Anchor Media · CreateX · ShellyX
Opening the markets is not the hard part and neither is translation. The constraint is which half of the operation we own, in what order, and which single missing function will quietly eat the conversion rate. This is the plan for that.
Start here
Before staffing anything, it is worth being precise about what is genuinely missing, because the answer turned out to be much less than the list suggests.
Shelly's store already runs in five locales, English, German, French, Italian and Polish. It bills in a single currency, euro, everywhere. It has live support, contact, returns and terms pages. Buyer reviews are already arriving in German, French, Italian and Spanish, which means real customers are already being served in those languages today.
So checkout, fulfilment, warranty, returns and post-sale support exist across Europe right now. Single currency plus the EU one-stop-shop scheme means the tax position is one return rather than twenty-seven. None of that is ours to build, and none of it is a reason to delay.
The gap
This catalogue is unusual. A meaningful share of buyers cannot qualify themselves, and the questions arrive before the sale rather than after it.
The energy advertorial currently tells readers to photograph their consumer unit and send it to the manufacturer's support before ordering. That is a deliberate line and it is the honest advice, but it means the page is routing conversion work into a service queue. A support team is measured on tickets closed and response time, not on orders recovered. Every hour that question waits is an order cooling. It will show up in the numbers as a support cost, and it will behave like a leak.
The fix is not a support team. It is a qualifier that sits on the advertorial and the bundle page, answers the four questions above from a photo and two taps, recommends the right kit, and hands anything genuinely ambiguous to a human. It is bounded, high volume and rule-based, which is precisely the kind of thing worth building rather than staffing.
It absorbs the support spike before it reaches anybody's inbox. It captures the email address before the sale rather than after it, which is the only mechanism by which we end up with a list at all if checkout stays on Shelly's store. And it records every question asked, which is the offer and angle backlog writing itself from real objections instead of from our guesses.
Sequence
Phase A can start this week and depends on nothing but a discount code. Phase B is the version worth having, and it is gated on a single commercial term rather than on any amount of work.
Gated on: the HOSTKIT and ENERGYAUDIT discount codes existing on the store. Nothing else.
We are pure demand generation. Traffic lands on the advertorial, the cart link loads the kit, Shelly's checkout and Shelly's people handle everything after that. We add zero headcount and carry zero inventory, VAT or returns risk.
What we learn: which angle sells, at what cost, and what people ask before buying. That last one is the input to Phase B, so Phase A is not a placeholder. It is the research.
What we accept: no list, no email, no SMS, no lifetime value. Every order is a one-off and the customer is Shelly's. That is a real cost and it is the reason Phase A is not the destination.
Gated on: one agreement with Shelly on who owns the marketing list. Not on engineering.
The qualifier goes live in front of the checkout and captures the address. We own pre-sale, the list, email, messaging, offers and testing. Shelly keeps checkout, fulfilment, warranty, returns and the books.
What changes commercially: this is where repeat purchase and the second-property kit actually become reachable, and it is the only route by which Shelly Cloud Premium can be sold to people we acquired. Without it the ladder in the strategy brief is decorative.
Ownership
| Function | Phase A | Phase B | What is actually required |
|---|---|---|---|
| Pre-sale qualification | Shelly support | Build it | Software, not a team. Escalation to a human only for the ambiguous cases, and only in German and French to begin with. |
| Post-sale support | Shelly | Shelly | Nothing. It exists, in five locales, and taking it on would make us a reseller. |
| Warranty and returns | Shelly | Shelly | Nothing. Do not touch this. |
| Accounting | Shelly | Shelly | Single currency and one-stop-shop already solve it. Not a staffing question. |
| Margin reporting | Us | Us | Contribution margin per market per offer. Blocked on the per-SKU margin figure, which is still outstanding. |
| Not possible | Us | Needs list ownership agreed first. Until then there is nothing to send to. | |
| SMS and messaging | Not possible | Us | Per-market channel choice, not one channel. See below. |
| Offers and creative | Us | Us | Already ours. Needs a cadence and a kill rule more than it needs people. |
| The weekly decision | Us | Us | A named owner, not a hire. Somebody has to be accountable for what gets killed. |
Phase A needs zero new people. Traffic goes to an operation that already runs. Phase B needs part of one person, handling escalations in German and French, the two largest of the five locales, plus the qualifier itself. That is not a support department, and anybody proposing one for this volume is solving a problem we do not have yet.
Rollout
Localising five pages before knowing whether the angle works means paying to learn the same lesson five times. The rule is that nothing gets translated until it has earned it.
An angle is ready to localise when it has beaten its target cost per acquisition over at least twenty-five to thirty purchases, not before. Below that you are reading noise and promoting luck. The second gate is that the click-to-purchase rate holds, not just the click-through rate, because a cheap click into a page that does not convert is the most expensive thing in this business.
The whole economic argument in the energy advertorial is priced in electricity. Household electricity runs about 40 cents a kilowatt hour in Ireland and about 11 in Hungary, with Germany near the top and Bulgaria near the bottom. The same worked example is three to four times more persuasive in one market than another, and in the cheapest markets the payback story stops being true at all.
Heating type moves it again. The water heater angle only works where water is heated by electricity, which is common across Southern and Eastern Europe and much rarer in Germany, where gas dominates. Every number in the page has to be recomputed per country, and one of the two lead angles has to be swapped in some of them. Send a translator at this and you will ship a confident article that is locally wrong.
The structure, the mechanism, the reframe and the offer ladder. The argument that you are billed a total you cannot decompose is true in every market in Europe.
Every euro figure, the tariff assumptions, the payback claim, the choice of lead angle, and the short-term rental rules, which differ sharply by city and are restrictive in several.
Native-language buyer reviews are already on the store in German, French, Italian and Spanish. Localised pages should carry real reviews in the reader's own language, and no new proof needs gathering.
Cadence
You are right that some offers will work and some will not, and the only thing that separates a testing programme from expensive thrashing is deciding the rules before the results arrive.
Read, kill, promote. Anything below target after its allotted budget is stopped, without a conversation about whether it deserves another week.
One new angle from the backlog per funnel. There are ten in the strategy brief plus your existing approved library, so this is a scheduling problem, not a creative one.
Every offer gets a defined test budget, roughly three times target acquisition cost, before judgement. Gut calls at half that spend are how good angles get buried.
Change the offer or the angle, never both, and never in two markets at once. Otherwise nothing is attributable and the whole programme becomes opinion.
A loser in Germany is not a loser in Poland. Electricity price, heating type, rental regulation and average order value all move by market, so an offer is killed per market, not globally, and a winner is promoted the same way. Rolling a German result out across five countries is the fastest way to lose four of them.
Critical path
Only the first item blocks launching this week. The rest block scale, in the order listed.
HOSTKIT and ENERGYAUDIT. Without them the live cart links load both kits at full price and the offer in the copy is a lie. This is the only thing standing between the current pages and paid traffic.
Every acquisition cost target in the strategy brief runs on a 45 percent placeholder. Until this is real we can launch, but we cannot tell you what a good result looks like, which means we cannot tell you when to spend more.
One term. It decides whether email, messaging, repeat purchase, the second-property kit and Cloud Premium are available to us at all. Everything else in Phase B is work we can simply do.
Starts the moment item three lands, and the question set is already known from the four objections above. Phase A traffic will refine it before a line is written.
The only subscription in the catalogue. Everything else in both kits is a one-off sale, so this is the whole of the recurring revenue question.
Decides whether order bumps sit in checkout or on the cart page, and therefore how much of the offer ladder is reachable at all.