E-commerceOperating model
A product nobody had ever bought, launched into three countries on $10,000 total, sold out in three months.
Artopad had no sales history anywhere. No audience data, no pixel history, no idea what a customer should cost. The budget had to cover the fee, the media, and everything else.
~800orders in three months
~2,000units sold
Read the full case ↗E-commerceOperating model
Average order value went from $240 to $760 after we stopped working on the ads.
A US luxury furniture retailer trading as VIG Furniture asked for a marketing fix. The audit covered sales, marketing, logistics, and IT together, and the margin turned out to be leaking somewhere nobody was looking.
$240 → $760average order value
4% → 18%customer retention
Read the full case ↗B2B marketplaceOperating model
The ad with the best click rate was the worst ad in the account.
A US marketplace selling ready jobs to HVAC and appliance technicians needed customers. The research found the audience already paying $180 to $200 a lead to a company they openly warn each other about.
127technician leads
$4.70lowest regional cost per lead
Read the full case ↗Non-profitOperating model
Reached 12 million people and got aid to 17,850, starting with almost no budget.
In the first weeks of the 2022 invasion, a humanitarian platform had to connect people who needed help with people offering it, fast enough to matter, and trusted on sight by both sides.
12Mpeople reached
17,850individuals received aid
Read the full case ↗Health technologyAdvisory
The hours nobody could get a psychologist were the hours everybody needed one.
An AI that stands in for a psychologist, 24/7, at a ninth of the price. The product worked. Nobody could see where users were dropping out, and the media buying was aimed at the wrong variable.
−70%customer acquisition cost
$0.03cost per click on the primary channel
Read the full case ↗AgricultureDiagnostic
Two hundred pages of analysis that told the client not to enter the US market.
A Turkish agricultural group wanted to know whether the United States was worth entering. After two months on the product line, the market, the paperwork, and the money, the answer was no.
~$450Kfirst-year entry cost avoided
200page go-to-market strategy
Read the full case ↗Public affairsOperating model
Seven candidates, six elected, on a channel most of their opponents weren’t using.
The 2020 Ukrainian local elections. Field work still decides these races, but digital turned out to be the part you can actually predict and control.
6 of 7candidates elected
20–25%of the result attributable to digital
Read the full case ↗E-commerceOperating model
Paid marketing reached profitable ROAS for the first time that year, six weeks after we stopped optimising the ads.
A national jewellery chain had a serious digital budget and no return on it. The funnel had been built for generic e-commerce, and nobody buys jewellery the way they buy a phone case.
ProfitableROAS, first time that year
−25%cost per sale
Read the full case ↗AerospaceFull-time employment
A factory producing sixteen aircraft a year at 64% of its twenty-five-aircraft capacity.
A Czech manufacturer selling into fourteen countries, running four unrelated business lines, and using at least three different names for the same aircraft. It started with a European grant and ended with a product line that didn’t exist when I arrived.
64%factory capacity utilisation
14countries in the dealer network
Read the full case ↗Aerospace and defenceOperating model
Reorganised a dual-use drone maker around buyer situations, and deal flow went up 30% in under two months.
Serious technology spread across too many use cases, civilian, industrial, military, flagship, custom. Everything mattered, which meant nothing was easy to buy.
+30%qualified deal flow
3buyer verticals defined
Read the full case ↗Technology distribution & wholesale retailOperating model
Cut project lead time by 40% at a major Ukrainian technology distributor — by making the official path the fast one.
ERC (erc.ua) is a major Ukrainian technology distributor and wholesale retailer. Across retail, wholesale, e-commerce, and systems-integration channels, ideas had no clean route to a named owner, a test, or a decision.
−40%project lead time
+12%average cart size
Read the full case ↗Public sectorDiagnostic
Turned scattered wartime data into a working list of priorities for credit and recovery.
Ukraine’s small and medium businesses were losing people, capacity, and suppliers. The National Bank needed analysis it could act on, built from sources that were late, incomplete, and politically loaded.
−33%MSME workforce decline
−50%capacity reduction
Read the full case ↗AerospaceOperating model
Turned an aircraft dealer into a full-lifecycle provider and won three partnerships on a ten-year horizon.
Heavy operational load, thin returns. We opened the books to see where the money actually came from, and the answer changed the product.
3global partnerships
$1M+annual turnover potential per partnership
Read the full case ↗AerospaceOperating model
A factory that could build 150 aircraft a year was calling itself a shop for parts.
Aviatex was built for export from day one, with exclusive rights to two Czech manufacturers and its own production line. It was selling all of that as parts on request.
~$2.8Msupply pipeline made forecastable
6 moforward demand visibility
Read the full case ↗TravelOperating model
Built an intercity chauffeur business from the logo up, buying clients worth years of repeat bookings for $46.
No brand, no site, no way to get a lead. The economics worked because we weren’t buying one ride: we were buying a two-to-three year relationship.
234booking leads
$45.97cost per lead
Read the full case ↗Marketing servicesOperating model
132 leads at $19.85 — and, more usefully, a process that outlived the campaign.
The agency could deliver work in any state and had no repeatable way to attract a client or onboard one. This ended up being more consulting than campaign management.
132client leads
$19.85cost per lead
Read the full case ↗Health and wellnessGrowth
The cheapest leads in the account were the worst ones.
A new franchise studio in Prague, opened by a Russian-speaking owner who needed Czech customers and said plainly he had no idea how to reach them. The advertising worked. What it sold was a service nobody in the market recognised.
384leads at about $3.30 each
~$2.23best cost per lead, and the worst leads
Read the full case ↗Legal servicesOperating model
Built a caseload big enough that some of those matters are still in court.
A US debt resolution firm was spending on Google Ads and LinkedIn with no strategy and no audience analysis. Rebuilding around one channel produced volume that needed a 24/7 call centre to absorb.
3,000+leads per month
<$1cost per lead
Read the full case ↗HR technologyAdvisory
Stopped speaking HR-tech, started speaking founder, and demo bookings went up 52% in four weeks.
An HR-tech platform matching businesses with pre-vetted virtual assistants had a good product and a page describing the platform instead of the problem.
+52%demo bookings
−38%landing page bounce rate
Read the full case ↗AutomotiveContract
An NPS of 70, sixteen years of auction licences, and no way for anyone outside the client list to know.
AutoAmerica buys cars at US auctions for Ukrainian buyers. The customers were loyal and the reputation was real. None of it existed anywhere a new buyer could find it.
70NPS at the time of the audit
16 yrsof auction licences never shown publicly
Read the full case ↗EducationOperating model
A marketing course with no theory in it — six modules where every framework is something you can go and do.
Ten years of practice existed as habit rather than method. Turning it into a teachable framework meant breaking every part of it into steps somebody else could follow without asking a follow-up question.
6modules built
15+named frameworks
Read the full case ↗FinanceOperating model
Fixed a junk-lead problem by making people prove they meant it.
A California mortgage business was generating plenty of enquiries and burning its team on the ones that went nowhere. One confirmation step raised the share of usable leads by about half.
~50%increase in quality leads
156leads
Read the full case ↗Sports tradingOperating model
He knew tennis for thirty years and had no way to act on it. Now a script does.
A client who had spent his working life around tennis wanted to bet on it seriously and had no way to turn what he knew into decisions. We got the knowledge out of his head, put thirty years of results behind it, tested around three million combinations, and shipped a system that wins close to four bets in five.
30 yearsof match data behind the model
~3Mcombinations tested
Read the full case ↗HealthcareOperating model
Filled a hospice to capacity on $1,000 a month, in the hardest category on the platform to advertise in.
The goal was two admissions. About twelve arrived. Most of the work was finding a way to speak honestly to families that Meta would actually approve.
~12new patients
6×the stated objective
Read the full case ↗Real estateOperating model
Strong traffic was hiding a weak business model, right up until developers started renewing.
A real estate listings platform had visitors it couldn’t turn into leads its developer clients would pay for. Rebuilding the funnel around buying urgency lifted paid conversion 2.5× and renewals 40% quarter over quarter.
2.5×paid traffic to lead conversion
+40%B2B renewals, quarter over quarter
Read the full case ↗FinanceOperating model
Communication work that moved a credit risk number: defaults among new borrowers fell 12%.
A rural microfinance programme run under the National Bank of Ukraine read low engagement as low interest. The borrower interviews said otherwise: people weren’t uninterested, they were scared, and nobody had explained it like a human.
+28%loan programme enrolment
−12%default rate, new borrowers
Read the full case ↗TechnologyProduct and go-to-market
Forty owners said yes before a line of code was written, and eight committed to use it.
A SaaS dispatch tool for small US trucking companies, taken from idea to a validated, scoped MVP. The work was the validation and the commercial model, not the software.
40owner interviews before development
8companies committed to use it at launch
Read the full case ↗TechnologyAdvisory
An NFT you could stand in the street and film, shipped to the App Store instead of a deck.
Influence’AR was an augmented-reality app built around influencer-linked NFTs. Buyers could place the objects in the real world, create content with them, and later resell them.
3named alternatives to displace
3customer segments defined
Read the full case ↗ManufacturingOperating model
Quotes went from two days to four hours, and the guarantee went on the website.
A packaging manufacturer producing runs from 10 units. The buyer isn’t comparing suppliers, he’s deciding whether the category is worth any money at all.
2 days → 4 hrsquote response time
~12×faster time to quote
Read the full case ↗EducationOperating model
The most valuable conversation was the one everyone else was charging for. We gave it away and sold what came after.
Student CZ sells Czech tuition and university preparation to families relocating to the Czech Republic. The product was good. There was no path from a worried parent to a paying student.
3segments researched, two funnels built
~$24tripwire price point
Read the full case ↗Public affairsOperating model
Political consulting was negotiated from scratch every time. Now it starts at about $6.70.
Political consulting is usually sold as a bespoke engagement negotiated from scratch every time. This packaged it into four tiers, with field software underneath instead of slides.
4product tiers, laddered
~$6.70frontend event price
Read the full case ↗BiotechnologyAdvisory
Made veterinary cryopreservation explicable to the three different people who would buy it.
A biotech startup founded by veterinary scientists needed a digital presence that carried real science without reading like a lab report.
3distinct audiences served by one site
Read the full case ↗AerospaceOperating model
Built an aviation community that outlived my involvement in it.
Not a client project: a venture I started with a partner to get aviation people into the same room. It reached self-sustaining, paid for its own events, and kept going after I sold my share.
Key resultAn industry network that covered its costs and survived a founder leaving, which is the harder half of building one.
Read the full case ↗TravelOperating model
Built a national travel agency’s marketing department from scratch, and made 60% of leads traceable for the first time.
It arrived as a campaign problem. It was an infrastructure problem. Tools didn’t talk, leads couldn’t be followed, and nobody could see what was working.
+35%online bookings
−18%customer acquisition cost
Read the full case ↗Non-profitDiagnostic
Made a grant portal readable for first-time applicants and lifted their conversion 22%.
A cultural grant programme was reaching people who then didn’t apply. Eligibility was never the barrier. The process just felt hostile to anyone who hadn’t done it before.
+22%first-time applicant conversion
−40%information search time
Read the full case ↗Consumer electronicsDiagnostic
Four official pages, 430,516 followers, and no agreement on what any of them were for.
A three-month audit of Samsung Ukraine’s social channels. The finding was not that engagement was low. It was that each channel was behaving like a different company.
430,516followers audited across four channels
321publications analysed
Read the full case ↗AerospaceOperating model
Cut the scope in half, which is why the project sold.
Strong engineering, and a story that kept drifting across SaaS, service contracts, defence, logistics, grants, and investors. Narrowing it to one path put it on two grant shortlists and made it saleable.
Key resultConvinced a very capable team to stop building half of what it was building, then took the rest to a completed sale.
Read the full case ↗IT servicesOperating model
A company building digital products for banks had a website from the 2000s.
CITAUA had real delivery experience across finance, banking, and travel, and no way to prove any of it. We rebuilt what the company was: product line, positioning, brand, and where the engineering hours went.
+60%website traffic
2clients propping up a legacy product line
Read the full case ↗Aerospace and defenceOperating model
Ten product lines into NATO STANAG compliance, and a deal built on technology transfer rather than a shipment.
An established Czech defence and aerospace group needed more than tidier materials. The work was business development, compliance, and getting a technology transfer over the line.
10product lines brought into NATO STANAG compliance
Read the full case ↗Non-profitOperating model
Gave scattered volunteer teams enough structure to work on their own without tripping over each other.
Dozens of separate channels connected volunteer groups to regional hubs. We replaced improvised coordination with light SOPs people could use without asking permission every five minutes.
Key resultEmergency coordination that let local teams act semi-independently while staying aligned, less duplicated work, and resource use you could actually trace.
Read the full case ↗