Peptide Clinic SEO: From Competing Service Pages to 363 Monthly Clicks in Three Months

The clinic came to us wanting more articles. We delivered fewer than the plan called for, deleted several pages that were already live, and the average position moved from 33.5 to 14 in three months.

That is not a contrarian pose. When we pulled the query-to-URL map in week one, four separate pages were surfacing for the same commercial intent, and none of them ranked above position 40 for it. Publishing into that structure would have added more competitors for the client's own pages. So the first month was mostly subtraction, the second was architecture, and only in the third did the informational library start doing the work it was built for: pushing relevance and link equity into two pages that actually take bookings.

The client operates a peptide therapy practice in a single metropolitan service area, generating patients through consultation requests rather than ecommerce transactions. Before the engagement, non-branded organic traffic was uneven month to month, average position across the tracked set sat around 34, and topical coverage was thin enough that competitors with generic health-directory content were outranking a practice with real clinical staff.

Month one baseline: 18,407 impressions produced 129 clicks, a 0.7% click-through rate, and five booked consultations. The impressions were not the problem. The site was being shown often enough; it was being shown for the wrong slice of demand, at positions where nobody clicks, on pages that could not tell Google which one deserved the query.

Three constraints shaped everything we planned. Local competition included two well-funded regional groups with larger content teams. Content production was capped by the client's clinical review process, since every page needed sign-off from a practitioner before publication. And in a regulated health category, no claim could go live without a citable basis, which ruled out the outcome-promise copy the competitor set was using freely.

The crawl and the Search Console query export disagreed in a specific, diagnosable way. Google was rotating between a service page, a location page, a thin comparison page, and an older blog post for the same commercial cluster. Rotation is the signature of an unresolved intent conflict: the algorithm cannot decide which URL is the canonical answer, so it tests them, and each test resets whatever position equity the previous URL had accumulated.

Three consequences followed from that single structural fault:

• No page could stabilise. The commercial terms were parked between positions 23 and 51, which is the band where impressions accumulate and clicks do not. A 0.7% CTR against 18,407 impressions is the arithmetic of being visible on page three and page four.
• Internal links were diluted. Existing blog posts linked to whichever page the writer remembered, so anchor equity was split four ways instead of concentrated on the page meant to convert.
• The booking path was too long. Consultation-request intent resolved to a page that then required two more clicks to reach a form. The transactional query sat at position 40 with a page that technically existed for it.

Secondary to that, indexation was noisy: paginated archives, tag URLs, and parameterised filter pages were consuming crawl budget on a site with roughly 100 backlinks and a domain rating of 19. A site that small cannot afford to spend crawl attention on URLs nobody will ever search for.

The rule we worked to: fix what breaks the signal before amplifying the signal. That ordering decided the whole calendar.

Month one: triage, not content

We ran a full crawl against the indexed URL set and separated pages into keep, merge, and remove. Tag archives and filter parameters were excluded from indexing, canonicals were corrected on the templates that were duplicating meta content across service variants, and a batch of internal 302s and soft-404 responses were resolved so link equity stopped leaking through redirect chains. Renderability and Core Web Vitals were checked and found adequate, so we deliberately spent nothing further there; on a site this size, template duplication was the bigger constraint on ranking than a few hundred milliseconds of load time.

In parallel, we built the intent map: every tracked query assigned to exactly one owning URL, with the SERP composition recorded for each (local pack present or absent, directories dominating or not, informational results mixed into commercial queries). That map became the arbitration document for every later decision.

Months two and three: one owner per intent

We consolidated the commercial cluster into a single service page and kept the city page as the local entry point, then 301'd the thin comparison page and the stale blog post into the surviving service URL. The consultation flow was shortened: booking intent now lands on a page with the request form above the clinical detail, not below it.

Internal linking was rebuilt as hub and spoke. Each informational article links up to its cluster hub, and each hub links to the service page or the city page depending on whether the reader's question is clinical or logistical. Anchor text was distributed deliberately rather than repeated, mixing descriptive service phrasing with question-shaped anchors, because uniform exact-match anchors across a small site read as a pattern rather than as editorial linking.

The informational library that did the actual lifting

This is the part of the engagement we would defend hardest. Fourteen articles across six clusters were commissioned, structured so that each cluster covers a complete decision area rather than a scatter of keyword targets:

• Peptide therapy mechanisms and how the treatment category works
• Candidacy, screening and lab work before treatment
• Safety, sourcing standards and regulatory status
• Metabolic and weight-management applications
• Recovery, sleep and performance applications
• Cost, consultation logistics and what insurance does and does not cover

Five were live at the end of month one, eight by the end of month two, fourteen by month three. The topical authority index we track for this client moved 28, then 41, then 52. By the close of the quarter the library was ranking for 367 informational queries, most of them low-volume long-tail questions that individually mean very little.

Collectively they mean a great deal. Each article gives Google another piece of evidence that this domain covers the subject rather than the transaction, and each one is a contextual internal link into a money page from a page that is topically adjacent. That is the mechanism: informational depth creates the relevance and the internal equity, and the commercial pages convert that into position. The service page did not climb from 50 to 8 on a commercial term because we rewrote its headline. It climbed because six clusters of supporting material now sit beneath it and point at it.

Editorial guardrails in a regulated category

Before a word was drafted, our team sampled the client's approved patient-facing material to extract tone and, more importantly, claim boundaries. We built a reviewer checklist that blocked outcome language, comparative efficacy statements, and anything implying a guaranteed clinical result. Several drafted passages were rewritten at review stage for exactly this reason. This slowed publication and it was the correct trade: a health-category page that overstates is a page that invites both regulatory and quality-signal risk.

Entity and answer-surface work, deliberately last

Only in month three did we clean up Organization and Service schema, align the practitioner entities named on the pages with their citation footprint elsewhere, and add short answer-ready summary blocks at the top of the cluster hubs. Those blocks exist because assistant-style search surfaces extract compact, attributable statements, and because a reader scanning for the cost range should not have to read 900 words first. We sequenced this last because structured data describing a confused architecture just describes confusion accurately.

Links, in small numbers

Digital PR ran only in month three: recovering links that had broken during old site migrations, cleaning up inconsistent directory citations, and pitching two industry resource pages. Referring domains moved 22 to 28 to 32 and total backlinks 103 to 137. We rejected several placement opportunities that did not meet the quality threshold. On a domain at DR 19, a sudden link spike is more likely to attract scrutiny than to help.

Weeks 1 to 4: subtraction

Almost nothing visible shipped. We were excluding low-value URLs from the index, untangling redirect chains, correcting canonicals, and building the query-to-URL map. Five articles went live at the tail of the month, chosen from the two clusters where the SERPs showed informational results mixed into commercial queries (cost and candidacy), because those were the fastest routes to relevance that a money page could inherit. Traffic barely moved, which was expected. The average position needle also barely moved. What did change was crawl distribution: Google started requesting the pages we cared about more frequently once it stopped being offered hundreds of parameterised near-duplicates.

Weeks 5 to 8: architecture

Consolidation landed. The thin comparison page and the stale post were redirected into the service page, the city page was rewritten around the local entry intent, and the hub-and-spoke internal linking went in. Three more articles published. This is where the numbers first broke pattern in an instructive way: impressions rose only 5.6% while clicks rose 36% and average position improved by ten places. We were not being shown to more people. We were being shown in positions where being shown means something, and we had stopped splitting those impressions across four competing URLs.

Weeks 9 to 12: the pivot, then the payoff

The original plan queued a larger article batch for month three. We shipped six and stopped the pipeline. The reasoning came straight from the month-two data: pages published into the newly consolidated structure were gaining position faster than pages published in month one had, and the pages we had pruned had cost nothing to lose. Adding more surface area was clearly worth less at the margin than reinforcing what already ranked. So the remaining hours went into pruning two further weak pages, deepening the three articles that were already earning impressions on candidacy and cost questions, and adding the E-E-A-T proof blocks (practitioner credentials, review process, sourcing standards) to the surviving money pages.

Entity and schema cleanup ran alongside, and the link recovery work closed out the month. That combination produced the steepest movement of the engagement: average position dropped into the mid-teens, CTR doubled against the baseline, and booked consultations reached fourteen for the month.

Across three months, clicks moved from 129 to 363, impressions from 18,407 to 25,899, click-through rate from 0.7% to 1.4%, and average position from 33.5 to 14. Booked consultations went from five to fourteen per month.

The before view shows the pattern we diagnosed: a wide impression base against a click line that stays flat, which is what page-three and page-four visibility produces. Nothing about that chart says the site lacked demand. It says the site was never in the click zone for the demand it was already being shown for.

The after view shows the two lines converging in behaviour. Impressions grew 41% across the quarter while clicks grew 181%, and that ratio is the whole story: position improvement compounds against a fixed impression base far faster than impression growth does.

We anonymized the practice, its service area and its consultation revenue. The numbers preserve the scale and direction of the work without exposing those commercial details.

Where the data is untidy

Analytics sessions tracked Search Console clicks closely in month one (127 against 129) and month two (176 against 175), then diverged in month three: 293 sessions against 363 clicks. Two plausible causes, and we did not resolve them cleanly within the quarter. Consent-mode gaps suppress a portion of session recording, and the redirect wave from consolidation means some clicks landed on a redirected URL where session stitching is less reliable. We treat the Search Console figure as the upper bound and the analytics figure as the conservative floor, and we report conversions from the booking system rather than from either.

The tracked set is masked below: the client-identifying service term is replaced with the bullet token, and the query structure, intent, search volume and positions are the real ones. That keeps the pattern legible without publishing the client's exact competitive footprint.

Which intents won and why

The transactional pair moved most decisively. The appointment-shaped query went from 40 to 2 because a single page now owns booking intent and the form is the first thing on it; the consultation query improved from 23 to 15 on the same page. Nothing else explains a 38-place move on a term with difficulty 26 in twelve weeks, since no new links were built to that URL.

Pricing and fee intent performed better than we forecast. Two of the six clusters address cost and insurance directly, and the internal links from those articles land on the money page section that explains fee structure. The fee query reached position 8 and the affordability variant 16. Competitors in this category tend to hide pricing entirely, which leaves the SERP open to anyone willing to publish a defensible range.

The high-volume local query improved from 51 to 13 and the shorter local variant reached 8. Both sit behind a local pack, so position 13 in the organic listing understates and overstates the outcome at the same time: fewer clicks than a top-three organic placement in a pack-free SERP, but the pack presence itself was reinforced by the citation cleanup.

Which regressed, and the honest reasons

The superlative comparison term slid from 35 to 49. That query had been served by the thin comparison page we removed. Its intent skews toward third-party listicles and directories, and our consolidated service page is a first-party page, which is structurally the wrong result type for that SERP. We accepted the loss rather than rebuild a page whose only purpose was to argue that the client is the best, a claim our review process would have blocked anyway.

The immediacy-flavoured local variant fell from 36 to 47. The city page was rewritten around scheduled consultations because the practice does not accept walk-ins, which removed the open-hours signals the previous version carried. At 140 monthly searches with a demand profile the clinic cannot serve, ranking for it would have produced calls the front desk turns away.

Two commercial terms stayed volatile between 20 and 21 after climbing from the low 40s. Both SERPs mix directory aggregators with practice sites and reshuffle week to week; our tracked positions bounced by five or six places within single weeks. We are treating those as unresolved and expect them to settle only as the cluster library deepens.

The third-party visibility view captures the same trajectory from outside the client's own property, including the long tail the tracked set does not cover: 367 informational queries ranking by month three, up from a coverage base that was effectively negligible.

Monthly booked consultations went 5, then 8, then 14. Across the quarter that is 27 consultations and $10,260 in modelled value, with month three alone at $5,320 against a $1,900 baseline. For a single-location practice, nine additional consultations a month is a scheduling change, not a rounding error.

The composition of that traffic matters more than the count. Before the engagement, the clicks arriving were largely on the service page from mid-funnel commercial queries. By month three, a substantial share of sessions entered through the informational library, and those visitors behave differently: someone who reads about screening requirements or what lab work is needed before treatment arrives at the consultation form already partially qualified. They ask better questions on the call. Some do not convert this month and return in six weeks, which is the part of the value that no first-touch attribution model captures and the part that compounds.

That is the argument for building depth rather than buying clicks. The 14 articles were a one-time production cost with clinical review attached, and they will keep earning impressions across those 367 long-tail queries next quarter without further spend. A paid campaign contributes nothing after its budget is switched off. An informational library at cluster depth does the opposite: each additional article strengthens the internal link graph feeding the money pages, so the marginal cost of ranking the next commercial term falls over time.

Discovery beyond the ten blue links

We built the cluster hubs with compact, sourced summary blocks near the top, and we aligned the practitioner entities named on those pages with their citations elsewhere. The intent is that when someone asks an assistant what to expect from a consultation in this category, the material that gets extracted is attributable to a real clinic with named clinicians rather than to an aggregator. We can observe that the structured statements exist and that the entity references are now consistent. We cannot report a citation share, because no reliable measurement of that exists for a practice this size, and we did not promise the client one. Treat it as positioning we control, not a metric we measured.

The trust work carries its own return. Practitioner credentials, an explicit review process, and sourcing standards on the money pages address exactly the questions a cautious patient asks in a category with a reputation problem, which is as much a conversion-rate intervention as an E-E-A-T one.

• Three months is short. Two commercial terms were still oscillating between positions 20 and 21 at the close. We do not know yet whether they settle in the top ten or drift back, and anyone reading a twelve-week window as a stable trend is reading it wrong.
• Revenue is modelled, not reconciled. The figures apply a flat average new-patient value to booked consultations. We had no access to CRM close rates, no-show rates, or treatment-plan values, so a booked consultation that never became a patient counts the same as one that did.
• Sessions and clicks disagree in month three. 293 against 363. Consent-mode suppression and redirect-affected session stitching are the likely causes, but we did not isolate them within the engagement window.
• Seasonality is unaccounted for. A single quarter in a category with metabolic and weight-management demand cannot be separated from seasonal interest cycles. A repeat measurement across the same months next year would be needed to do that honestly.
• Two intentional losses are still losses. The superlative term at 49 and the immediacy variant at 47 represent roughly 620 combined monthly searches we chose not to compete for. A different practice, one accepting walk-ins, should make the opposite call on the second.
• Authority barely moved. DR 19 to 21 across the quarter. The link profile is not yet strong enough to defend these positions if a larger competitor targets the same cluster deliberately.
• The content cap was a constraint we worked around, not solved. Clinical review limited output to 14 articles. A team able to publish 40 with the same review rigour would likely have reached these positions faster and held them more securely.

Indexation cleanup came first because crawl attention on a DR 19 domain is finite. Removing tag archives and parameter URLs from the index concentrated crawl frequency on the pages that mattered, which is why later changes were picked up within days rather than weeks.

Consolidation resolved the intent conflict. When four URLs compete for one query, Google rotates them and none accumulates stable position history. Designating one owner per intent and redirecting the rest ended the rotation, and the redirected URLs' equity flowed into the survivor. This is the single largest contributor to the average position moving from 33.5 to 23.5 between months one and two.

Internal linking converted the informational library into ranking pressure on the commercial pages. Fourteen articles across six clusters, each linking up to a hub and each hub linking down to the service or city page, means the money pages now receive contextual links from pages that are topically adjacent rather than from a generic sidebar. Relevance passed through those links is the reason the commercial cluster moved into single digits and low teens on a link profile that grew by only ten referring domains.

Topical authority is the mechanism underneath all of it. The index moved 28 to 41 to 52 as cluster coverage went from partial to complete, and 367 informational queries now rank. Each of those queries is trivial alone. Together they constitute the evidence Google uses to decide that this domain is a subject authority rather than a landing page with a form, and that judgement is what allows a small site to outrank larger ones on commercial terms.

Position improvement then converted into clicks: CTR doubled from 0.7% to 1.4%, and because a top-ten result attracts a multiple of a position-thirty result's clicks, a 41% impression increase produced a 181% click increase. Shortening the booking path turned those clicks into consultations, taking conversions from 5 to 14 monthly at a broadly stable conversion rate, which tells us the traffic quality held as volume grew rather than diluting.

Links and schema entered last and did the least in the measured window. They matter for what happens after this quarter: authority reinforcement that makes the current positions harder to displace, and entity clarity that makes the practice legible to answer surfaces which increasingly summarise rather than list.

• Audit intent collisions before commissioning a single article. Publishing into a cannibalized architecture adds competitors for your own pages. The 40-to-2 movement on booking intent cost nothing but a redirect and a page rebuild.
• Watch the impression-to-click ratio, not impression growth. A 5.6% impression rise against a 36% click rise in month two was the clearest signal we had that the structural work was correct and that more publishing would have been the wrong spend.
• Cluster completeness beats article count. Six clusters covering entire decision areas moved the authority index from 28 to 52 with only 14 pieces. Fourteen scattered keyword-chasing posts would not have.
• Some rankings are worth losing. Dropping an immediacy-flavoured local query the clinic cannot serve removed a source of wasted front-desk time, and dropping a superlative comparison term avoided writing claims the review process would rightly have blocked.
• Sequence links last on small domains. Ten referring domains in a quarter contributed less than one consolidation decision. Spending the budget on links first would have amplified a broken structure.