How the German GLP-1 telehealth market builds its consultations, sets its clinical rules and advertises

01Summary

Eight operators, no solved form. The German field splits into two incomplete strategies. Four operators run a light questionnaire and carry the legal risk of a questionnaire-only prescribing decision. One pushes the clinical depth into a mandatory video call and carries the operational cost instead. Only one attempts both, at around twenty-four screens, and it is the longest form in the market by a wide margin.

On one design choice the field has converged. Three of the eight tell the patient early whether they qualify: GoLighter returns a verdict on the screen immediately after height and weight, DoktorABC calculates a body mass index live in the page and states that it qualifies the patient, and Fit fürs Leben declares eligibility after two clinical questions. Two more, Voy and Juniper, give the verdict at the end. The operators that do not return anything are taking the patient's two most personal numbers and giving nothing back at the point where forms usually lose people.

Wellis's own German service sits outside this eight-operator crawl and is assessed here on its public marketing, legal and FAQ pages, plus a follow-up funnel attempt that reached screens one to nine of what is likely a twenty-screen-plus form, a real advance on the first pass, which stalled on screen one. The remaining stall is a timing limitation in the form's own field-confirmation behaviour, not a bot-wall, so several funnel-level questions stay open: consent structure and save-and-resume were not reached, though a pregnancy and contraception disclosure ending in a genuine "do not continue" choice, and an ethnicity question in the same family as Zava's and Voy's, were both confirmed on the way through. Its FAQ states plainly what the crawls could not observe directly: no in-form eligibility verdict, payment taken before any doctor contact, and prior GLP-1 use disclosed on a mandatory phone call rather than in the questionnaire. Prescribing and dispensing run through Netherlands-registered doctors and Wellis Pharmacy B.V. rather than a German entity, with no alternative pharmacy stated anywhere on the site, the same cross-border structure the pending Court of Justice reference concerns. Section 03 sets that out in more detail, and section 11 records the evidence that is still missing.

The follow-up crawl closed the largest gap in the earlier report. GoLighter's clinical questionnaire, previously recorded as unreachable, was captured in full: ten questions on a separate prescribing domain, with an outcome-graded prior-use question and an exclusion set wider than its question count suggests. It is the clearest evidence in the file that a short form and a thin clinical screen are not the same thing.

02Scope, method and how to read this

Eight operators were walked end to end on identical terms with synthetic intake data: MedExpress, Zava, GoLighter (commercial brand of Wellster Medical), Fit fürs Leben, DoktorABC, Dokteronline, Voy and Juniper. TeleClinic was checked manually and set aside as a different service category, a general reimbursed consultation rather than a GLP-1 prescribing funnel. Every crawl stopped at the first gate that could not be passed honestly: payment, account creation where terms forbid it, identity verification and photograph upload. No real or synthetic identity document or photograph of a person was ever generated or uploaded.

Wellis (not to be confused with GoLighter's parent Wellster Medical, a different company) is not one of the eight and was not walked end to end. Its public marketing pages, category page, FAQ and terms were read on 31 August 2026. A first attempt to walk its intake questionnaire with synthetic data stalled on screen one; a follow-up attempt using a different automation approach progressed to screen nine of what is likely a twenty-screen-plus form before a recurring timing issue in the form's own field-confirmation behaviour stopped further progress, a tooling limitation rather than a bot-wall or CAPTCHA. Consent structure, save-and-resume and the full clinical question sequence remain unreached, so Wellis is treated the way this report treats Dokteronline and Juniper for their unreached funnels: recorded as unknown rather than absent, and never averaged against the eight. The configuration assessed is the live service as operated on 31 August 2026 and may differ from the planned launch design.

Three claims carry different weight and are marked accordingly. Findings sourced to page text captured during a crawl are stated plainly. Findings drawn from an operator's own internal planning material, supplied for this study, carry an INTERNAL marker, they are not publicly observable and could not be checked against a rival on equal terms. Legal holdings rest substantially on court press releases and professional commentary rather than full published judgments, and are flagged where the underlying text was not read.

The reader most likely to misuse this document is one who reads a design difference as a preference. In several places it is not: German law removes options from every operator in this market, and a report that praised one operator for restraint or criticised another for caution would be wrong in both directions. Section 03 separates the two before any comparison is made.

03What the law forces and what operators choose

Three design choices in this market look like competitive positioning and are not.

• A synchronous step is a legal position, not friction. The Landgericht München I held on 3 March 2025 (4 HK O 15458/24) that remote treatment of obesity by questionnaire does not meet generally accepted professional standards, and that personal medical contact is required before prescribing a weight-loss injection, relying on the patient guideline of the Deutsche Adipositas-Gesellschaft, which contemplates blood and urine investigation. Fit fürs Leben's mandatory video call and DoktorABC's doctor-triggered contact are the most plausible responses to that ruling. The judgment is an interim injunction and is not final.
• The cross-border question is live and points at the prescribing structure, not the form. The Bundesgerichtshof referred the cross-border dimension to the Court of Justice on 26 March 2026 (I ZR 118/24, registered as Case C-265/26, lodged 31 March 2026), asking whether freedom to provide services under Article 56 TFEU precludes a ban on advertising remote treatment by doctors established in another member state. GoLighter's Ireland-registered prescribing practice on a separate domain and MedExpress's Spain and Portugal prescriber model are the two closest analogues in the crawled set. Outside the crawl, Wellis runs the same kind of structure: its doctors carry Dutch BIG registration numbers and its dispensing pharmacy, Wellis Pharmacy B.V., is Netherlands-registered, per its own published terms. Judgment is unlikely before 2027.
• Pharmacy-choice wording is compliance, not generosity. DoktorABC's statement that the patient may choose a partner pharmacy or take the prescription themselves, and Zava's offer of collection from any of 6,500 German pharmacies, are what §11 of the Pharmacies Act, extended in 2019 to pharmacies established elsewhere in the European Union supplying German patients, together with consent for transmitting health data to a named pharmacy, require. Wellis's own terms read the other way: they describe the contract for medicines as being with Wellis Pharmacy B.V. specifically, and nothing on the pages reviewed offers a choice of a different pharmacy. That reading is against the operator's stated terms rather than an observed funnel screen, so it should be checked rather than relied on.

Two further constraints shape every funnel here. Weight-management GLP-1 medicines are excluded from statutory reimbursement as lifestyle medicines under §34 SGB V, following the Gemeinsamer Bundesausschuss decision of 21 March 2024 in force from 15 June 2024, so every journey is private self-pay and the form is a commerce funnel as much as a clinical one. And the Heilmittelwerbegesetz restricts claims of certain success (§3), advertising of remote treatment (§9) and advertising of prescription-only medicines to the public (§10), which is why several operators' approach to social proof is a risk rather than a tactic. That is the subject of the Advertising view.

Health data adds a third layer that shows up directly in form design. Explicit, separable consent is required, so an itemised consent step is the expected construction and a single bundled tick is both a compliance question and a conversion advantage the operator taking it is enjoying at the others' expense.

04The fourteen mechanisms

Each row is something an operator does at a named step to move more patients through. Nature records where the row comes from: an internal decision means the operator could have done otherwise; a regulatory requirement means German law pushes every operator to it regardless of preference.

Sources: funnel crawl page text, 14 and 15 August 2026, follow-up 19 and 20 August 2026, for the eight crawled operators. Wellis column from its public marketing and legal pages, read 31 August 2026; a follow-up attempt walked its questionnaire to screen nine of a likely twenty-plus-screen form before a timing limitation in the form's own field confirmation, not a bot-wall, stopped progress, so cells marked "not observed" are unknown rather than absent.

05Architecture at a glance

Screen count is not clinical rigour, and the table is ordered longest to shortest to make that visible rather than persuasive. Fit fürs Leben's two questions defer the history to a call. GoLighter's ten carry a wider exclusion set than the count suggests. Voy's twenty-four reflect the most complete exclusion and interaction screening captured anywhere in this market. The clinical substance behind each count is compared in the Policies view.

MedExpress's form sits in the least favourable position available: long enough to feel like Voy's, without Voy's named clinician, transfer-in evidence or off-ramps, and without Fit fürs Leben's call. Wellis cannot be placed on this spectrum with the same confidence, its questionnaire opens in the same one-question-per-screen format, but the only architecture claim this report can make with evidence is the one made off the page: a free phone consultation described as a required step, earlier and more explicit than anything the eight offer inside their forms.

06Where each operator is strongest

07The unclaimed ground

Four mechanisms nobody in this market has taken, in order of how cheap they are to take.

08Design choices that carry risk

09What the market overlay predicted

The pre-crawl working hypotheses, tested. The overturned ones are the more useful half.

The pattern in the overturned rows is consistent: the market is markedly more exposed on public product advertising than a reading of the law would predict, and markedly less demanding on identity than a reading of the consumer-protection pressure would predict. Both gaps run in the same direction, towards conversion.

10Evidence and limitations

• Text only, no screenshots. Every competitor finding is sourced to page text read during the crawls of 14 and 15 August 2026 and the follow-up of 19 and 20 August. Nothing here is verified against an image.
• Two funnels were not reached. Dokteronline's questionnaire sits behind an account wall and Juniper's quiz stalled immediately after its contact-details screen, so their clinical content, exclusion sets and consent architecture are unknown rather than absent. GoLighter's questionnaire, recorded as unreachable in the earlier report, was captured in full on the follow-up crawl.
• Stated policy is not practice. Investigations by the BBC, Which? and ITV, and the German consumer bodies' test purchases, all found operators whose actual checks fell short of their published ones. A page is evidence of what an operator says, not what it does.
• Wellis sits outside the crawl; its questionnaire is now partly observed. Unlike the eight, most of its entries in this report are read from public marketing, legal and FAQ pages rather than a completed walkthrough. A first attempt to progress its intake form stalled on screen one; a follow-up attempt reached screen nine before a recurring timing issue in the form's own field-confirmation behaviour stopped further progress, a tooling limitation, not a bot-wall or CAPTCHA, so anything still marked "not observed" for Wellis is unknown rather than absent, and no comparison between Wellis and the eight is made on funnel mechanics that remain unseen. Separately, its German site carries two Dutch-market artifacts not otherwise checked across the field: a footer badge linking to the Dutch medicines register rather than a German equivalent, and a help-centre link on the German checkout page pointing to the Dutch-language help centre.
• The crawled set is not the nominal set. The pre-crawl brief named Zava, Fernarzt, TeleClinic, Juniper, Spring, Weight Doctors and Treated. Zava and Juniper appear in both. Fernarzt, Spring, Weight Doctors and Treated were not crawled; GoLighter, Fit fürs Leben, DoktorABC and Dokteronline were crawled and not named. Recorded rather than reconciled.
• The wider field is larger than eight. Operators appearing in the advertising evidence but outside the crawled set include DocMorris, HealthExpress, DrAnsay, GreenMedical, Apomeds, Kapsel, PharmaPremium, PrioOne, Solean, Oladoctor and the coaching provider Oviva. Kranus and Younimed appear in no evidence held. That is a gap, not an absence.

01Summary

This market has one settled policy and no settled clinical rule set. Every operator gates on body mass index 30, or 27 with a weight-related comorbidity, because that is the licensed indication for Wegovy and Mounjaro and because the consumer bodies test against it. Beyond that threshold the operators diverge widely enough that two of them screening the same patient could reach different answers, and the divergence is not a spectrum from light to thorough, it is different judgements about what matters.

Three positions exist on the commercially most valuable question. On transfer-in, Voy demands documentary proof before continuing a claimed dose and defaults to the lowest starting dose without it; GoLighter grades prior use by outcome, more or less than four per cent of body weight lost in three months, and asks for no documents; Zava takes the patient's word with a two-week gap rule. MedExpress asks the question and attaches no consequence to the answer, which leaves the reviewing clinician deciding without structured evidence. Wellis does not ask the question online at all, its FAQ states that prior GLP-1 use is disclosed verbally on the mandatory phone consultation instead, with the doctor deciding on a higher starting dose from there.

On exclusions the surprise is GoLighter. Its ten-question form is among the shortest in the market and its exclusion set among the widest: an age ceiling at seventy-five, severe psychiatric illness including suicidality, categorised blood pressure, and, harder to justify on current evidence, HIV and AIDS and Creutzfeldt-Jakob disease. Zava runs the opposite policy, converting most named conditions into comorbidities that qualify a patient rather than exclude them, and adjusting the healthy-weight threshold by ethnicity. Voy screens most thoroughly of all, with two exclusion checklists and three medicine-interaction screens carrying explicit monitoring timelines.

The most transferable idea in this view is not an exclusion at all. It is Voy's construction of a monitoring disclosure that ends in a genuine choice, made before payment. Wellis runs a version of the same construction at its own pregnancy and contraception screen, confirmed on a follow-up funnel attempt, and within the intake questionnaire itself, which places it before payment on the same logic as Voy's, though the exact sequence relative to checkout was not directly observed.

Competitor positions are sourced to page text captured during the crawls. Wellis positions are stated only where its public, legal or FAQ pages, or the follow-up funnel attempt, support them, and are marked as such throughout this view. No line-by-line reconciliation of any single operator's exclusion list against the whole field exists yet; the table in section 03 covers the three funnels where the complete set was captured.

02Eligibility and thresholds

Two things follow. GoLighter's flat 27 is the loosest published threshold in the market and sits below the licensed indication for a patient with no comorbidity, which makes it the most likely of the eight to fail a consumer-body test purchase. And the operators that state a threshold early are the same ones that return a verdict early: the threshold and the feedback are one design decision, not two, which is why an operator that publishes the threshold but says nothing in the form is paying twice for the same caution.

03What the field screens out

Recorded from the three funnels where the complete exclusion set was captured. A dash means the item did not appear, which is evidence of absence for these three forms only.

Four differences that matter

An age ceiling exists at one operator only. GoLighter excludes seventy-five and over; nobody else does. Frailty and the current safety signal make it a defensible position, and its absence elsewhere is a considered position only where somebody has actually considered it.

Severe psychiatric illness including suicidality is screened by GoLighter alone. Against the volume of regulatory attention on psychiatric adverse events with this drug class, that is the most substantive single gap the crawl found across the market, and it sits at seven of the eight operators rather than at one.

HIV, AIDS and Creutzfeldt-Jakob disease read as inherited rather than reasoned. Neither has a plausible mechanism-based link to GLP-1 prescribing, and both appear only in the operator with the widest list. Width is not the same as quality, and this is the row that shows it.

Ethnicity as a threshold adjustment is run by two of eight, Zava and Voy, both with an explanation attached and Voy explicitly linking it to prescribing outside the licensed indication. It is a clinical position with a data-protection dimension, not a form question, and the six operators that do not ask have taken the opposite position by default rather than by decision. Outside the eight, Wellis asks the same question on screen four of its intake form, listing South Asian, East and Southeast Asian, Middle Eastern, North African, African and Afro-Caribbean background as a lower-threshold risk flag, a third operator running the same clinically legitimate practice, though its questionnaire was not reached far enough to confirm what the answer then changes.

04Transfer-in, switching and dose continuation

The widest divergence in the market, on the most valuable patient in it. Someone arriving on an existing dose is the easiest patient to convert and the easiest to get clinically wrong.

Voy's wording is the most useful artefact in the crawl, because it converts a clinical control into a conversion mechanic: the patient is told exactly what to supply and exactly what happens without it, which turns a refusal into a lower starting dose rather than a lost patient. GoLighter's outcome grading is the more interesting idea and the weaker control, it captures response rather than exposure, so it cannot distinguish a patient stepping up from one restarting after a long gap. The unoccupied position is the combination: ask for evidence, accept its absence, default low, keep the patient.

Entry route matters as much as evidence. Zava is the only operator routing a returning patient to a pre-filled repeat questionnaire, and GoLighter asks at the top of its funnel whether the patient is already a patient, warning that answering yes in error will delay or reject the request. Neither is a clinical control; both express the same policy as a shorter journey.

05Interactions and monitoring

The sharpest divergence is not what an operator excludes but what it does with a medicine it will not exclude.

06Evidence of identity and weight

The weakest link in this market, and the one a consumer body has already tested to destruction.

The test purchases run by the Verbraucherzentrale Nordrhein-Westfalen and its Rhineland-Palatinate counterpart between 2 and 11 March 2026 are why this matters. Of ten platforms identified, four required no proof of identity or weight at all. None dispensed to a tester at a body mass index of 21.6. After the tester adjusted the entered weight to a body mass index of 32.4, five platforms prescribed without difficulty and one accepted artificial-intelligence generated body photographs, with no personal doctor contact anywhere. That finding is the empirical spine of every enforcement argument now being made in this market, and it lands on the questionnaire-only model rather than on any one operator.

Two consequences follow for the market as a whole. Requiring nothing puts an operator in the same group as the four platforms the consumer body singled out, and the threshold the report itself implies, that a normal-weight tester, or one entering a falsified weight, cannot obtain a prescription, is a test most of this field would currently fail. And GoLighter's deferral pattern shows the requirement and the friction can be separated: ask for the photograph, let the patient submit without it, chase by email, and hold the prescription rather than the form.

07Pregnancy, contraception and fertility

Two of the eight go well beyond a yes-or-no question, and both do it in the same place: an information screen between the clinical questions and the decision. Voy states that treatment is unsuitable in pregnancy or when trying to conceive, that contraception is required if the patient is sexually active, that Mounjaro requires a non-oral method such as a condom or a coil, and that patients on an oral contraceptive need an additional method because its effectiveness may be reduced, then offers "I understand" or "I do not want to continue". GoLighter recommends a non-oral method for the first four weeks and at every dose increase, names all three molecules in the warning, and adds a point nobody else makes: that these medicines can increase fertility through weight loss, and that a patient trying to conceive, pregnant or breastfeeding should stop the medication. Outside the eight, Wellis runs the same construction at the one screen a follow-up funnel attempt reached: it names Mounjaro directly, requires a non-oral method, and ends in the same "Verstanden" or cancel choice as Voy's, a third, confirmed example rather than one inferred from marketing copy.

The fertility statement remains GoLighter's alone, the only place in this market where an operator tells a patient something that works against the sale, and it is clinically correct. All three examples share the structural point that runs through this view: a disclosure ending in a genuine choice is worth more evidentially than a tick, and costs one screen.

08Stopping, continuing and missed doses

Only Juniper publishes a full set: a recommended treatment duration of six to twelve months, no known withdrawal symptoms on stopping, appetite and satiety effects persisting for weeks afterwards, some patients regaining weight, and a missed-dose rule worked through with a calendar example, take within five days of the due date, otherwise skip and resume on the next scheduled day. GoLighter routes switching decisions to the medical team and states an interval of about a week between injections. No other operator publishes a stopping policy at all.

The continuation floor is the market-wide gap. Published floors run from body mass index 19 to 23, with 21 the most common, and none of the eight crawled forms surfaces one to the patient. The Munich court's reasoning shows why this is not a detail: it relied partly on the defendant's own advice to discontinue if less than five per cent of body weight was lost within three months, using an operator's stopping policy as evidence that the condition requires monitoring a remote prescriber cannot provide. A published stopping rule is therefore both a clinical safeguard and, on that reading, a piece of evidence against the model that publishes it.

09Prescription routing and withdrawal rights

Zava runs the broadest routing policy in the market and states it in full: free same-day collection from any of over 6,500 German pharmacies, with the prescription transmitted electronically to the pharmacy the patient chooses when the request is approved before 15:30; free or express partner-pharmacy shipping including cold-chain medicines; or a paper prescription by post for a fee. Delivery to near-border parcel shops is refused outright, citing medicine-import restrictions, and such orders are cancelled. Its footer discloses that medicines may be supplied by a named Netherlands pharmacy, which is the shipping route and not the other two. GoLighter offers a prescription-only route so the patient fills at their own pharmacy, and a route out of the online funnel to find a local doctor. DoktorABC states the choice in the market's clearest single sentence. Wellis reads the other way on the pages available: its terms describe the medicines contract as being with its own captive pharmacy, Wellis Pharmacy B.V. in the Netherlands, and nothing reviewed offers a German alternative, the position §11 was written to discourage, on the wording of the terms rather than an observed funnel screen.

None of this is generosity. §11 of the Pharmacies Act, extended in 2019 to pharmacies established elsewhere in the European Union supplying German patients, plus consent for transmitting health data to a named pharmacy, is what produces the wording. A Frankfurt ruling of 28 May 2025 (2-06 O 150/25) found no unlawful steering where the customer can choose to take the prescription and redeem it at a pharmacy of their own choice, which makes GoLighter's prescription-only route the safe-harbour construction rather than a commercial concession. An operator that never hands the patient a transferable prescription has built commercial lock-in out of a transmission choice.

On withdrawal rights the field splits several ways. GoLighter excludes the fourteen-day right for cold-chain medicines and names Wegovy and Saxenda directly in doing so. Zava bundles the withdrawal notice with its terms and privacy policy into a single tick. Wellis's terms state a fourteen-day right for non-prescription products only, and say plainly that prescription medicines cannot be returned "for health and safety reasons", the same substantive position as GoLighter's exclusion, reached by a blanket rule rather than a cold-chain carve-out, and on the wording of the terms the clearest of the positions recorded here under §312g(2) BGB, the provision that excepts certain goods, including sealed goods unsuitable for return on health protection or hygiene grounds, from the fourteen-day withdrawal right. The June 2026 change is a separate obligation: §356a BGB requires an electronic cancellation button for distance contracts, which governs how a contract is cancelled, not whether a dispensed medicine can be returned. LEGAL REVIEW

10Commercial model and price position

Because these medicines fall outside statutory reimbursement, price is a live competitive lever and the pricing structure is itself a policy choice. The figures below are from a competitor pricing tracker with data to July 2026 and should be treated as a snapshot in a fast-moving market. INTERNAL

Three corrections to the common reading of this market follow from the table. First, there is no medicine price leader: six of eight operators start Wegovy at €171.96 or €172, which is convergence rather than competition, and the only material undercut is Voy's €152 on a twelve-month commitment. Wellis's published €172.00 sits inside the same convergence, for the record, though it is outside the eight and its payment structure was not confirmed against an actual checkout. Second, the fee is where the real spread sits (€30 to €49 among those charging one, against four operators that bury it in a subscription), so an operator competing on price is competing on the fee and on payment structure, not on the medicine. Third, payment model splits the field more cleanly than price: six operators run subscriptions with the fee included and two run one-time orders with a visible fee. The subscription operators are also the ones offering guarantees, which is the same commercial logic viewed from the other end.

Review volume tells a different story from review score. Ratings cluster between 4.5 and 4.8, so the score discriminates poorly; Zava's 4.8 rests on more than 42,000 reviews while a category page elsewhere in the field displays under 2,000. Two scores moved materially in a single month to July 2026, DocMorris from 4.6 to 4.0 and Oviva from 4.5 to 4.0, and no source held explains either fall.

Source: DE GLP-1 competitor pricing tracker, data to July 2026; category-page review counts from the 14 and 15 August 2026 advertising capture.

11Guarantees and refunds

An outcome-conditional guarantee is a different order of claim from a described typical result, and it is not primarily a refunds question. §3 of the Medicines Advertising Act restricts claims of certain success, and a contractual promise of a specific weight loss sits closer to that line than any percentage on a landing page. The court that enjoined GoLighter's advertising did so partly over the phrase "reach your target weight". Wellis's refund is conditional on a clinical suitability decision rather than a weight-loss figure, which keeps it out of that argument entirely, a materially different, and lower risk, construction from GoLighter's and Voy's. The guarantee mechanics belong here; the claim risk is in the Advertising view.

12Where the clinical positions diverge

Recorded, not resolved. Each is a point where operators in this market hold materially different clinical positions, and the difference is a judgement for a market medical lead rather than for a funnel comparison.

01Summary

German law forbids naming the medicine, naming the class, discounting it, promising an outcome from it, or advertising the remote consultation that leads to it. §10 of the Medicines Advertising Act bars advertising a prescription medicine to the public at all, and since the Bundesgerichtshof ruling of 26 March 2026 naming the class plus its indications and offering a route to a doctor is enough to breach it. §9 bars advertising the remote consultation itself unless recognised professional standards say personal contact is unnecessary, which a Munich court has held a questionnaire-only obesity pathway does not satisfy. What is left after those two provisions is narrower than a product page but wider than nothing: naming the condition, explaining the service, showing named clinicians, a plain treatment-service price, a genuine review rating, and offering the patient a paper or PDF prescription so they can choose their own pharmacy are all open, and none of the eight operators in this set has built its public estate around that list. Section 03 sets out the register in full.

The field is not competing on compliance and it is not, on the evidence, competing on the register that is actually open to it either. It is competing on how far past the line to go, and on how well structured each operator is to survive being caught. That is a second, separate failure from the legal one: German buyers rank trust and medical credibility above price in every sample this market has been asked, and the constraint on this category's growth is reported as legitimacy rather than awareness, roughly two-thirds of adults already know the drug class exists and roughly four in five say they would not use it themselves. The register the law leaves open (condition education, named clinicians, transparent pricing, verified reviews) is also the one this market's own research says converts. Restraint here is a commercial argument as much as a legal one, not only a legal constraint dressed up as a compliance instruction.

Four of the eight name a medicine on a public page with no gate: MedExpress, Zava, Dokteronline and Fit fürs Leben. Three of those four also place a percentage weight-loss claim and a price in the same view. Only Voy and Juniper gate the molecule, and only Voy's is verified, a direct text search of its public site returns no brand or active-ingredient name anywhere, with the first mention several screens inside the questionnaire. Dokteronline is the most exposed of the eight, listing twelve named products with efficacy figures including four investigational compounds not authorised in Germany.

The market's second axis is social proof, and it runs the opposite way. Five of eight use named or initialled testimonials carrying a specific weight and timeframe. Two operators, Juniper and Voy, have built their paid social almost entirely on it: twenty-seven active creatives, three-quarters video, before-and-after imagery and customer reviews in nearly half at one; all video, a review badge on every ad and a quarter influencer-generated at the other. This is the part of the market where §11 and the Cologne appeal court's known-person doctrine sit closest to the commercial engine, and where the enforcement record is thinnest so far.

Enforcement is not coming from the medicines regulator. The Bundesinstitut für Arzneimittel und Medizinprodukte has taken no published advertising action against a telehealth platform, and there is no German equivalent of the United States programme that sent warning letters to thirty telehealth companies in February 2026. It comes from competition associations, pharmacists' chambers and, faster than any of them, the advertising platforms themselves. Three of the eight are currently dark or disabled on paid social.

The most useful finding for any operator in this market is that the two most exposed positions are separable. An operator can hold the strongest verification and certification posture in the field and still carry the most exposed public product page, because they are different decisions taken by different functions. Section 10 works one operator's position through in detail to show how that happens.

02The five provisions that bite

• §10: no advertising of prescription medicines to the public. The sharpest and least contestable, implementing Article 88 of Directive 2001/83/EC, so it is fully harmonised and no country-of-origin argument rescues it. On 26 March 2026 the Federal Court of Justice held that naming a class of prescription medicine with its indications, and enabling treatment requests to co-operating doctors, is prohibited public advertising even where no product or manufacturer is named. The Munich court went further and held that the colloquial "Abnehmspritze" itself identifies the class, because the relevant public knows which medicines are meant.
• §9: no advertising of remote treatment unless generally recognised professional standards say personal medical contact is not required. The advertiser carries the burden of proof, and the standard is measured against §630a BGB rather than professional-conduct rules, so a treatment can be lawful to provide and unlawful to advertise. This is the provision now before the Court of Justice as Case C-265/26, lodged 31 March 2026, with judgment unlikely before 2027.
• §3: no misleading advertising and no promise of certain success. This is what the phrase "reach your target weight" broke, and it is the provision an outcome-conditional money-back guarantee points at.
• §11: content restrictions on lay advertising, including statements of gratitude and recommendation, and advertising with known people. An appeal court in Cologne held on 11 September 2025 (6 U 118/24) that general fame is not the test and that recognisability below that threshold suffices, and that the advertiser is liable for the influencer as its agent. The case concerned an eighteen-second social video for a non-prescription cold remedy by an account with around 120,000 to 130,000 followers.
• §7: no inducements or gifts in product-related advertising. This is why every operator in the field discounts the consultation fee rather than the medicine, and why value-adds are framed as service extras.

All of these operate as market-conduct rules through §3a of the Unfair Competition Act, which is what gives competitors and qualified associations standing to sue. A separate layer, the audiovisual rules on commercial communications for prescription-only medicines, applies to any video or connected-television creative.

03What the law actually leaves open

The five provisions above describe what is closed. Read the same statutes and the case law for what they leave standing and a genuine register emerges, narrower than a product page, but real, and mostly unused in this market. The generalisation that holds across every row below: the compliant version of an asset is usually a different asset, not the same asset with a disclaimer added.

Two structural moves sit between fully open and fully closed and are worth naming because operators in this market use both, with mixed results. Separating the advertisement from the page it lands on, a splash page carrying no product, with the questionnaire or product page a deliberate second step, genuinely weakens the argument that the advertisement itself is product-sales advertising, though it does nothing about a class term placed on that same splash page. And bidding on a product name as a paid search keyword while keeping it out of the visible advertisement is a real, untested argument (the keyword is never seen by the consumer), but it collapses the moment dynamic keyword insertion pulls the term into the ad automatically, which is what one operator's own account appears to have done. Both are treated in the register above as contested rather than open, because neither has been tested in this market and both fail outside a narrow configuration.

Two developments make the open register more valuable than it looks on paper rather than less. First, the March 2026 ruling that closed the class-euphemism route removed the entire market's working compliance device at once, and on the evidence in section 05 nobody has moved off it, which means every operator currently relying on "Abnehmspritze" in a page title is exposed, and an operator that builds clean from the open register now is not behind the market, it is ahead of a market that has not noticed the ground shifted. Second, foreign establishment has protected nobody in the reported case law so far, two of the enjoined operators were established outside Germany, so the open register above, not country of establishment, is the more reliable place to build a defensible public position.

04Who enforces, and how fast

The procedural shape matters more than the identity of the claimant. A cease-and-desist letter is followed, often within days, by an application for a preliminary injunction, which can be granted in weeks and carries penalties of up to €250,000 per breach. Main proceedings then run for years. Every GLP-1 advertising decision in this market so far is interim and appealable, which means the law shaping the market is provisional and the commercial consequences are not.

05Where the medicine gets named

Half the market names a medicine publicly, which is the market norm and not a safe harbour: the two injunctions of March 2025 both landed on this posture, one against a Netherlands pharmacy and one against a Munich-established operator, so neither foreign nor domestic establishment protected the advertiser. Wellis's category page, outside the eight, sits in the same exposed group, with brand and price together and no gate of any kind.

Paid channels behave differently again, and a competitor advertising review held in this market records the pattern. Search ads largely avoid brand names and use "Abnehmspritze" or "Diät Spritze" instead, with Zava the exception, running Mounjaro in a search headline while keeping it off the landing page. Shopping ads invert it and put product names in the headline, HealthExpress and Dokteronline among them. One operator's own internal review records two aggressive ad variations, built around pen wording, a discount code and the full price in the headline, delivering 45 per cent of impressions and conversions, with shopping ads on drug-related terms carrying around 80 per cent of shopping traffic. INTERNAL The commercial dependence on the exposed side of the line is measurable, which is what makes this a decision rather than a cleanup.

The United States brand name appearing in a German footnote reads as reused source material rather than a deliberate choice, but it still places a named medicine on a public German page.

06Quantified claims and social proof

Two constructions in this table are riskier than the percentages that dominate it. A quantified personal result inside an unverified customer review is a testimonial for §11 purposes whether or not the operator wrote it, so an operator running no named testimonials may still be carrying testimonial exposure through a review widget. And a projection tied to the patient's own entered weight (two operators do this) converts an average into an individual promise at the moment of highest intent. Wellis is the outlier on both counts: no outcome figure and no named testimonial on any page checked, which is the most conservative public claims posture in the set.

07Thirteen ways operators stay out of trouble

Mostly not compliance strategies. Most of these are techniques for advertising something that cannot lawfully be advertised, in a form that is harder to enjoin, and graded against the register in section 03, most of them do not actually survive contact with the statute. A handful are the genuine article. The grade after each title says which.

08Who got caught, and for what

Two lessons. Establishment protected nobody: one injunction hit a Netherlands pharmacy and the other a Munich company, on materially the same conduct. And nothing on this list was brought by a regulator. Every action came from a chamber or a competition association, which means the trigger is a complaint rather than an inspection, and the shortest route to being enjoined is being the most visible.

09The paid media picture

Creative mix and search visibility from a competitor tracking programme with data to July 2026. INTERNAL

The market churns fast enough to make all of this provisional, three of the top five operators by search visibility in October 2025 had left the leaderboard by February 2026. Cost pressure is the constant: cost per click across the market rose 67 per cent between October 2025 and February 2026 on aggressive bidding, and one operator reports acquisition cost running at €170 against a €140 budget with twelve-month lifetime value against acquisition cost at 0.3. INTERNAL That pressure is the mechanism by which advertising risk increases in this market: it is created by the auction, not by any operator's appetite.

10The platforms as second regulator

Platform policy is stricter, faster and less negotiable than the courts, and in this market it diverges from the law in both directions. Google's healthcare and medicines policy makes prescription-drug services a restricted category requiring certification, and Germany is on the list of locations where certified advertisers may run them, so a campaign can be technically permitted by the platform and still breach §10. Meta does not permit prescription-drug promotion outside a short list of countries that does not include Germany, so the question does not arise there.

The practical consequence is that platform enforcement is where this market actually gets policed, and it arrives without a hearing.

11Reading Wellis's public position against the field

Wellis is read here against the same claims a market operator might make about itself, using only its public marketing and legal pages. Nothing below draws on an observed questionnaire, and each line says so where it matters.

Three of these six lines rest on Wellis's own published terms and marketing copy rather than an observed questionnaire, and are marked as such in the table. A terms-derived reading is a reasonable starting point for Medical and Legal, not a substitute for walking the funnel: the pharmacy-choice and cross-border findings in particular should be confirmed against an actual completed session before either is relied on in a compliance assessment.

Set against the open register in section 03, three moves are available to Wellis now, independent of anything else in this document. It already displays a genuine review rating (4.7 from 2,778 reviews) and carries no outcome-figure or named-testimonial exposure, both squarely in the open column, and worth keeping rather than adding to. What is missing is the register's other side: nothing on the pages checked discloses the Dutch establishment of its doctors and pharmacy prominently rather than in the terms, no PDF-prescription or alternative-pharmacy option is offered to answer the single-pharmacy exposure above, and its category page still carries the same struck-through pricing and dual brand names as the market's most exposed operators, where a plain, all-in service price would sit in the open column instead. None of the three requires waiting on the Court of Justice.

12The riskiest constructions in the market