpened across multiple DTC engagements, applied to your business specifically.
Engagements
Custom-priced. Shaped to your business.
Every engagement is a one-time setup plus a monthly retainer. The depth, cadence, and add-ons scale with your media spend and the questions you need answered. We'll size it on the call.
Foundation
Earlier-stage DTC brands
First serious model of your business, with monthly intelligence after the setup.
- Setup engagement covers core funnel and primary channels
- Monthly six-page report plus 45-min analyst call
- Weekly data refresh, quarterly calibration check
- One add-on study included in the first quarter
Established
Multi-channel DTC operations
Wider model, deeper monthly work, room for the harder strategic questions.
- Setup engagement covers full funnel, secondary channels, retention loops
- Monthly report plus bi-weekly analyst calls
- Weekly data refresh, quarterly incrementality testing
- Two add-on studies included per quarter, plus ad-hoc scenario work
Add-on studies priced separately, one-time or monthly.
Who runs this
Built and delivered by Andrija (Andy) Radica.
Vantage Statistics is a solo practice based in Zagreb, working with DTC brands across Europe and North America. The work draws on probabilistic modeling, Bayesian inference, and the open-source tooling now used by enterprise data science teams, applied at the scale operators actually need.
Every engagement is run by me. No handoff to a junior, no offshored analysis. The call you book is the analyst you keep.
Questions
The questions before the call.
Bayesian Marketing Mix Modeling at the core, paired with customer-journey modeling (the state machine on this page) and periodic incrementality testing. We build on PyMC-Marketing, the same open-source Bayesian framework that enterprise teams at HelloFresh-scale brands use. The work was historically reserved for companies with internal data science teams. We deliver it boutique, for sub-enterprise DTC.
Those are last-click attribution dashboards. They tell you which click happened immediately before which purchase. We measure causality: what would have happened if you had not run that channel. Different problem, different math. Both can sit on the same desk; only one tells you what your marketing actually caused.
Those are platforms. You log in and look at dashboards. The judgment is your job. We deliver one model, one report, one analyst call per month. The judgment is the deliverable, not the homework.
If you have $120K a year of budget for it, sure. Most DTC brands at $1 to $5 million in revenue cannot justify a senior hire, and an in-house analyst only ever sees inside one business. We are fractional quant rigor, sharpened across multiple engagements, applied specifically to yours.
Twelve or more months of weekly spend by channel plus revenue is the sweet spot. We can work with less, but the model is sharper with more. If you are under twelve months in, we will tell you on the call whether the timing is right.
Yes. Part of the setup engagement is walking you through how your business is modeled, what the priors are, and where the uncertainty lives. You can ask questions of the model at any point during the engagement.
It is your data and your model. We hand over the artifact and the methodology at the end of any engagement, no friction.
Yes. NDA and data processing agreement signed before anything moves. All platform access is read-only. Storage is encrypted in transit and at rest, and data is destroyed at the end of the engagement on request.
Yes. We share an anonymized sample on the strategy call. We do not host it publicly because real reports name real channels, campaigns, and competitor positions.
Get started
Book a strategy call
Thirty minutes. Free. We figure out together whether Vantage is the right shape for what you need.
Andrija (Andy) Radica
Free Strategy Call
30 min
Video call
Google Meet
June 2026
SuMoTuWeThFrSa
Mon, Jun 8
Monday, June 8 at CET
You are booked
Monday, June 8 at CET
A confirmation has been sent to
Times shown in CET.