The German Lever Board

Each finding below is a claim about the German market that a photograph, a base rate, or both will support. Each one also names the levers it produces. The detail sits in the correspondingly numbered section.

Three things need to be true before a merchandising observation becomes a recommendation: it has to have been seen properly, it has to be counted against an honest denominator, and the judgement about whether it travels has to be written down as arithmetic rather than asserted as taste. This section covers all three, and states plainly where each one is weak.

The underlying practice is roughly a decade of walking convenience stores, quick-service restaurants and charging estates across Europe, Turkey and Israel — motorway corridors, city sites and countryside stops, mostly on long road trips rather than store tours. The measured subset analysed here is the photographic archive: 538 geotagged photographs and 22 video clips, 2017–2026, clustered into 51 cities and stops across 13 countries and segmented into 174 discrete store visits. Prevalence is computed only across the ten markets carrying enough visits to support a rate; the other three appear in the narrative but never in a base rate.

The distribution matters, because it shapes what the findings can see. The June 2026 trip alone is 226 frames — 42% of the whole archive — and it is overwhelmingly Greek, Bulgarian and Polish. Greece is the single richest unworked market in the set at 152 frames. Israel and several other markets sit in the lived practice but not in the photographic record, and are therefore not part of any count below.

Reading an archive this size one frame at a time is the expensive failure mode, and it is also the least accurate one — a store visit read as nine disconnected pictures loses the thing that makes it a scene. The method inverts that.

Reliability. A 54-photograph overlap between the main pass and an independent check gives Cohen's κ = 0.70 on mechanism codes and 0.75 on observability — substantial agreement. The disagreement has a known direction: the checker found more observability flags than the main pass (22 against 7 unique), so denominators are likely undercounted and the prevalences below likely overstated by roughly a fifth.

The standing caveat, which applies to every number in this report. This is a convenience sample of what was worth photographing, not a random sample of European retail. Absolute prevalence is soft. The cross-market contrasts are the load-bearing output, because the selection bias applies roughly evenly across markets. Proof is ordinal, not cardinal.

"Would this survive in a German forecourt under cost pressure?" is a real question with a fuzzy answer, so it was made a scored one. Every candidate is rated 0–5 on five friction dimensions — higher always means less friction — and those roll into a Transportability Index on 0–100.

The weights are the argument. Capex and ops carry half the model between them, because with a German operator under pressure the thing that kills an idea is almost never that it is a bad idea — it is that it needs a remodel or a headcount. Stack-independence gets a heavy 20% for the same reason: a large share of what is photographed here sits downstream of a loyalty app or a modern POS, and an operator without one cannot buy their way into it this quarter.

Friction alone produced a ranking that could not respond to evidence. A model scored only on capex, ops, stack-dependence, speed and culture judges every candidate from the nature of the mechanism rather than from what was actually photographed. It has no prevalence term, no evidence-class term and no readability term, so more fieldwork cannot move it. It also ranks a vinyl decal against a legally protected demand window in a single list, which is a category error. Four terms fix it, and one structural rule.

Three deliberate properties. It separates ease from value, so a cheap idea cannot out-rank a valuable one on cheapness alone — that is what the multiplicative form is for. It penalises unfalsifiable claims: readability asks whether a lever could ever be proven with data an operator can actually obtain, and a lever with no read design is discounted to 0.40 no matter how attractive it sounds. And it makes low scorers arguable rather than dismissed — every rank is a hypothesis that can be overturned by disputing a specific number.

Structural rows are ranked separately from, and above, execution rows. Structural advantage comes from a market condition — a legal window, a closed competitor, a captive dwell. Execution advantage comes from doing something better than the site next door. They are not sequential and they are not comparable: execution levers are the delivery mechanism for structural ones. A decal cannot outrank a protected trading window, and a protected trading window delivers nothing without the decal.

The first pass scored thirty mechanisms on friction and impact alone, before any German-specific term existed. It is kept because the shape of the field is the useful part: the upper-right quadrant is the working list, and the upper-left is where the interesting arguments are — high-value ideas that a capital-constrained operator cannot simply buy.

Prevalence across the 174 visits, with per-mechanism denominators. These are the numbers the Proof term is derived from, and the numbers the German gap map in Finding 1 is computed against.

Every claim below names the photographs behind it. Frame references are live against the archive index, so any assertion in this report can be taken back to the picture that produced it.

This is the finding that most changes the brief, and it runs against the intuition of almost everyone who works on a struggling German site. The instinct is to fix the food. The measurement says the food is not the problem.

The chart below compares the prevalence of fifteen merchandising mechanisms in Germany against the rest of Europe, using the same denominator discipline on both sides — a mechanism is only scored against visits where that part of the site was actually in frame. Read the bottom five rows first.

The single highest-value open question follows directly from this chart. What signage a private operator may legally place on or approaching a German road governs the largest measured gap in the set (+17pp). That question — logged as To Verify 3 — gates a €200 lever, which makes it worth clearing before a recommendation is made rather than after.

Most competitive advantage has to be built. This one is granted by statute and then left unused.

German shop-closing law removes most of the competitive set from the late-evening and Sunday windows. A forecourt site can generally trade through hours in which the QSR on the same street legally cannot — which is an advantage no amount of capital could otherwise buy, sitting on the balance sheet of every station in the country. Across the archive, extended-hours claims appear on signage at 14% of exterior-observable visits, in seven countries. The practice is entirely normal in Europe. What is unusual about Germany is not the tactic; it is the legal asymmetry underneath it.

The precedents for trading that window hard are photographed and specific. A Żabka fascia trades 6–23, seven days IMG_9776. A Madrid hot-dog operator makes the hours the hook itself — ABRIMOS HASTA LAS 7AM IMG_2687. A Sofia EKO and a German unattended snack pavilion both trade 24/365, the latter with the cheapest possible personality on the format with the least of it: "LET'S TALK ABOUT SNACKS BABY? 24 HOURS 365 DAYS" #294–295. And Germany already has the physical answer for hours when the shop itself is shut: an Aral/REWE To Go site runs a Nachtschalter — a night serving hatch IMG_4144.

Hot food in the late window is the highest-value occupant, and the reason it survives a German cost curve is that its marginal labour is near zero — the oven is already on site and the staff are already rostered. Żabka prints the operating model on a counter card: "WEŹ Z PÓŁKI I PODGRZEJ PRZY KASIE" — take it from the shelf and the cashier heats it at the till. Panini 12,99, Burgery 11,99, Rollo 15,99, Tosty 15,99, Zapiekanki 15,99, Pizza na wynos 19,99–22,99. The photograph behind the counter shows the entire apparatus: two contact grills and one oven. No kitchen, no cook, no separate food counter #532 · IMG_9883 #523 · IMG_9730.

The night trade itself is visible in the video record rather than the stills, which is why three photograph passes under-read it. A Dutch forecourt trades at 02:18 with a "Lunch deal 7,-" A-frame sign still standing between the pumps #230 · IMG_6486. FEBO's automat wall is being used by two workmen in hi-vis at 01:08 #228–229. An urban London Shell runs fuel under the canopy and 175 kW chargers in the same yard after dark IMG_8985.MOV. And a German open-air food destination at night shows packed terraces, lit kiosks and children still on the playground IMG_5269.MOV.

Levers produced. DE-01 (trade the protected hours), delivered by DE-25 (A-frame sign on the pavement), DE-29 (bestseller named on the entrance glass), DE-30 (time promise on the signature item) and DE-02 (hot food into that window). Poland runs the entire stack simultaneously: a 6–23 fascia, the hot dog named and priced on both door leaves, "ready in 3" on the glass. The legal moat is German; the kit is Polish.

A charging session is fifteen to forty minutes of captive, bored, standing dwell, twenty metres from a shop. It is the single most valuable unpurchased attention in forecourt retail, and across six countries it is being spent almost entirely on the charge point operator's own funnel.

This claim was originally recorded in the standing lever inventory as an absolute negative — "zero charger-to-store provocations found." A negative of that shape is fragile: one counterexample kills it. So it was audited properly. 57 charging frames across 29 distinct sites in six countries, every one read at full resolution, plus the 22 video clips, which turned out to matter.

Against that bar, four exist across six countries — and one of them is German

Kaufland and Lidl run ties, but they are inverted — they discount the electricity for cardholders and do nothing to the basket IMG_4163 · IMG_4167.

The negatives carry the claim better than the positives ever did:

Price the electricity against the opening hours. A McDonald's site prices its charging cheaper while the store is open and higher once it shuts. This is the strongest structural provocation in the set because it is a price mechanism, not signage — it cannot be ignored the way a poster can, it needs no creative, no screen and no staff, and it inverts the usual logic. Rather than asking the driver to come in, it makes the hours in which coming in is possible the cheapest hours to charge.

It also pairs directly with Finding 2. An operator trading through legally protected hours, when competing retail is shut, should price electricity down precisely in those hours — because the site is then the only place in the catchment to spend the dwell. Nothing in this archive runs that combination. Filed as DE-38 at Confidence C, because it rests on recall rather than a photograph; To Verify 13 asks for the market, the site count and the spread before it goes near a board. At TI 88 it has the highest transportability of anything scored, and clearing that one question moves it from ninth in the structural tier to roughly third.

Condition the offer on what the charge point already knows. A charge point knows three things a billboard does not: the weather at that exact spot, the daypart, and — derivable from arrival state of charge and connector power — roughly how long the driver is about to be standing there. Serving a hot drink in cold rain, and a twenty-minute item to a twenty-minute session, is a rules layer over hardware already installed. Filed as DE-39.

Norway is the most EV-saturated market observed, and it has no destination hubs at all — no services-scale food hall, no flagship charging forecourt. What it has is Circle K and Esso doing ordinary forecourt retail unusually well:

This finding changes the shape of the conversation more than any other, because it removes the objection a German operator raises first: that a Polish or Greek practice will not work here.

Nearly every lever sourced from abroad turns out to have a German precedent already in the archive — at Christmas markets, behind bakery counters, on motorway table tents, and in a supermarket produce hall filmed in 2017. The gap is not knowledge. It is that these practices are quarantined into seasonal and specialist formats and never cross onto the everyday forecourt.

Consequence for the boards. Eight levers across both boards can cite a German precedent alongside the foreign one. That does not change any priority score — the model measures friction and evidence, not vividness — but it materially changes the argument, and the argument is what gets a lever adopted.

The most widely copied idea in European forecourt foodservice is usually described as a Polish fry box. That description is too small, and it makes the idea look like a national quirk rather than the pricing architecture it actually is.

Worth naming as its own sub-mechanic, because it is the cheapest possible way to run a ladder and it recurs with striking consistency: a single flat number to move up a tier. McDonald's France, premium fries swap at +1,50 € #65 · IMG_0903. Pans & Company Madrid, upsize at +0,60 € #159 · IMG_2837. Sushi Daily Madrid, "CREA TU BANDEJA 10 PIEZAS 10,25€" then "HAZLO MENÚ +2,5€" #172 · IMG_3011. And the German A2 soup tent, Beilagensalat +3,40 € / Getränk +2,00 € #269 · IMG_4760. One number, one decision, no menu complexity — and it works on a kiosk screen, a printed card or a table tent equally well.

The strongest occasion device in the archive is also the cheapest: naming the moment rather than the product. A Bulgarian grab-fridge topper reads "Вземи за из път" — take it for the road — over a cola and chocolate-bar combo at 2,04 / 3,99 #383 · IMG_0135. Nine hundred kilometres away, Żabka prints Weź na drogę along the chilled case. The same three words, in two languages, on the same occasion — and the German version writes itself.

Two further devices ride on the same instinct and cost nothing. Heritage dates as everyday signage: "GRILLING SINCE 1954" on a Burger King building #97, "RODILLA — SINCE 1939" on a sandwich fascia #171, "STICKY CARAMELISED LARDY CAKES SINCE 1890" on an Oxford A-frame sign #22 — free, and Germany has more of it to draw on than any of these markets. And the premium sub-line with provenance: Burger King's floor tower reading "Gourmet King's · Peppercorn Collection · Aberdeen Angus Beef · Big taste, elevated" #98, or a case card reading "The World's Best Bocadillo — 5,50€ — WE'RE FARMERS" #179.

Print the eat-in and take-out prices side by side. A London bakery writes both on the same handwritten card: "SPECIAL TART — STOP IN 3.50 / TO GO 3.00", and beside it "CONTAINS PEANUTS · SALTED CARAMEL PEANUT BUTTER BROWNIES — STOP IN 3.45 / TO GO 2.95" #24–25. The gap is VAT, but it is presented as a customer choice rather than a tax artefact — and it quietly rewards the takeaway trade.

Germany has exactly the same split, at 7% for takeaway against 19% for eat-in, and most German forecourts absorb it silently into one price. Publishing both numbers turns a compliance detail into a visible saving, and nudges volume toward the lower-VAT channel — which is also the faster one, and the one that does not consume a seat. Very low load; it needs nothing but a price card. Filed as DE-47.

Most levers in this report compete for a share of existing traffic. This one inherits a catchment that has been vacated, and it is the only lever on either board that scores maximum on both readability and proof.

German bakery closures are dated, public and structural — they appear in trade registers and chamber data, which means the opportunity can be located on a map before a single euro is spent. The customers do not disappear with the shop. They redistribute to whoever, within a short drive, most looks like a bakery.

That resemblance is the whole intervention, and the archive prices it. A closed bakery's trade is not recaptured with par-baked goods in a plastic case. It is recaptured with a glass case, a printed card per item and an oven left visible — which is what a Norwegian forecourt bakery costs IMG_9261 #88 · IMG_9964. A Frankfurt bakery puts the entire promise on one chalkboard #267. The devices are cheap; the demand transfer is not.

The archive also holds an unintended longitudinal asset for this thesis. The oldest frames in the set are a REWE baking-mix aisle shot in Bavaria in December 2017, with own-label interleaved against Dr. Oetker at the same eye level under electronic shelf labels #3 · IMG_8227. As a private-label placement observation it is useful. As a baseline photograph of the German home-baking category taken before the bakery-closure curve steepened, it is considerably more than that.

Levers produced. DE-03 (recapture the closed bakery's catchment, the highest-scoring structural lever on either board), DE-28 (compete with the discounter bake-off), DE-27 (copy the town bakery's one item), DE-09 (sell one local pastry by name). DE-03 earns its position on two maxima the other rows cannot reach: readability 1.00, because closures are dated and public, and proof 1.00, because the claim rests on trade data rather than photographs.

A large share of the photographic archive is deliberately composed food imagery, shot to record where the European bar actually sits. The bar is real. But it is not set by the markets an operator would instinctively fly out to benchmark.

Splitting the 35-code codebook into craft (handwritten cards, per-item flavour cards, glass roundels, shelf product photography, named size tiers, open kitchen, "made here" claims) and systems (promo fixtures, physical loyalty, kiosks, digital menus, charging ties, meal-deal programmes, dietary and regional navigation) separates Europe cleanly into two models.

The Dutch sophistication is in the orchestration, not the fit-out budget — which is what makes it affordable to copy. Note what each of these theatrics costs: an illuminated word, a wooden crate, a green LED strip.

Honest limits on the two-model table: 13–26 visits per country and only 3–11 food-observable visits, so these figures are indicative rather than precise, and two markets with very small samples were excluded for that reason. What survives the small n is the ordering and the craft/systems inversion, which is large and consistent across both measures.

For an operator with no capital, the most important category in this report is the one where the capital comes from a third party. It is the only genuinely zero-capex group of levers, and it does not compete with the others — it funds them.

The honest caveat, which should be priced into any recommendation. Supplier fixtures buy space with assortment control. The margin gain is partly paid for in range freedom, and that trade — not the fixture — is the thing to negotiate. An operator who takes three coolers without negotiating the planogram has sold the shelf cheaply.

Why this finding sits last but funds the rest. Supplier-funded fixtures are the only lever on either board with genuinely zero capital cost, which is why they enter Board A high and displace nothing. Everything else on that board — the A-frame sign, the decals, the crate of fruit — is small money, but it is still the operator's money. This is the row that pays for it.

A single ranked list assumes every operator starts from the same place, and they do not. A site with a working food programme and charging on the forecourt has a different next move from a site near the brink with a QSR taking its lunch trade. So there are two boards, gated differently.

Starting from a Low Base assumes nothing: no fresh-food programme, possibly no charging, and a competitor opening on the same street. The gate is strict — no capex above about €500 a move, no new skill, no partner, live inside 30 days. Anything requiring a fresh-food programme, a charging estate or a loyalty stack is disqualified regardless of how well it scores. Starting from Mid-Tier assumes a working food programme, or charging on site, or both, and asks how a competent operator becomes a differentiated one.

Every card below carries the same fields as the standing lever inventory, plus the two things a ranked list cannot show: what was actually photographed, and why the move survives a German cost curve. The priority number is auditable — open the strip at the foot of any card and the arithmetic is there.

Priority against capital cost for all eleven moves on the low-base board. The point of the chart is the top of it.

The two highest-scoring moves on the board cost €0 and €200. All eleven together come to €890 in capital. The real cost is incremental payroll on two of them — and both of those were selected specifically because their marginal labour is close to zero.

Two open dependencies to carry into any recommendation. The A-frame lever is gated on To Verify 3 — what signage a private operator may legally place near a German road. That single unanswered question governs the largest measured gap on the board, so it is worth clearing before a meeting rather than after. And the opening-hours tariff lever remains a citation resting on recall: at TI 88 it has the highest transportability of anything scored, and clearing To Verify 13 would move it from ninth to roughly third in the structural tier.

Every recommendation above should be read against these six absences. They are also, straightforwardly, the shot list for the next trip.

The thirty mechanisms behind the effort-against-payoff chart, scored on friction and impact before the German-specific terms were added. Read up the transportability column rather than the priority column when the operator in question is capital-constrained.

A standing inventory of field notes predates this analysis. The large majority held up, often with a second or third instance in a market the note did not cite. Recorded below are only the entries where the photographic record adds something — a correction, a sharper specification, or a mechanism the note missed.