/ Growth infrastructure for EU DTC brands
Measurement you can steer by.
Your ad account reports one number. Your bank reports another. Every budget decision you make runs on the first one. I find the gap, in five working days, with the evidence printed next to every finding.
Five questions, about two minutes. You find out which of the three tiers fits your size, then you buy the one you want. 149, 700 or 1,900 EUR, excluding VAT, every one of them credited in full against the first project.
Who this is for: Shopify DTC brands, 1M to 20M EUR a year or 30k+ EUR a month in ad spend, a physical product of your own, selling into at least one EU market. EU and UK. Founder, CEO, CMO or head of growth, because the answers only mean something from someone who can act on them.
- Price
- 149 to 1,900 EURThree tiers, excluding VAT. Whatever you pay is credited in full against the first project you sign within 30 days.
- Turnaround
- Minutes, or 5 daysSignal is automatic and arrives in minutes. Audit and Roadmap start the clock when read-only access is granted, not when you pay.
- Availability
- Scope
- 6 areas, 33 checksOne scorecard, one impact estimate per area, one recommended next step.
How it runs
- 01Qualify, two minutes. Five questions: site, platform, spend, markets, and what is bothering you. No call to book first.
- 02Pick a tier. You see the ones that fit your size, with what each can and cannot see stated plainly.
- 03Pay. Card, EUR, VAT worked out at the checkout before you enter it. Nothing is charged before the total is on screen.
- 04Minutes, or five working days. Signal is automatic. Audit and Roadmap send the access guide and the questionnaire the same day, then the report and a 45 minute call.
01 The situation
Ads Manager says one thing. The bank says another.
Scale this, kill that, enter that market. All of it downstream of a number nobody has ever audited.
The gap is rarely one dramatic bug. It is usually five quiet ones stacked:
- 01Purchase events that never reach Meta at all.
- 02Events that reach it twice, because the pixel and the Conversions API both fire and nothing ties them together.
- 03A consent banner that silently drops part of your signal, in exchange for a compliance posture nobody checked.
- 04Non-EUR orders reporting the wrong value: 18,900 HUF read as 18,900 of your account currency.
- 05An attribution window that flatters everything and is never compared against blended.
Each one is small. Together they decide where your money goes. And none of them announce themselves: the numbers keep arriving on schedule, they are just wrong.
Ads Manager, last 30 days
Shopify, same window
Those fields stay empty on purpose. I have no clients yet, so I have no numbers to put in them that would mean anything to you. The audit fills them in with yours.
02 What the audit actually is
Six systems, 33 checks, one scorecard.
A forensic review of the six systems that decide how much you keep from every euro you spend. Every check carries its evidence: a table, a query or a screenshot. A check that cannot be evidenced is reported as not verified and does not enter the score. It is never estimated.
| Area and the question it answers | Example checks | Checks | Output |
|---|---|---|---|
| Measurement Is the number you steer by real? |
|
7 | Score, evidence, impact in EUR |
| Ads efficiency Is the spend structured to learn? |
|
6 | Score, evidence, impact in EUR |
| Cross-border Which markets are already asking and cannot buy? |
|
6 | Score, evidence, impact in EUR |
| Checkout What happens after the click you paid for? |
|
6 | Score, evidence, impact in EUR |
| Creative Are you testing enough to find a winner? |
|
4 | Score, evidence, throughput gap |
| Operations How much of the business runs on a person remembering? |
|
4 | Score, hours and risk |
Creative and operations carry no euro estimate. A number there would be guesswork, and guesswork in a report about measurement would be an odd way to start.
03 The deliverable
This is the format you get back.
One page, six areas, a score out of 100, a state, and the formula behind the euro figure. The formula is printed in the report next to the number it produced, with its inputs, so you can argue with the assumption instead of trusting the output. The score and state columns below are blank because the numbers that go there are yours.
How the state is decided
- Fix now Score under 55. Money is leaving now, and the fix is worth doing before anything else on the list.
- Fix later Score 55 to 79 with a euro figure still attached. Real, and not urgent.
- Fine Score 80 or above, or nothing left to recover. Leave it alone and spend the attention elsewhere.
- Not verified No score. The check could not be evidenced, so it is reported as unverified and kept out of the score rather than estimated.
Those four bands are the rule the report runs on, not a description of anyone's account.
Recommended next step The area with the largest impact x (100 - score), chosen only from the work we actually sell. Everything else is reported as an observation, with no price attached to it.
Read this bit. The score and state columns are blank on purpose. Filling them with example numbers would put the first invented figure on a page whose whole argument is that we do not invent figures, and you would have no way to tell which other number on the site was made up. So: no sample scores, no average, no benchmark. What is real above is the layout, the six areas, the state rule and the formulas. Your numbers are computed from your own accounts, and every line of the finished scorecard points at the evidence behind it in the appendix.
04 Five working days
What you get, and what I need from you.
What lands on day five
- 01The scorecard. 0 to 100 per area, with the checks behind each score and the evidence behind each check.
- 0233 checks, each marked verified or not verified. Anything we could not evidence is labelled as such rather than estimated.
- 03An impact estimate in euros per area, with the formula and the inputs printed next to it. Conservative, and a range, never a promise.
- 04One recommended next step. Chosen by impact, not by what is most expensive to buy.
- 05The list of findings your own developer can close in a day. Listed separately, with the fix, because we would rather lose that work than bill for it.
- 06A PDF, 10 to 14 pages. Yours. Share it, act on it with anyone, including a competing agency.
- 07A 45 minute presentation call with the person who ran the audit. There is no account manager in between.
- 08A baseline dataset. If we work together afterwards, this is what the work gets measured against.
| Day | What happens |
|---|---|
| Day 0 | Call, then payment. Questionnaire and access guides sent automatically. |
| Day 1 | Access confirmed. Automated collection runs. First pass scorecard generated. |
| Day 2 and 3 | Manual review. Every finding checked by hand before it enters the report. |
| Day 4 | Report finalised, impact estimates validated against their inputs, recommendation decided. |
| Day 5 | PDF delivered, then the 45 minute call. |
What I need from you
| System | Access | What it is for |
|---|---|---|
| Meta Business Manager | Partner access, Analyst role on the ad accounts and Events Manager | Spend, structure, Event Match Quality, deduplication |
| Shopify | Collaborator: Orders, Products, Analytics, Settings (view) | Orders by market and currency, checkout, markets configuration |
| GA4 (optional) | Viewer | Sessions by country, which is what the dormant demand check runs on |
| Questionnaire | 12 questions, about 15 minutes of your time | Operations, goals, constraints, and the things no API reports |
- •Read-only, always. An audit never needs write access. If a project starts later, write access is scoped as narrowly as the platform allows and revoked at handover.
- •The clock starts at access, not at payment. Step by step access guides are sent with the questionnaire.
- •If access does not arrive within 10 working days, the audit is delivered from public data and the questionnaire, with every access-dependent check marked not verified, plus one free re-run within 60 days once access arrives.
- •Mutual NDA on request, signed before access. That is standard here, not a favour.
05 Disqualification
Who this is wrong for.
Written down in public so you can rule yourself out before paying, rather than finding out from a report. If you are one of these, the qualification step will say so in two minutes, and it will say it for free.
-
Under 30k EUR a month in ad spend
At that level the measurement work costs more than the misallocation it recovers. The cross-border and checkout findings still stand on their own, so the audit can be worth 700 EUR to you, but I will tell you on the call that the package after it probably is not.
-
No product of your own
Dropshipping or generic reselling. This work assumes a margin structure and a brand that can carry a market entry. Neither holds, and the honest answer is that better tracking will not change the economics.
-
Your home market is not working yet
Expansion multiplies whatever you already are. If home economics are marginal, four more countries produce four more marginal businesses and a much harder operation to run. Fix the first one, then come back.
-
You want a guaranteed ROAS
I guarantee two things on the measurement package: Event Match Quality at 7 or above, and purchase variance between Meta and Shopify under 5% at 30 days. Nothing about your revenue. Anyone guaranteeing a ROAS is either lying or has not been doing this long.
-
You want ads managed while tracking is broken
No. Running spend on false data is how an agency produces excellent reports and a dead business. Measurement first, or I decline the ads work. That is not a sales ladder, it is the order the work has to happen in.
-
You cannot grant read-only access
Then the audit would be guesswork with an invoice attached. If policy genuinely blocks it, buy Signal instead: it reads only what is public, it costs 149 EUR, and it is honest about the six questions it cannot answer from out there. Write to me before buying Audit, not after.
06 Who runs this
An operator, not an agency.
Photograph pending: real working photo, shot on a camera. Not stock, not generated.
I am George Carpusor. I co-founded Carpathian Jerky in 2023 and I still run it. We sell a physical food product across 14 EU markets.
Server-side tracking, per-market storefronts and currencies, local payment methods and carriers, a custom checkout, and the operational agents that handle orders, returns and customer support. Built in-house, by a small team working with AI agents, because nobody sold it as one thing.
Food is the hard version of this: shelf life, customs, labelling, per-country compliance. Everything I know here, I learned by getting it wrong on my own money first, on a business that has to pay people at the end of the month.
So when the report tells you your Purchase events are not deduplicated, or that a country has been sending you traffic for a year that cannot check out in its own currency, it is not a finding from a methodology slide. It is a thing I have had to fix on my own account, under time pressure, with revenue running through it.
I run the audits myself, and I am on the presentation call. Capacity is 10 to 15 audits a month, because every finding is reviewed by hand before it goes in the report.
Verifiable: Carpathian Jerky, carpathianjerky.ro.
A real shop with real products. Order something if you want to test the checkout.
No case studies: this company is new. I have no client results to show you and
I will not invent any. Carpathian Jerky is the only proof I have, and you can inspect
all of it yourself.
07 Questions
The ones that decide it.
01Why is the audit paid?
Because a free audit is a sales call wearing a costume. You know it, so you discount everything in it, and the work gets done at the depth a free thing deserves. Paid means five days of real work and a document you can hand to someone else.
It also does the qualifying in both directions. A brand doing 3M a year that will not spend 700 EUR to find out where the money leaks is not going to sign a 15k project either, and we would both find that out slowly and expensively.
That argument has a floor under it now. Signal is 149 EUR and it runs the same engine on everything visible from outside your store, so nobody has to be sold on the idea before they can see the work. The free check at /check shows you the score and the coverage before you pay anything at all.
02I already have an agency. Does this step on them?
The report is yours. Hand it to them. Most findings are implementation work that somebody has to do, and nothing in the report requires that somebody to be me.
A good agency will be glad that someone reconciled the numbers they are judged on. If yours objects to being measured, that is information too. What I will not do is use the presentation call to pitch against them.
03Do you resell or share my data?
No. You stay the data controller, I am a processor, and a DPA is signed before access. Collection pulls aggregates and samples, stored on EU infrastructure, deleted 12 months after delivery or the day you ask, whichever comes first.
Findings feed public writing only in anonymised, aggregated form. Anything specific to you needs your written permission, and access is read-only and revocable at any moment.
04What happens after the audit?
A 45 minute call, then your decision. Do nothing, hand it to your developer, hand it to your agency, or work with me. If you sign a package within 30 days, the 700 EUR comes off it in full.
Three of the findings are the same piece of work every time, so they are priced fixed rather than quoted: multi-currency purchase value fix, 1,500 EUR. CAPI deduplication and Event Match Quality fix, 2,000 EUR. Consent Mode v2 without losing signal, 1,500 EUR. The larger packages are FIX (the full measurement stack plus a blended MER dashboard, 4,000 to 8,000 EUR setup, plus monitoring at 1,000 to 1,500 EUR a month) and EXPAND (launching 2 to 5 new EU markets properly, 15,000 to 30,000 EUR a project).
05Can my own developer implement the findings?
Yes, and the report is written for that. Every finding names what has to change and where, and there is a section that lists which findings are a one-day in-house fix. Those are usually free money and you should take them whoever you hire.
There is no lock-in and no obligation. If your developer closes everything in the report, the audit paid for itself and I would rather you tell people that than keep you on a retainer you do not need.
06What if you find nothing wrong, and what about refunds?
Then you get a document saying your measurement is sound, with the evidence, and you can stop wondering. That is a legitimate result and it is the deliverable.
Refunds follow the work, not the calendar. The question is never how many days have passed, it is whether I have done the thing you paid for, so the three tiers get three different answers:
- Signal. No refund once the report is generated, because generation is delivery and it takes minutes. Two automatic exceptions, refunded without you asking: the engine could not reach your storefront, or coverage came out under 60 percent of the checks, which means I answered too little to charge for it. You keep the partial report either way.
- Audit. Full refund before you grant access, because nothing has run yet. Half after access but before manual review starts, and you keep what the collection produced. None after review begins, which is 09:00 CET on the second working day after access. That boundary is a fixed clock moment on purpose, so neither of us can argue about it later.
- Audit and Roadmap. The same, plus one more: if I cannot schedule the working session within 15 working days of access because of my calendar, you take the Audit instead and I refund the difference without being asked.
If I miss a deadline or cancel for reasons on my side, full refund at any stage, no pro rata and no stage table.
07Shopify and Meta only?
That is the base: Shopify and Meta, with GA4 optional. WooCommerce works too. If most of your spend sits on Google, TikTok or Pinterest, say so on the call: the measurement checks transfer, the ads efficiency checks are Meta specific today and I would rather tell you that now than pad the scope.
08How does it start, and do I have to get on a call first?
No. You answer five questions, about two minutes: the site, the platform, monthly spend, the countries you sell into, and what is actually bothering you. Two minutes later you know which of the three tiers is right for your size, and you buy the one you want.
The questions are not a hurdle, they decide what you are shown. Above roughly 30k a month in ad spend or 1M a year in revenue, all three tiers are open to you. Below that, you see Signal and the threshold that would change the answer, because the deeper audit would cost you more than the misallocation it could find. Nobody is turned away and nobody is sold something that will not pay for itself.
A call still happens on Audit and Roadmap. It happens after the work, to present the findings, which is the only point at which I have anything worth your time to say.
/ Next step
Find out which number is the wrong one.
Six areas, 33 checks, five working days, evidence behind every line, and a scorecard you can put in front of your board.
Two minutes. Then you see your tiers and buy the one you want.
149, 700 or 1,900 EUR, excluding VAT, paid up front, every one of them credited
in full against the first project signed within 30 days.
Signal in minutes. Audit and Roadmap: 5 working days from access granted, not
from payment. Read-only access, revocable, on your own accounts.
Mutual NDA on request.
No performance guarantees, here or anywhere else on this site.