What is the pattern?
The pattern is retail thinking applied to a regulated catalogue, and it involves nobody's ill intent. A pharmacy owner commissions a website as any small business would: the portfolio looks competent, the price is reasonable and the agency has built shops before. The agency then does what agencies do well. It applies the e-commerce patterns that maximise conversion — product grids, prices, baskets, reviews, urgency messaging, abandoned-cart emails — to a product catalogue that happens to consist of medicines.
For parts of that catalogue, every one of those patterns falls somewhere between inappropriate and unlawful. Neither party realises it. The owner assumed web professionals understand web law; the agency assumed the client understands its own industry's rules.
The site launches and trades, and the trouble surfaces later. An acquirer's underwriter declines, a certifier queries the prescribing flow, an inspector examines the prescription-only pages, or a competitor reports the advertising. The build was cheap. The investigation that follows is not.
What does the generic build omit?
The missing layer is the one this site's library documents piece by piece — and a pharmacy site inherits all of it at once.
The GPhC's standards treat the website of a registered pharmacy as part of the registered service, with the voluntary logo and registration transparency expectations that follow. Medicines advertising law divides the catalogue three ways. GSL retail is retail. P medicines may be sold online only within a supervision and intervention architecture. And POMs may not be advertised to the public at all — which prohibits, for prescription lines, precisely the product page agencies build first: the priced, basketed, review-decorated listing.
Service models bring the prescribing-flow transparency the certification and remote-prescribing frameworks require: consultation before checkout, the model described honestly, prescriber independence visible. Health data makes the forms layer consequential — a symptom checker feeding a marketing system is a special-category data incident with a newsletter attached. The policies layer must state what consumer law actually provides, not what a template supposed. And the payments layer reads all of it, because the acquirer and the certifier underwrite from the rendered site, not from the intentions behind it.
None of this is obscure and all of it is discoverable. Yet none of it appears in a generic brief. That is the failure in a single observation.
What do the failure modes comprise?
A quick survey of what reaches this site's inbox.
The prescription-only shop. Prescription medicines displayed on public pages with prices, stock indicators and baskets. That is promotion of POMs to the public, built as diligently as any retail grid.
The inverted checkout. Payment first, then a questionnaire, then — notionally — a prescriber. This is the sell-first architecture every regulator and certifier in this library treats as the primary indicator.
The unsupervised P shelf. Pharmacy medicines sold without quantity limits, without an intervention step and without a pharmacist appearing anywhere in the flow.
The testimonial wall. Reviews attached to POM services: patient endorsements doing the promotion the law forbids the pharmacy to do directly.
The leaking forms. Health enquiries submitted through a general contact form, routed to a shared inbox and a marketing platform, and kept indefinitely on no identified lawful basis.
The retail policies. Fourteen-day returns promised on dispensed prescriptions — the precise opposite of the statutory position — drafted from a template.
The delayed launch. The completed, paid-for site that cannot pass acquirer underwriting or certification review — discovered in week eleven of a twelve-week programme.
Each failure is competent execution of the wrong specification.
Why do subcontracting chains amplify the failure?
Here is an industry mechanic agency proposals seldom mention: much of the agency market does not build websites, it sells them. Production is subcontracted to white-label studios and offshore teams, sometimes through more than one layer. The economics are rational — sales and account management stay onshore, production goes wherever it is cheapest. For a bakery's website the model is harmless.
For a pharmacy it fails structurally, because a brief degrades at every handover. The compliance paragraph from the first meeting becomes a bullet in a scope document. That becomes a ticket describing product pages as per the mock-up. The ticket sits in a queue worked by a developer who has never encountered the Human Medicines Regulations and is measured on delivery speed. Nobody in the chain owns regulatory outcomes, and no layer is paid to raise objections. The one person who could spot the errors — the superintendent — reviews at the end, once the information architecture, the checkout order and the data flows are settled.
The remedy is not to ban subcontracting — the brief below addresses it candidly — but to refuse anonymity. The client knows who is building. Compliance requirements travel as acceptance criteria that survive handover, and a named person with authority signs each one off. Visible chains are manageable. Invisible ones are where pharmacy sites fail.
What are the economics of the inexpensive build?
Priced honestly, the cheap build fails on arithmetic before you ever reach principle. The generic quote wins the tender by thousands. Then the ledger starts. Weeks of launch delay while an acquirer's decline is diagnosed and the site is rewritten for underwriting review. Premises costs, stock finance and staff go on consuming resources in a business that cannot yet accept a card. Remediation done under time pressure and, in the worst cases, under regulatory attention — the most expensive procurement environment there is. A second build, because architecture-level failures in checkout order, data flows and catalogue structure are rebuilds described as corrections. And the indirect costs: the certification application delayed by the website, the advertising accounts that cannot open, the founder-months spent on investigation.
Set against that, the specialist premium — or the extra weeks of a properly briefed generic build — is not a luxury. It is insurance priced at a fraction of the claim. The industry's pricing illusion is that the build cost equals the figure on the proposal. For a regulated site, the build cost is the proposal figure plus the probability-weighted cost of failing four audiences — and the second term dominates. Owners who have lived through one delayed launch need this paragraph only once.
Who are the five audiences?
A pharmacy website is read by five audiences; a generic build serves one. Patients, for whom the agency optimises. The GPhC, whose standards and inspectors treat the site as part of the registered pharmacy. The MHRA and the advertising rules, under which every page is an advertisement with a legal classification. The payments layer — acquirers, certifiers and advertising platforms — which underwrites the business from the rendered site, as an entire cluster of this library established. And the future: the bank, purchaser or investor whose diligence will one day read the site as evidence of how the business is run. The protective brief exists because four of the five never attend design workshops. A site that satisfies the first audience while failing the other four is the industry's default output.
What does the protective brief contain?
| Clause | What it requires | Acceptance test |
|---|---|---|
| Catalogue classification | Every product tagged GSL, P or POM, with page behaviour specified per class | No POM page displays price and basket to the public, and the P flow includes intervention and limits |
| Service flow order | Consultation and prescriber decision preceding payment for POM services, with the model described truthfully on-page | A test patient cannot reach payment without completing the clinical step |
| Transparency set | Registration details, superintendent, prescriber information, ownership and complaints route displayed per the standards | Every item locatable within two clicks from any page |
| Data map | Every form's fields, destination, lawful basis and retention documented, with health data segregated from marketing | The data map exists as a deliverable and corresponds to the build |
| Policies | Refunds per the statutory matrix, together with privacy, terms and cookies, drafted for pharmacy rather than adapted from retail | Policies cross-checked against the site's actual flows without contradiction |
| Payments readiness | Gateway per the cluster's criteria, card data architecturally excluded, and the site written for underwriting examination | The certification and acquirer questionnaires may be answered from the site without difficulty |
| Accessibility and performance | A stated accessibility standard and performance budget, tested rather than aspired to | Audit reports delivered at handover |
| Named builders and sign-off | Who actually builds, disclosed, together with a named compliance owner and superintendent sign-off gates at architecture, content and launch | A sign-off record exists for each gate |
| Ownership and access | Client ownership of domain, content, code and design, with administrative access to hosting, DNS and analytics from the first day | Access verified before final payment |
| Change control and exit | Post-launch changes to regulated surfaces passing through review, with documentation and priced exit assistance | The change log exists and the exit clause contains figures |
Two clauses carry the greatest practical weight. Sign-off gates move the superintendent from end-of-project reviewer to architectural authority. That is where the failures described above become preventable: the checkout order is a first-gate decision, not a launch-week discovery. And ownership and access is the clause that stops a compliance problem becoming a dependency. A pharmacy that cannot edit its own prescription-only page on the evening a regulator writes about it does not have a supplier relationship. It has a dependency with invoicing.
Everything in the table is ordinary professional practice. The only unusual part is writing it down — which is precisely why writing it down sorts the market.
Generic, specialist or in-house?
The honest position, with the disclosure repeated: this site's publisher is a specialist, so examine the reasoning rather than the conclusion.
A generic agency is a sound choice for a brochure site: locations, hours, team and service descriptions, with no commerce or health-data capture. It should still work to the brief's transparency, policy and ownership clauses. The compliance surface there is small enough to specify completely.
The specialist case strengthens with each regulated feature: P sales, POM services, prescribing integration, subscriptions. Compliance there is architecture, not copy. A builder already familiar with the five audiences turns the brief from an education programme into a checklist.
In-house building suits groups with genuine product teams and continuous development needs. Carry the same brief internally: employment does not confer regulatory knowledge either.
The wrong answer is the same in all three: any builder, whatever its label, working without the brief. The label was never the variable. The specification always was.
What does the site require after launch?
A pharmacy website is never finished, because its compliance surfaces track a moving framework. The brief's change-control clause is where the ongoing care lives. Three habits keep a launched site inspection-ready.
Change control on regulated surfaces. Product pages, service flows, policies and claims change only through the sign-off route, however minor the edit. A small, apparently harmless amendment to a prescription-only page is how compliant sites drift into advertisements.
A claims review within the content rhythm. Every article, landing page and campaign faces the question the certifier and the MHRA will ask: does the claim fall within the licence and the evidence? Ask it before publication, not after a complaint.
A standing audit against the five audiences. Quarterly, someone reads the site as a patient, an inspector, an advertising regulator, an underwriter and a diligence lawyer, using the brief's acceptance tests as the script. Frameworks move — this year alone has revised label rules and payments regulation, among other things. A site that passed in January may fail in July without any code having changed.
The builder relationship should survive launch as retained capacity for exactly this, at documented rates — or the in-house team should own it explicitly. The failure mode is the orphaned site: compliant at launch, drifting afterwards, reproducing the generic-build failure pattern slowly through neglect.
Which questions examine any builder?
Seven questions you can ask in one meeting. There is no wrong way to put them, but there are wrong answers.
Which pages of a pharmacy site may lawfully show a price and a basket, which may not, and on what basis? Where does the pharmacist appear in a P-medicine purchase on your build? What happens, in sequence, between a patient reaching a POM service page and money being taken? Where does the data from a health enquiry form go, and who else can access it? Who, by name, will write the code and the product content, and who signs off compliance-relevant changes after launch? What will our acquirer and certifier read on this site, and have your builds passed those reviews before? And what do we hold, from the first day, that lets us leave you?
A specialist answers from experience. A capable generalist answers from the brief you supplied. A builder who answers with design awards has answered the only question that mattered. The pattern this article opened with persists because these questions are not asked — and seven questions end it.
Key takeaways
- Generic pharmacy-site failure is structural, not incompetent: retail patterns competently applied to a catalogue where those patterns are unlawful, under briefs that never mention the rules.
- The missing layer is the library itself — catalogue classification, service-flow order, transparency, health-data handling, statutory policies and payments readiness — all inherited at once.
- Subcontracting chains amplify failure by degrading the brief at each handover. The remedy: named builders, compliance written as acceptance criteria, and superintendent sign-off gates at architecture, not at launch.
- A pharmacy site has five audiences — patients, the GPhC, advertising law, the payments layer and future diligence. Default builds serve one.
- The protective brief runs to ten clauses with acceptance tests. Sign-off gates and ownership-and-access carry the greatest weight, and writing it down is what sorts the market.
- Generic agencies suit brochure sites, specialists earn their premium as regulated features accumulate, and in-house teams need the same brief. The variable was always the specification, not the label.
- Seven questions test any builder in one meeting — including this site's publisher, which is as it should be.
Sources and further reading
FAQs
Built to the brief, by the brief's authors.
Our publisher builds pharmacy websites to exactly the brief above — catalogue classification, compliant flows, the data map, underwriting-ready pages — with Dataforge PMR behind the clinical services. Send us the brief from this article and hold us to every clause. That is the point of publishing it. See how it works.
See Dataforge PMR