Peptide Store Ecommerce SEO: How Pruning and Ten Content Clusters Reached 1,146 Monthly Clicks

The clearest signal in this engagement arrived in month 5, and it looked like bad news. Sessions had climbed to 180, their best figure so far, while average position went backwards from 29.3 to 32.1 and clicks stayed flat at 177. Two facts pointing in opposite directions usually means the site is competing with itself. It was: several collection variants and a handful of thin promotional URLs were all bidding for the same commercial intent, and Google was rotating between them.

So we stopped publishing at pace and spent a month deleting, merging and redirecting. From that point, average position improved every single month through to 16 in month 12, and clicks went from 146 in month 1 to 1,146.

The client runs a peptide ecommerce store selling nationally into a healthcare-adjacent market. When we took over, the site averaged position 33.6 across its tracked query set, pulled 24,264 impressions and 146 clicks in a month, and converted one order at a $90 average order value. Sessions were 131. In other words, the store existed in the index but sat below the fold of the second page for almost everything that mattered commercially.

Authority was thin but not absent: domain rating 14, 26 referring domains, 172 total backlinks. That profile matters, because it ruled out any plan that depended on outranking established competitors by link volume alone. Whatever we built had to earn relevance rather than borrow strength.

Three constraints shaped every decision. The store faced entrenched competition on its core commercial terms. Content production was capped at roughly six to nine articles a month, so we could not brute-force coverage. And the category is one where careless wording creates real risk, which meant no outcome claims, no implied medical benefit, and no invented trust signals of any kind.

The diagnosis split into three problems, only one of which was a content problem.

• Intent collision at template level. The store had generated multiple collection permutations for the same product set, plus separate promo and sale URLs. Sixteen tracked commercial and transactional queries were spread across those variants. None of them accumulated enough signal to hold a position, so rankings drifted week to week.
• No informational floor. Topical coverage was shallow. Buyers in this category research handling, purity documentation, storage and comparison before they spend, and the store answered almost none of it. Our internal coverage index, which tracks the share of a mapped cluster question set where the site has a ranking URL, started at 23 out of 100.
• Trust signals that could not be improved by copywriting. The reviews-modifier query sat at position 28 and stayed there all year. That SERP was owned by third-party aggregators, and no amount of on-site work moves a query where users are explicitly asking for an outside opinion.

There was also a commercial trap. The obvious first move in an ecommerce brief is to write more product and category copy. Our crawl said that would make the collision worse. The site did not need more commercial pages; it needed fewer, plus a body of informational content underneath them.

The first six weeks were spent on crawl and indexation triage rather than writing. We pulled the full URL inventory, sorted it by indexed status and impression share, and found crawl budget draining into faceted and parameterised collection variants that had never earned a click. Canonicals were inconsistent across those variants, several internal links pointed to redirect chains, and a template-level duplication issue repeated the same descriptive block across every collection view. We fixed canonical logic, cleaned the redirect graph, resolved the template duplication, and verified render and status codes on the priority paths. None of that produced traffic in month 1. It produced a site where a later change could be attributed to something.

Alongside the technical pass, we built the intent map. Every tracked query was checked against the live SERP to establish what page type actually ranked, then assigned to exactly one URL. The result was blunt: twelve of the sixteen commercial and transactional queries belonged on a single collection page, and the informational research query belonged on a supporting guide. That map became the constraint everything else obeyed.

The consolidation, and then the pivot

In month 3 we executed the merge: the strongest collection URL was rewritten to carry the full commercial intent (product structure, comparison block, sourcing and documentation section, shipping and handling detail, FAQ), and the variants were redirected into it. Internal links were rerouted so the money page sat two clicks from the homepage on every path, with anchor text distributed across the intent set rather than repeating one phrase.

Month 5 forced the second decision. Position regressed even as sessions grew, and the query-level data showed remaining thin URLs (promo, sale, subscription landing pages with almost no unique content) still surfacing intermittently for commercial terms. We pruned them, redirected what had equity, and cut publication to four articles in month 6 so the team could finish reconciling internal links instead of adding new ones on top of an unstable structure. That is the pivot: we stopped producing content volume for a month and reinforced the page that converts.

The informational library that did the heavy lifting

Once the commercial layer was singular and stable, the content system carried the campaign. We planned 10 topic clusters and shipped 89 articles across the twelve months, at a pace that respected the production cap: research fundamentals and terminology, purity and third-party testing documentation, storage, reconstitution and handling, dosing units and measurement, product category explainers, laboratory equipment and accessories, shipping, cold chain and import considerations, regulatory and research-use status, published safety literature summaries, and buyer comparison guidance.

Every article was written to answer one question set completely and to link upward into the money page or the supporting guide using the anchor language of the intent it served. That is the mechanism: informational depth across clusters earns the site relevance for thousands of research queries, and the internal link graph channels that accumulated relevance into the pages that sell. By month 12 the library was ranking for 4,224 informational keywords, and the coverage index had moved from 23 to 36.

Entity clarity and claim control

Two supporting workstreams ran in parallel. We cleaned organisation and product schema, aligned author and reviewer entities so the people behind the content were identifiable, and standardised citations of the store's name and details across the places it appeared. We also added short, answer-ready summary blocks at the top of cluster articles, written so a machine reading the page can extract a complete answer without inferring anything the page does not state.

The editorial guardrail mattered more here than in most niches. The store’s already-cleared product and policy copy became the editorial baseline. Reviewers converted those examples into an allowed-claims checklist, then stopped any draft that implied a treatment outcome. Several articles came back for rewrites. That cost us a week or two of throughput and removed the category's largest downside risk.

Link work stayed modest and started in month 4: recovering links that had broken or been dropped, cleaning inaccurate citations, and pitching genuinely useful resource pages, with the recovered placements landing around month 7. Referring domains moved 26 to 69 and domain rating 14 to 25 over the year. No month showed a spike, because we declined placements that failed the quality threshold.

Months 1 to 3: foundation, no visible reward

Clicks moved from 146 to 142 and impressions barely shifted. Average position improved slightly, 33.6 to 31.8, mostly from the indexation cleanup removing weak URLs from consideration. The client saw a flat quarter while we spent it on canonicals, redirects, template duplication, the intent map and the first 19 articles. We were explicit in advance that this phase would look like nothing.

Months 4 to 6: first lift, then the correction

The consolidated money page went live at the end of month 3 and month 4 responded: impressions jumped to 30,064 and clicks to 180, with average position at 29.3. Month 5 then regressed on position while sessions rose, which triggered the prune. Month 6 was deliberately quiet on output (four new articles) and closed at position 29.6 with 183 clicks. Two months of near-flat clicks bought a structure that could compound.

Months 7 to 8: the content layer starts feeding the money page

This is where the mechanism became visible. Clicks rose to 232 then 259, CTR ticked from 0.6% to 0.7%, and average position dropped to 26.2. The cluster articles published in months 2 to 6 had matured enough to rank and to pass internal link equity upward. Recovered links landed in the same window, reinforcing rather than causing the movement.

Months 9 to 10: compounding

Impressions went 48,853 then 60,234, clicks 391 then 482, sessions 353 then 468. Orders moved from two a month to three and then four. The library passed 72 articles and the coverage index reached 34. Position broke into the low twenties, which is the range where a commercial query starts producing real click volume in this category.

Months 11 to 12: the money page reaches page one for several intents

Month 11 delivered 86,971 impressions and 783 clicks; month 12 closed at 104,204 impressions, 1,146 clicks, 969 sessions, 9 orders and average position 16. The premium, ingredient, shipping and alternatives-style intents all landed inside the top ten. Part of month 11 and 12 growth coincides with the category's stronger buying season, which we address in the limitations.

The opening month is the honest baseline: 24,264 impressions, 146 clicks, a 0.6% click-through rate and average position 33.6. Almost nothing was on page one, and the impressions that existed were being distributed across collection variants rather than accumulating on one URL.

Month 12 shows 104,204 impressions and 1,146 clicks at a 1.1% click-through rate with average position 16. The merchant is anonymized; the displayed measurements preserve the scale used to evaluate this twelve-month program.

• Clicks: 146 to 1,146 per month, with the inflection at month 7 rather than at launch.
• Impressions: 24,264 to 104,204, driven mainly by the informational library entering the index across 10 clusters.
• Average position: 33.6 to 16, improving every month from month 6 onward after the prune.
• Click-through rate: 0.6% to 1.1%. The gain is a position effect plus better title and summary alignment with the query intent, not a formatting trick.
• Sessions: 131 to 969 per month.
• Orders and revenue: 1 order and $90 in month 1, 9 orders and $810 in month 12, at a $90 average order value. Across the full twelve months the site recorded 33 orders and $2,970 in modelled revenue.
• Authority: domain rating 14 to 25, referring domains 26 to 69, total backlinks 172 to 300.

The number we watch most closely is the ratio between the two ends of the funnel. Impressions grew roughly 4.3x while clicks grew roughly 7.8x. Traffic did not just get bigger; the queries the site appeared for matched what it actually sells more closely than they did in month 1.

Bullets replace identifying category words in the table. Search demand, intent labels and both ranking observations remain unchanged.

The pattern is consistent. Queries where a buyer wants specification, sourcing or comparison detail moved hardest: the alternatives-style query reached position 4, and the ingredient, shipping and premium modifiers all landed inside the top ten. Those are exactly the intents the informational clusters answered in depth, and exactly the intents the rewritten collection page absorbed from the redirected variants.

The regressions are real and mostly self-inflicted. Deal-led intents (the sale modifier and the subscription query) fell from 48 to 65 and from 41 to 80 because we deleted the thin promo and subscription landing pages that had been ranking weakly, and we chose not to rebuild them. Rebuilding would have reintroduced the collision we had just resolved, for query volumes of 210 and 140. The price-modifier query slipping from 35 to 43 was less deliberate: the merged page carries pricing inside a broader commercial context, and competitors running dedicated price comparison pages hold that SERP shape better than we do. It remains on the open list.

The reviews query at 28 never moved. That result did not surprise us and we stopped spending on it after month 4.

The third-party view tracks the same shape: visibility and estimated organic traffic curve upward from around month 7, which is when the first two content quarters had matured and the internal link graph had settled. The informational footprint reached 4,224 ranking keywords by month 12 across the 10 clusters, and our coverage index rose from 23 to 36. That library is the asset; the commercial positions in the table above are its output.

Monthly orders went from 1 to 9 and monthly modelled revenue from $90 to $810 at a $90 average order value, with 33 orders and $2,970 across the year. For a store of this size the absolute figures are modest, and we would rather state them plainly than dress them up. The important part is the shape: seven of those nine month-12 orders came in the final two months of the engagement, on a traffic base that had roughly quadrupled its impressions and grown clicks nearly eightfold.

Order conversion rate is the honest counterweight. Month 1 converted 1 order from 131 sessions; month 12 converted 9 from 969. That is a near-identical rate, which tells you the gain came from more qualified sessions rather than from a checkout improvement (we did not touch checkout). It also tells you something about the traffic mix: a large share of the new sessions arrive on informational articles, and those visitors buy later or not at all in the same session.

Which raises the fair question of whether informational traffic is worth funding for an ecommerce store. In this case, yes, for three connected reasons. First, the clusters that gained the most traffic (purity and testing documentation, storage and handling, comparison guidance) are read by people at the point of choosing a supplier, and they feed the money page through in-content links, which is visible in the movement of the comparison, ingredient and alternatives intents. Second, the library is the reason the collection page could reach position 16 on a head term at domain rating 25; a store with 26 referring domains and no topical footprint does not get there. Third, the asset keeps working. Positions earned by depth do not switch off when a budget cycle ends, which is the structural difference between this and paid acquisition on the same terms.

There is a further upside we treat as plausible rather than measured. The cluster articles carry answer-ready summary blocks, explicit entities, and claims tied to cited page evidence, which is the format AI assistants and AI search surfaces can quote without inference. The pages are extractable and the entity signals no longer conflict. Citation monitoring was outside this engagement, so the case assigns no outcome to assistant visibility.

The trust work has a measurable business edge too: because the editorial checklist blocked outcome-implying language before publication, the store's expanded footprint grew without adding regulatory exposure. In a category where a single unsupported sentence can cost a payment processor relationship, that is part of the return.

Nine orders in a month is a small sample. A single unusual week can move that figure by 20% or more, so we treat the month-11 and month-12 conversion counts as directional, not as a stable rate. Any forecast built off them should carry a wide band.

Seasonality is unresolved. Months 11 and 12 delivered the sharpest jumps in impressions and clicks, and this category buys more heavily in that part of the year. Our view is that the position curve (improving steadily from month 6, well before the seasonal window) carries most of the explanation, but we cannot separate the two effects cleanly with twelve months of data and no prior-year baseline of comparable quality.

Attribution runs on last-click sessions and a fixed $90 average order value. That understates the informational library, because a visitor who reads a storage or purity article in week one and buys through a branded search in week four is credited elsewhere. It also means we cannot tell you which specific cluster produced which order.

Three query problems remain open. The reviews-modifier intent has not moved from 28 and probably will not without third-party review volume the site does not control. The price-modifier intent regressed and our current page structure is the likely reason, which means the fix is a structural decision we have not yet made. The subscription query at 80 is the clearest cost of the prune, and if the client relaunches a subscription offer that page needs to be rebuilt properly rather than patched.

The research guide intent is also noisier than the table suggests. It moved from 22 to 12, but it swung across a wide range week to week all year, which is normal for an informational SERP with mixed result types. Reading a single week of that query would give a misleading picture in either direction.

Finally, domain rating and the ranking-keyword count come from a third-party index, not from the search console view. They are useful for tracking direction and poor for precision.

Nothing in this campaign worked in isolation, and the order was the point.

Indexation cleanup made attribution possible. Fixing canonicals, redirect chains and template duplication in months 1 to 3 removed the low-value URLs that were absorbing crawl attention. That change produced no traffic on its own, but without it, every later movement would have been indistinguishable from technical noise.

Consolidation converted scattered signal into one accumulating page. Twelve commercial and transactional intents were mapped to a single collection URL, the variants were redirected, and internal paths were shortened to two clicks. This is why average position began moving in one direction after month 6 rather than oscillating, and why the head term could climb from 31 to 16.

The informational library supplied the relevance the domain did not have. Eighty-nine articles across 10 clusters, ranking for 4,224 informational keywords by month 12, with the coverage index rising from 23 to 36. Each cluster answered a distinct stage of the research path and linked upward with anchor text matching the commercial intent it supported. Volume mattered, but so did depth: shallow coverage of ten clusters would not have earned rankings on the research queries, and without those rankings there would have been no equity to pass. Content was the cause; the top-ten positions on the specification and comparison intents were the effect.

Pruning protected the mechanism. Removing the thin promo, sale and subscription URLs in month 5 stopped weak pages from intercepting commercial impressions. We paid for that with two declining queries and gained a money page that stopped competing with its own site.

Entity and schema clarity converted position into clicks. Clean organisation and product markup, identifiable authors, consistent citations and answer-ready summaries lifted click-through from 0.6% to 1.1%: the same or better positions now presented an unambiguous result.

Links reinforced rather than led. Recovered and earned placements moved referring domains from 26 to 69 and domain rating from 14 to 25 at a steady monthly pace. That growth made the top-ten results on higher-difficulty commercial terms defensible; it did not create them.

Qualified sessions became orders. Sessions grew 131 to 969 at a roughly unchanged conversion rate, which is how monthly revenue moved from $90 to $810.

• Map intents to URLs before writing a word. The single most valuable artefact of month 1 was the sheet showing twelve queries pointing at one page. It determined the merge, the anchor plan and what we refused to build.
• A flat quarter is an acceptable price for a clean structure. Clicks moved 146 to 142 across months 1 to 3. Setting that expectation in the kickoff, in writing, is what allowed the prune in month 5 to happen without panic.
• Slowing production can be the fastest available move. Cutting to four articles in month 6 finished the internal link reconciliation and preceded the strongest sustained run of the year.
• Choose which queries you are willing to lose. We traded a 210-volume sale query and a 140-volume subscription query for a stable money page, and we would make the same trade again.
• Depth across clusters, not word count in one place, is what carries a low-authority domain. At domain rating 25 the store held top-ten positions on commercial terms with difficulty scores in the 50s and 60s because 4,224 informational keywords sat underneath them.
• Some SERPs are not yours to win. The reviews query never moved, and recognising that in month 4 freed budget for clusters that did.