Methodology, version 1.0

Purpose

This methodology exists to make every conclusion on the site reproducible: given the same evidence records, a reader should reach the same numbers and the same qualified findings. It governs what gets captured, how prices are normalized, when a claim may publish, how conflicts and staleness are handled, and how commercial relationships are controlled. Where a rule here conflicts with making any provider look good, the rule wins.

Scope

The methodology covers compounded semaglutide and compounded tirzepatide programs sold by U.S. telehealth providers, in injectable, ODT, sublingual, and oral forms, across standard-dose and marketed "microdose" plans. Manufacturer and brand-access channels (LillyDirect, NovoCare, Amazon Pharmacy, savings-card and insurance pathways) are profiled as a separate class and are never ranked against compounded programs. Clinical-evidence pages follow the research-review rules summarized here and detailed on each research page. Nothing in scope constitutes medical, legal, or financial advice.

Definitions

Advertised price is the price a provider displays publicly for a plan before checkout, captured with date and source. Effective monthly cost is total mandatory cost for the treatment period divided by verified months of treatment supplied. Mandatory cost includes medication, required membership, required consultations, provider fees, laboratory fees, onboarding fees, mandatory shipping, dose surcharges, required refill charges, required supplies, and any required advance payment.

First-month cost is everything required before or during month one, including one-time fees. Renewal cost is the recurring price after any introductory pricing ends. Commitment-adjusted cost presents the total contractual commitment, the amount due today, the number and frequency of payments, medication shipped per shipment, and treatment months represented.

Page vocabulary is fixed: "cheapest" pages analyze advertised or upfront price; "most affordable" pages analyze verified normalized all-inclusive cost; "best value" pages additionally weigh dose coverage, clinical support, pharmacy transparency, commitment, renewal pricing, cancellation terms, and evidence quality. The three are different questions and are never merged into one list.

Entities and identity

Records distinguish four identities that marketing usually blurs: the consumer-facing brand, the legal business entity behind it, the medical group employing or contracting the prescribing clinicians, and the dispensing pharmacy. A price belongs to a brand's program; a license belongs to a pharmacy or clinician; a lawsuit or warning letter belongs to a legal entity. Conclusions are attached to the correct identity, and where the corporate structure is undisclosed — common in telehealth — the gap is recorded as an evidence gap on the provider's page rather than papered over with the brand name.

Verification statuses and confidence

VERIFIED first-party checkout or document capture, archived. PROVIDER REPORTED taken from official provider pages, unconfirmed at checkout. CALCULATED arithmetic we performed from captured figures, shown to the cent. CONFLICTING sources disagree, or a claim contradicts itself — suppressed from rankings and flagged on-page. PENDING known entity, capture incomplete. Additional ledger statuses: Estimated, Not Publicly Disclosed, Stale, Historical, No Longer Current, Not Applicable. Confidence (High/Moderate/Low) reflects source quality and internal consistency, and is recorded per record, not per provider.

Comparison population

All cost conclusions are computed for a generally available cash-pay adult consumer in the United States: no insurance offsets, no employer subsidies, no eligibility-restricted savings cards, and no assumption of clinical suitability. This population choice is what makes figures comparable across providers — and it is also a limitation, because an individual reader with coverage, a subsidy, or a restrictive medical history will face different real numbers. Category pages restate this population so that no conclusion travels without it, and insured or subsidized pathways are analyzed on their own pages under their own assumptions rather than blended into cash-pay rankings.

Evidence standards

Sources are used in a fixed hierarchy: FDA, NIH, PubMed, ClinicalTrials.gov, CMS, state pharmacy and medical boards, court records, the Federal Register, official drug labels, peer-reviewed journals, manufacturer medical and pricing pages, official provider pages, checkout evidence, provider terms, official pharmacy pages, archived first-party captures, reputable news organizations, and secondary editorial sources — in that order. Another comparison or affiliate site is never the primary source for pricing, pharmacy status, availability, credentials, cancellation terms, FDA status, warning letters, or ownership. Authoritative identifiers (DOI, PMID, NCT, NPI, license, FEI, warning-letter and recall numbers) are stored when available and are never printed from memory.

Core inference rules: unknown never equals zero — unknown shipping is not free shipping, an unknown membership fee is not no fee, unknown availability is not nationwide, and an undisclosed pharmacy is not transparent disclosure. Provider Reported is not Verified. An archived price is not a current price. A starting dose is not an all-dose price; a starter offer is not renewal pricing; a coupon price is not generally available pricing. A monthly equivalent does not mean monthly billing, and a multi-month commitment does not mean monthly shipments. A 503B registration does not mean any product is approved. Evidence for an FDA-approved product does not automatically support a compounded formulation, and injectable evidence does not support ODT, sublingual, oral, or troche formulations.

Worked examples of the formula

Bundled provider, prepaid plan: NexLife's captured 12-month standard semaglutide plan is $1,740 with no reported membership, shipping, or surcharge, so effective monthly cost = ($1,740 + $0 + $0 + $0) ÷ 12 = $145.00. The same provider's month-to-month plan computes $165 ÷ 1 = $165.00 — both publish, labeled by billing structure, because a prepaid equivalent and a monthly bill are different consumer propositions even when the arithmetic looks similar.

Membership-model provider: Mochi Health's captured medication price is $99/month with a required $79/month membership (resolved to its current first-party value), so the effective cost is $178.00 — the $39 advertised first month is a promotion that renews at the full rate. The two examples show the method's point: identical-looking headlines produce different true costs, and the formula makes the difference visible instead of hiding it in a headline.

Rounding is display-only. A $1,764 twelve-month prepaid plan computes $1,764 ÷ 12 = $147.00 exactly; where a division isn't exact, we publish the calculated figure to the cent and show a provider's rounded number only when labeled as rounded or provider-advertised. Precision matters less for a fraction of a dollar than for the habit it enforces: numbers on this site are reproducible from their inputs, or they don't publish. (Figures currently under recapture, such as certain NexLife microdose totals, are suppressed until re-verified rather than displayed as settled.)

Why the exclusions are excluded

Coupons, influencer codes, and referral credits are excluded because they are not generally available: a comparison population can't reliably obtain them, so including them makes the ranked price a lottery ticket. FSA/HSA advantages are excluded because they depend on individual tax situations, not the program. Conditional rebates and hypothetical cancellation refunds are excluded because they pay out only under conditions the buyer can't count on at decision time. Financing is excluded because it changes payment timing, not cost. Insurance savings are excluded from cash-pay comparisons because coverage varies per person; insured pathways are analyzed separately. Each exclusion trades a lower-looking number for a number more people will actually pay — the correct trade for a comparison site.

Research-review standards

Clinical-evidence pages separate FDA-approved-product evidence from compounded-formulation evidence, and randomized trials from observational, real-world, provider-sponsored, post-hoc, and preclinical work. Every study summary must state design, population, sample size, intervention, comparator, duration, primary endpoint, results, adverse events, funding, conflicts, and — always — applicability limits to compounded products. Provider-sponsored studies additionally disclose author affiliations, employment, randomization and control status, attrition, and data-source limits. Identifiers (DOI, PMID, NCT) publish only after confirmation, never from memory. Investigational agents (retatrutide, cagrilintide, CagriSema, orforglipron and peers) receive research-only coverage: no purchase, provider, ranking, compounding, or dosing content, and no suggestion of lawful compounded availability. Summaries remain noindexed until a named medical reviewer signs them.

Eligibility rules

A program is ranking-eligible for a verified category only when it has current Verified pricing (including all mandatory fees), a confirmed billing and shipment structure, a defined dose range, and no unresolved Conflicting Evidence on a ranked field. A provider receives a full indexable review only when sufficient evidence exists; otherwise it appears in the directory as "Not yet fully reviewed" with noindex applied to thin profiles. Manufacturer channels are categorically ineligible for compounded rankings. Programs with lapsed recheck dates become Stale and drop out of eligibility until re-captured.

Calculation rules

Effective monthly cost = total mandatory cost ÷ verified months supplied, computed to the cent and rounded only for display, with rounded figures labeled as rounded. Excluded from the calculation, deliberately: financing benefits, FSA/HSA tax effects, conditional rebates, referral credits, limited coupons, influencer codes, insurance savings unavailable to the general comparison population, and hypothetical cancellation refunds. Comparisons never cross these separations without explicit labeling: semaglutide/tirzepatide, injectable/ODT, standard-dose/microdose, monthly/prepaid, brand/compounded, medication-only/all-inclusive, starter/maintenance, introductory/renewal, provider-reported/verified, and differing dose ranges, quantities, or pharmacy formulations.

Ranking rules

No verified category winner publishes with fewer than three eligible programs holding sufficiently current Verified evidence. Every published affordability conclusion must state, in one passage: medication, dosage form, route, dose range, program type, plan duration, payment schedule, comparison sample and count, verification date, evidence status, methodology version, material exclusions, and material relationship. Advertised-price ("cheapest") findings may publish earlier with those same qualifications and an explicit advertised-not-verified label. Any provider — NexLife included — loses a category the moment another eligible program shows a lower verified normalized cost, and expired rankings are withdrawn, never silently retained.

Provider scoring

The 100-point model weights safety and clinical oversight at 25, total cost and price stability 20, pharmacy and medication-source transparency 20, program transparency 15, consumer policies and protections 10, and patient support and accessibility 10. Brand popularity, search volume, advertising spend, affiliate revenue, and conversion performance are never scored. Missing evidence reduces the available denominator or yields "Not yet scored"; missing information never defaults in a provider's favor. Every published score displays its denominator, excluded fields, evidence cutoff, verification status, methodology version, relationship disclosure, and unresolved gaps.

Conflicting and stale evidence

When sources disagree, or a provider claim contradicts its own figures, the field is stored as Conflicting Evidence, suppressed from rankings, disclosed on affected pages, and queued for first-party capture — the semaglutide-ODT savings claim on the NexLife review is the live example. Records carry recheck dates by freshness class: event-triggered for enforcement, recalls, and closures; weekly for prices and promotions; monthly for licenses, availability, fees, and terms; quarterly for full reviews and this methodology. Lapsed records are flagged Stale automatically and excluded from conclusions until refreshed.

Relationship and compensation controls

Every provider record carries a material-relationship value from a fixed taxonomy (none known, affiliate, referral, advertising, shared ownership, common beneficial owner, financial interest, management, editorial control, reviewer affiliation, data sharing, multiple, or verification pending), currently a dated publisher attestation of no relationships. If any compensated relationship begins, affected records update before the next publication cycle, compensated links are labeled, and compensation never influences rankings, scores, ordering, or editorial conclusions. The site uses no fake urgency, countdowns, stock levels, preselected providers, or compensation-based ordering.

Data pipeline and audit trail

Facts flow one way: source → ledger record (with URL, capture date and time, status, confidence, recheck date) → database → rendered page. Page templates cannot introduce a commercial figure that lacks a record, and superseded records are retained as history rather than overwritten, so any published number can be traced to what was captured, when, from where, and under which methodology version. Public JSON and CSV exports of the program database, pharmacy database, and ledger ship with every build so third parties can audit the arithmetic without trusting the prose.

Quality assurance

Every build runs automated checks: internal-link resolution, suppressed-claim leak detection, sitewide required-statement presence, noindex/sitemap consistency, and structured-data/visible-content match. Editorial gates before indexing follow the page-quality standard: the primary query answered near the top, at least two forms of original information gain, primary sourcing, verified/provider-reported separation, current verification dates, stated limitations, relationship disclosure, and no unsupported superiority claims. A shorter page that passes beats a longer page that repeats.

How ranking categories are constructed

A category is a tuple, not a vibe: medication × dosage form × program type × cost basis (advertised or verified) × plan-duration class. "Lowest verified standard tirzepatide injection cost" and "lowest verified tirzepatide microdose cost" are different categories with different eligible populations, and a program competes only in categories whose tuple it matches exactly. Category pages must name the tuple, the eligible-program count, and the excluded programs with reasons. New categories are created only when at least three programs could plausibly become eligible — a category built to have one entrant is a press release, not a comparison — and duplicative keyword-variant categories (cheap/cheapest/budget) are prohibited unless the underlying question genuinely differs.

Limits of the formula itself

Effective monthly cost deliberately measures money, not value: it cannot see monitoring quality, dose-range breadth, pharmacy quality, or cancellation fairness, which is why those live in the scoring model and the "best value" definition instead. It also assumes the months-supplied figure is real — the reason shipment cadence and beyond-use dating are verification fields, since twelve billed months served by medication that expires in eight is not a twelve-month supply. Where months supplied cannot be verified, the calculation publishes with that caveat or not at all.

Corrections

Errors are corrected visibly on the affected page with the date and nature of the change; publication dates are never refreshed cosmetically, and corrected conclusions are recomputed from the corrected records rather than hand-edited. Readers can report errors through the corrections route on the About page; substantiated reports enter the ledger like any other evidence, with the reporter's claim distinguished from what we independently verify.

Methodology changes

Material changes version this document and are logged with dates and reasons; conclusions derived under a prior version are recomputed or labeled with the version that produced them. Corrections to published errors are made visibly on the affected page with the change date — publication dates are never altered to appear fresh.

Limitations

Honest limits of the current dataset: most pricing is provider-reported rather than checkout-verified; secondary-source figures conflict for several providers; renewal terms, cancellation mechanics, and state availability are largely uncaptured; pharmacy license verification is pending; and no medical reviewer has yet signed the research summaries. These limits are stated on every affected page, drive the "no verified winner" posture, and define the verification backlog in priority order. Figures can change between our capture and your checkout; the checkout price is always the authoritative one.