Marketplace card SEO and audits: the full methodology for keyword research, diagnostics and visibility growth

Open any 2020-2022 course called "SEO for Wildberries" and you will find the same advice: collect 100 keywords from Wordstat, put them into the title and description, and the card will climb within a week. In 2026 that playbook works about 15% of the time. The marketplace will indeed index the card for the new words. But without purchases on those words, the algorithm will not believe the demand is real, and the card stays somewhere in the 60s.

Marketplace SEO is not about words in the description. It is about picking the right set of customer queries where your product has viable economics and which the marketplace will be happy to show more often. Words in the description are a necessary condition, not a sufficient one. If a keyword is missing from your SEO, there is no indexing at all and the product simply will not appear for it. But if the keyword is there and nobody buys through it, it is dead weight that dilutes the card's relevance.

This guide is a methodological breakdown of card SEO and audits. Within cluster D of our semantic core (~50 keywords about SEO and audits on WB, Ozon and Yandex Market), this is the Pillar. Topics:

• What proper marketplace SEO consists of in 2026.
• How to build a keyword set from real demand rather than from competitors' descriptions.
• What an MPO audit is and how it differs from a "general card analysis".
• What the 6 diagnostic layers reveal and where growth usually breaks.
• How to use the free forseller AI audit for a fast first diagnosis.
• The differences between SEO on Wildberries, Ozon and Yandex Market.

The Pillar on marketplace promotion is the parent article and we recommend reading it first: it explains the overall logic of visibility and the MPO system. The Pillar on buyouts and reviews is the neighboring one: SEO without buyouts locks in slowly, and buyouts without SEO land on the wrong keywords.

The marketplace algorithm decides to show a card more often for a given keyword not based on whether the word is in the description, but on whether people buy through that word. A word in the description is a signal that the card is relevant to the query. A purchase through that word is a signal that the card also converts.

The straightforward consequence:

• A card with correct SEO but no purchases on those keywords gets indexed but sits at positions 60-80. The algorithm is waiting for proof.
• A card without correct SEO but with purchases on keywords grows for a while on general weight, but the growth does not lock in for specific queries.
• A card with correct SEO plus purchases on those same keywords locks into the search results and keeps growing organically after support is switched off.

That is exactly the MPO strategy: SEO secures indexing for the right queries, buyouts provide the evidence of demand, and behavioral signals (CTR / add-to-cart / order pickup) confirm the quality. All three are mandatory.

This guide focuses on SEO, but you cannot ignore its link with behavioral signals and buyouts. It is a companion piece to Pillar C on buyouts, and it makes sense to read both together.

Full card SEO is not a single task of rewriting the description. It is a five-stage process:

1. 1Collecting real demand. Not words off the top of your head and not a copy of the top 3. Real search queries that customers use to look for the product.
2. 2Clustering by meaning and by search results. Grouping queries into clusters — low, mid and high frequency — with competition and economics in mind.
3. 3Removing keywords with poor economics and low conversion. Not all keywords are equally useful. Half of a typical "competitor keyword set" is dead weight.
4. 4Writing the SEO description for the platform's algorithm. WB, Ozon and Yandex Market rank by different rules, so the description is adapted to each.
5. 5Regular updates for seasonality and new queries. Queries shift in waves, so updating once a year is not enough.

Below we break down each stage.

Stage 1. Collecting real demand

The main mistake in marketplace SEO is collecting keywords from competitors. That is what most sellers do: they open the top 3 in the category, copy the words out of those descriptions and paste them into their own card. The logic looks sound: they are selling, so their keywords must be right.

The problem is that competitors get it wrong too. We regularly find words with zero real demand in top sellers' descriptions, word forms nobody searches for, and industry jargon instead of the plain wording customers actually type. Copy the mistake and, based on our audits, you lose 30-60% of the potential traffic.

The correct collection sequence:

• Collect real customer queries through specialist services (MPStats, Moneyplace, JET, or Wildberries own analytics via the Competitor Market report). Not Wordstat: those queries come from every platform, and on a marketplace they are often different.
• Check that products with your Subject ID actually appear for that keyword. If the query "cream" returns only shaving products while you sell a moisturizing face cream, the keyword is irrelevant and does not index what you need.
• Take not only full queries but also fragments. Customers type "face cream 50 ml", and the algorithm may rank you for "cream 50 ml" and for "face cream" as well.
• Account for slang and everyday wording. "Hair clipper" versus "trimmer" are different queries with different search volumes.
• Do not forget seasonal and trending queries. "School backpack" in August and in December are two different niches.

Stage 2. Clustering: low, mid and high frequency

You have collected 300+ keywords, so now you group them. The logic of the three frequency tiers:

High-frequency (HF). "Face cream", "serum", "yoga mat". Plenty of traffic, but maximum competition too. The top 10 for HF queries is held by players with millions of roubles in turnover. Getting in there is expensive.

Mid-frequency (MF). "Moisturizing face cream", "cork yoga mat". Less traffic, but far better targeted. Competition is moderate and the project economics add up.

Low-frequency (LF). "Moisturizing face cream with hyaluronic acid 50 ml", "cork yoga mat 6 mm non-slip". Each keyword brings 1-10 purchases a day. But any card has 100+ of them. Together that is a lot of traffic, and it is highly targeted: someone who types "6 mm non-slip" already knows what they want.

The core principle: you start pushing a card on LF, move on to MF, and only then go for HF. That is exactly how the marketplace algorithm works: first it tests you on safe low-frequency queries, and if conversion is good it expands to mid-frequency, and if that also works it finishes with high-frequency. This is the "test on small traffic" pattern, the natural logic of ranking.

A seller who tries to break straight into HF hits a ceiling: ads are expensive, buyouts do not lock in, and organic traffic never arrives. A seller who starts with LF has 50-80 purchases on precise keywords within two weeks, the algorithm sees that the card converts, and it starts expanding the reach on its own.

Stage 3. Removing keywords with poor economics

Not all keywords are equally useful. After collection you need to filter out:

• Keywords with zero real demand. The word is in the description, but nobody searches for it. Pure dead weight.
• Keywords with low intermediate conversion. People click through on that keyword but do not buy. They generate clicks without revenue, which lowers the card's overall CTR and its standing in the algorithm.
• Keywords with high promotion costs and low margin. To reach the top 10 for the HF keyword "cream" you need 300 purchases a month. At a margin of 100 ₽ per unit the economics do not work, so you are better off on MF.
• Keywords that return a different Subject ID. If you sell face cream but the keyword "cream" on WB mostly returns hand cream, it will not work for face cream.
• Words with different endings and word forms. For the algorithm, singular and plural forms of the same noun are often identical. There is no need to duplicate them.

Two to four weeks after the card is published, check each keyword in analytics: how many impressions it brings and what its conversion is. Keep the ones with high intermediate conversion, remove the low ones. This is called iterative SEO, and without it the keyword set loses effectiveness over time.

Stage 4. Writing the SEO description for the platform

This is the technical part, and it has plenty of traps. The main rule: the description is read both by the algorithm and by the customer. The marketplace punishes anyone who solves only one of those problems: an over-optimized description written for the algorithm looks like spam and lowers your ranking, while a description written purely for humans without keywords does not get indexed.

Baseline rules (universal across all platforms):

• Every priority keyword appears in the description at least once. Without that there is no indexing.
• Keywords are distributed across the text rather than dumped into the first paragraph. The algorithm understands context.
• Natural language is mandatory. "Buy serum for face moisturizing Moscow cheap" is spam. The card drops out of ranking.
• A unique description for every SKU. Both platforms penalize copied descriptions across cards.
• Structure: what it is / who it is for / how to use it / composition / specifications. The customer is looking for answers to these questions, and each block lifts conversion.

Two days after the updated SEO is published, the card is reindexed. If the keywords are right, the first impressions on the new queries start coming in.

Stage 5. Regular updates

"Do it once and forget it" is a common mistake. Customer queries shift in waves:

• Seasonal shifts. Summer means "moisturizing cream", autumn means "nourishing cream", winter means "cream for dry skin". If SEO is not updated, our observations show you lose 20-40% of seasonal traffic.
• New query formats. In 2025 a wave of "K-Beauty" queries arrived, in 2026 it was "slugging". Whoever added them first took the traffic.
• Trends and events. Releases, scandals and films create spikes on new queries. Reacting within a week is a big advantage.
• Search results evolve. The marketplace periodically changes its ranking rules. What worked in Q1 may stop working in Q3.

A healthy SEO update cadence is once every 2-4 weeks. For seasonal categories, once every 2 weeks at peak season. For stable non-seasonal niches, once a month.

The keyword set and the principles are shared. The technical rules are not. Here is the short version by platform.

Wildberries SEO

What you need to know:

• Card structure. Title (60 characters) plus description (the limit depends on the category — usually up to 2000, maximum 5000) plus attributes. Every block is indexed.
• Subject ID decides everything. Wildberries determines the category automatically from the title and description. If the Subject ID is wrong, the card will not get traffic on relevant keywords. This is the first thing an audit checks.
• Attributes matter more than people think. How completely attributes are filled in affects ranking and how well you match the filters customers apply.
• The description is read "from the bottom" in mobile results. The last 200 characters sit under the cut. That space is better used for handling objections than for keywords.
• Duplicates in the description do not help. Wildberries indexes unique keywords; repeats are not counted.
• Updating too often looks suspicious. Change your SEO daily and the algorithm treats the card as unstable.

Queries from cluster D3 — "seo optimization of wildberries cards", "wb seo optimization", "seo for wb cards", "wildberries card optimization" and so on — are all about these specifics. A deep dive into this exact mechanic is in the Hub guide "Card SEO on Wildberries".

Ozon SEO

What is different:

• A longer description. Ozon allows more text, but that also means higher demands on structure. The algorithm rates long unstructured descriptions worse.
• Rich content. Ozon indexes not only the description text but also the text inside Rich Content blocks (text baked into an image only is not read by search, so move keywords into text fields). Well-built Rich Content gives you extra keywords.
• Deeper categories. Ozon's category tree is deeper than WB's. Based on our audits, a card in the wrong subcategory loses up to 50% of its traffic.
• Filters affect indexing. If an attribute that customers filter by is left empty, the card does not appear. This is a common gap.
• Ozon search is more precise. The marketplace understands synonyms and slang better, but it does not tolerate copy-paste from other cards.

Queries from cluster D2 — "seo for ozon", "seo optimization of ozon cards", "ozon product card seo" — are all about these specifics. Porting a WB keyword set to Ozon without adapting it is a common and expensive mistake. A deep dive into the Ozon mechanic is in the Hub guide "Card SEO on Ozon": indexing of Rich Content text, 5-7 category levels, attributes as filters.

Yandex Market SEO

The specifics:

• Product card versus seller offer. On Yandex Market, one product can be sold by several sellers. The base product card SEO is shared, while the seller offer has its own SEO.
• Integration with Yandex search. Part of the traffic arrives through general Yandex search. That means external SEO weight matters for Yandex Market: brand, mentions and reviews outside the platform.
• Price and seller rating are part of SEO. On Yandex Market these parameters affect ranking more than they do on WB.
• Fewer noisy keywords. Competition is lower and niche keywords deliver results faster.

Cluster D4 queries — "yandex market seo", "seo yandex market" — are short, but in the right niches and with the right strategy Yandex Market delivers solid ROI.

Most services offering "wb product card analysis" give you a list of metrics: positions, rating, review count, conversion. That is not an audit. That is a status dashboard. It leaves you unclear on what to do next.

An MPO audit is a diagnosis with hypotheses and priorities. What you get is not "here is the current state", but:

• A list of "keyword gaps" — clusters where competitors get impressions and you do not.
• A map of the card's problems — exactly where the funnel breaks between impression and order pickup.
• Hypotheses on why it is not growing, with priorities and reasoning.
• A 30/60/90-day plan — what to change in SEO, which clusters to push, where reputation work is needed, where ads are needed, where you need to recover after an OOS.
• A comparison with competitors — where they are stronger and where they are weaker.

A side-by-side table:

The structure of the diagnosis:

Layer 1. Keywords and visibility

• Which keywords the product is visible for and which it is not.
• Share of search results across priority clusters.
• "Keyword gaps" — clusters where competitors have impressions and you have zero.
• Position dynamics for priority queries: growing, flat or falling.

The main problem is often found in this layer: across our audits, 60-80% of customer queries are not indexed at all because the keywords are missing from the description. This same layer is checked automatically in the free forseller audit — more on that below.

Layer 2. Behavioral funnel

• CTR in search results versus the category CTR (how many % above or below).
• Click to add-to-cart conversion.
• Cart to order conversion.
• Order to pickup conversion.
• Share of adds to favorites.
• Time spent on the card.

If CTR is 30% below the category, the problem is the cover image and first slide: what belongs on your slides. If cart conversion is low but order conversion is high, the problem is the description or the price. Each hole in the funnel is diagnosed separately.

Layer 3. Reviews and rating

• Count and format (text / photo / video).
• Dynamics — is the flow growing organically.
• What competitors have on the category's first page.
• Themes in the negative reviews — what exactly people write.

Audits often find that 3-5 negative reviews on the first page cut conversion by 20-30%. Covering them with fresh, realistic reviews fixes the problem within 3-4 weeks.

Layer 4. Warehouses and logistics

• Which warehouses the competitor uses and which ones you use.
• The impact of OOS on positions over the last 30/90 days.
• Which regions lag on delivery speed.

This is the layer most "card analyses" ignore entirely, and it is one of the most powerful. On WB, a warehouse in Elektrostal versus Podolsk creates a 15-20% difference in visibility across Moscow and the surrounding region.

Layer 5. Ad economics

• Ad spend share (DRR) and CPM across all active campaigns.
• Clusters where CPM is high but organic traffic is close by (potential to shift into organic).
• Campaigns with bids above profitability — where the budget is being burned.
• The ratio of paid to organic sales.

This layer often surfaces 20-40% of the budget that can be redirected into buyouts or cut entirely without losing revenue.

Layer 6. Competitive landscape

• The top 10 in search results for each priority keyword.
• Their strengths: price, rating, reviews, visuals, Rich Content.
• Their weak spots — where you can squeeze in.
• Their position dynamics — who is climbing and who is sliding back.

Without understanding your competitors it is impossible to work out which cluster is cheapest to win. Sometimes a card "stuck at position 100" can reach the top 10 of one cluster in 2 weeks because the competitors there are weak, while in another cluster that same card will run into a wall of million-rouble turnovers.

Trusty offers a free AI tool for a fast initial diagnosis of a WB card. It is not a replacement for a full MPO audit, but it covers the first and second diagnostic layers automatically.

How to use it:

1. 1Open trustyone.pro/forseller or go straight to the Telegram bot @TrustyAIasist_bot.
2. 2Enter the WB item number. A single number.
3. 3Get the PDF in Telegram within 60-90 seconds.

What is inside the PDF:

• Trusty Score — the card scored across the 6 diagnostic layers (on a 100-point scale).
• Complaint clusters — what customers write in negative reviews, grouped by theme.
• AI photo audit — analysis of the cover image and infographics: what reads at a glance on the first slide and what does not.
• Niche benchmarks — where you stand relative to the top 10 competitors on each metric.
• A 30/60/90-day growth plan — specific actions with priorities.

Limitations:

• Wildberries only (Ozon and Yandex Market are in progress).
• Only what is visible from outside the card — no access to the ad account or warehouse data.
• 3 free reports per user (lifetime). After that, you connect with the team via @michael_trustyone.

This is the ideal first point of contact. If the report shows the card needs a deeper diagnosis, you move on to a full MPO audit with account access and a call.

"Keyword gaps" are queries where competitors get meaningful traffic while your card does not show up at all or sits beyond position 50. Finding and closing them is the fastest route to growth, because:

• Demand is already proven by a competitor — you do not have to create a market.
• Competition is already known — you can see how much effort it will take.
• Queries can be ranked by potential (volume x conversion x your economics).

The search algorithm:

1. 1Collect every keyword the top 10 competitors in the category use.
2. 2Strip out keywords where you already rank well — there is no room to grow there.
3. 3Filter by economics — keep the ones where the margin works.
4. 4Rank by potential — where competitor revenue and reach are higher.
5. 5Group into clusters — which ones are related and can be pushed together.

The output is a list of 10-30 clusters with priorities. That is the "growth map" an MPO project works from.

The step-by-step methodology for finding keyword gaps is in a separate Support article, "Keyword gaps: a step-by-step methodology": 4 sources of gaps, 7 manual search steps, and the formula for potential revenue gain.

Trusty offers three audit formats for different needs and budgets.

Mini SKU audit. Free or discounted for leads. 1-3 cards. Baseline metrics: positions, share of search results, obvious "keyword gaps", 2-3 hypotheses. The ideal first touch — afterwards it is clear whether it makes sense to go further.

Standard MPO audit. Fixed price, agreed based on scope. Full analysis of the card and the category. A prioritized problem map, hypotheses with reasoning, a 30/60/90-day plan, and a presentation on a call. The output is a document you can show your team, your director or your agency.

Growth map / Enterprise audit. An audit of a 10-100+ SKU range. Comparative analysis against competitors, a strategy by product line and direction, and several calls with the client's team. Suited to large brands and agencies with big portfolios.

Any format can be extended with:

• Keyword selection for buyouts (a separate free service).
• A free card review during a 30-minute consultation.
• A free AI audit through forseller — for a fast first diagnosis.

Before launching any campaign, run through this list:

• [ ] Subject ID is correct. The card sits in the right category. On WB this is the most common gap.
• [ ] ALL attributes are filled in. Not just the mandatory ones — every attribute the category offers.
• [ ] Keywords come from real demand, not from competitors' descriptions.
• [ ] Clustering is done. You know which of your keywords are LF, which are MF and which are HF.
• [ ] Keywords with poor economics and zero demand have been removed.
• [ ] Every priority keyword appears in the title or description at least once.
• [ ] The description reads like a description, not like a keyword list. Natural language.
• [ ] A unique description for every SKU. No copy-paste between cards in the same product line.
• [ ] The description structure handles customer objections (what it is / who it is for / how to use it / composition / specifications).
• [ ] The platform is accounted for. SEO for WB / Ozon / Yandex Market means different texts, not one for all.
• [ ] An SEO update schedule is in place — once every 2-4 weeks.
• [ ] Position monitoring is running. Daily dynamics for every priority keyword.

If even half the points are open, those are your growth opportunities. Close SEO first, then launch the behavioral layer and buyouts.

"What is a card audit on Wildberries / Ozon?" It is a diagnosis that shows why the card is not growing and what to change first. It differs from a "general analysis" in that it gives you prioritized hypotheses and an action plan rather than a list of metrics.

"How much does a card audit cost?" A mini SKU audit is free or discounted as a first touch. A standard MPO audit has a fixed price calculated by scope. A growth map for a product line is quoted individually. Exact figures come on the consultation call.

"How does SEO optimization on WB differ from Ozon?" Different card structure, different category depth, different indexing rules (Ozon indexes text in Rich Content blocks, WB does not), and different description lengths. Porting a keyword set between platforms without adapting it is a common mistake.

"Can I order SEO optimization only, without buyouts?" You can. But without the behavioral layer, SEO delivers a limited effect: the words will appear in the description, but positions will climb slowly. At a minimum we recommend keyword selection before the start and launching buyouts 2-3 weeks after the updated SEO goes live.

"How long until SEO takes effect?" Indexing takes 2 days. First impressions on new keywords take 7-14 days. Positions lock in over 30-90 days (faster with behavioral support and buyouts).

"What if several contractors have already done our SEO — what should we change?" In that case an MPO audit is especially useful, because it shows exactly what you inherited from previous contractors. It often turns out that 50-70% of the keywords are dead weight or keywords with poor economics.

"How often should SEO be updated?" A healthy cadence is once every 2-4 weeks. At peak season, once every 2 weeks. For stable non-seasonal categories, once a month. Less often than that and you lose up to 40% of the traffic from new keywords.

"Do you do SEO for Yandex Market?" Yes. The logic there is different: the product card is shared while the seller offer is separate. SEO for Yandex Market is a separate piece of work, not a port from WB.

"What is Trusty Score?" It is a card score across the 6 diagnostic layers (on a 100-point scale) generated by our forseller AI audit. It is not a replacement for a full MPO audit, but it gives you a fast first picture.

"Are the AI audit and the MPO audit the same thing?" No. The forseller AI audit is a free automated diagnosis based on what is visible from outside the card (SEO, photos, reviews, niche benchmarks). A full MPO audit includes analysis of the ad account, warehouses and conversion history, and is run by an expert with account access.

"Can I get an AI audit for an Ozon card?" Right now forseller works with WB only. Ozon support is in progress. For an Ozon card, the standard MPO audit is the option for now.

"What about regulated categories — medical products, supplements?" We do run SEO audits there, but with allowances for vocabulary restrictions (some keywords are banned by the marketplaces). A full MPO project is discussed individually.

"Is an audit of a 50+ SKU range realistic?" Yes, that is the "growth map" format. We analyze every SKU, rank them by potential, and sort them into "rockets" (with plenty of headroom), "middlers" and "anchors" (no potential). It helps you trim the range and concentrate the budget.

"Who does the audit — AI or a human?" The forseller mini audit is AI. The standard MPO audit is an expert working with AI tools. The growth map is a team of 2-3 experts plus AI for data collection.

"What about confidentiality? Our business is sensitive." An NDA at the start if needed. Client data is not used in public case studies without consent. Account access is granted with the minimum necessary permissions.

"What if after the audit we decide to work with another contractor?" No problem. The document is yours, and you can use it however you like, including handing it to other contractors.

A few cases where we honestly say "not now" or "not us":

• A card with 100% visibility on the keywords it needs. SEO will not make it more visible. Growth means moving into other clusters, which is part of a full MPO project rather than a standalone audit.
• A recent audit already exists. If you had an audit 1-2 months ago and need a second opinion on a specific question, a direct consultation beats a repeat full audit.
• The card is only just being published. There is nothing to audit on an empty card with no data — go straight to a full launch plan instead.
• Systemic problems with the product. Defects, a bad batch, poor quality — SEO will not rescue that. Fix the source first.
• Fully regulated categories — some niches with special requirements are discussed individually.

• "Marketplace promotion: a systematic guide" — the parent Pillar on the MPO system. It explains how SEO fits into the 8-module approach.
• "Buyouts and reviews: the full methodology" — the neighboring Pillar. SEO plus buyouts is the pairing without which neither part locks in.
• "Micro-differences, macro results" — on how visibility and behavioral signals work. The foundation everything else rests on.
• "Buyouts 2.0" — a detailed breakdown of keyword collection, cleaning keywords by conversion, and the "plateau" pattern.
• "Safe buyouts" — on marketplace detection markers and the effective cost of a review.

If you have an active Wildberries card and want a fast diagnosis, enter the item number at trustyone.pro/forseller or in @TrustyAIasist_bot. A PDF with your Trusty Score, complaint clusters and a growth plan arrives in Telegram within a minute. Free — 3 reports per user.

If your task is bigger — a range of 10+ SKUs, a brand relaunch, entering new categories — get in touch for a consultation. We will show you what full-service card SEO optimization and a complete MPO audit with account access look like, and which growth opportunities will be the fastest in your case.