Irina Maltseva

Brand mentions only build trust when they appear on relevant, authoritative sites, because a private blog network placement reads as spam to a language model, and one genuine answer in a high-signal Reddit thread outperforms dozens of mentions that get flagged and removed.

• GEO without SEO foundations is described as guesswork, and anyone claiming otherwise is described as selling something.
• A related claim: listicles on a company's own site are losing visibility while listicles hosted on third-party sites are gaining it.

After auditing what is sold under the GEO umbrella, Maltseva describes an outreach automation startup placing brand mentions on private blog networks and charging roughly fifteen times the real cost, and a Reddit agency buying aged accounts and mass-posting mentions for every client from the same account, most of which are removed within thirty days. The closing line is that the bar for calling yourself a specialist has never been lower.

Seen’s launch post frames the shift from the client side: SEO once meant optimising for Google and measuring rankings. That is no longer enough when customers also search on ChatGPT, Gemini, Reddit and review sites, while most companies keep running old playbooks.

Read original · Original 2

The recommended moves: open with a direct definition, put the strongest claim in the first 20 to 30 percent of the page, turn headings into real user questions and answer them immediately, use named entities instead of generic categories, replace hedging with definitive statements, and close with a real summary rather than a footer.

• Balanced sentiment is preferred over hype or dryness.
• These rules come from Kevin Indig's analysis of 18,000-plus verified citations across 1.2 million ChatGPT responses, which Maltseva summarises rather than conducts; the cited findings include 44% of citations coming from the first 30% of a page and roughly 20% proper nouns in cited content.

An intro that starts “In today's fast-paced digital landscape…” has already lost. “X is Y” beats “In this article, we'll explore…”, and within a paragraph, AI most often cites the middle sentence—the one carrying the most complete idea.

Read original

That exact sentence already exists in the client's Gong transcripts, which is why Maltseva asks new clients for their sales calls rather than Google Search Console access. The underlying point is that marketing already has the relevant information and is not listening to it.

The example: a buyer types “what can I use to book meetings for my cybersecurity SaaS without hiring another SDR” — a query with zero volume in Ahrefs that appeared in the previous week's call transcripts.

Read original

Those pages are now declining because readers and algorithms both recognise invented expertise, and because the pattern becomes impossible to miss when produced at scale or programmatically. Maltseva argues very little original, differentiated, first-hand B2B content exists and almost all of it is ghostwritten for search or AI visibility.

The example is a set of “I tested the X software” guides on G2, a pattern Maltseva credits Gaetano Nino DiNardi for spotting. One author claims to have tested the eight best review management tools, while her bio describes her as a content marketer specialising in ERP content.

Read original

After checking the partner's site for relevance, the founder makes the introduction and Maltseva pitches a reciprocal article exchange, each side writing a relevant piece for the other, aimed at the angles where AI-answer visibility is wanted. The claim is that roughly nine out of ten warm pitches turn into published articles, against far worse conversion from cold outreach.

The example: a sales automation tool with integration partners in the CRM niche — the same audience, similar content and a genuine openness to collaboration.

Read original · Original 2

An audit is about 5% of the job; execution is the remaining 95%.

• A company without an in-house team or the resources to act on recommendations is a bad fit for both sides.
• The post’s closing idea: knowing who isn’t a fit attracts more of the clients who are.

Maltseva rejected a $12,000 project that would have taken two weeks because it covered paid search, paid social and demand generation alongside organic search, and was a one-off audit.

Maltseva knows paid search, paid social and demand generation, still uses those channels to grow Seen, and has managed marketing teams at SaaS companies. The change is that they no longer take client work outside SEO and GEO, not that those skills have gone away.

Read original

The agency homepage was rebuilt around live outputs from ChatGPT, Claude, Google and Perplexity for the commercial searches clients want to be visible for, replacing logos, testimonials and stock illustration, with an invitation to check the same queries in your own AI tool.

• The stated logic is that different clients in different niches appearing where it matters says more than a long case study.
• The same instinct shows up in what Maltseva says to founders who ask for a guaranteed ChatGPT ranking within three months.

It took four months to settle on the hero section. The concept — real AI answers for money-driving searches instead of a title, subtitle, call to action and random illustration — led to a full site rebuild.

Read original · Original 2

Newcomers should start here.

Maltseva announces Seen, an AI SEO agency for SaaS companies, and traces the path to its launch: more than ten years in SaaS SEO, working in-house, advising companies and serving as a fractional CMO across martech and cybersecurity.

• They argue that optimizing for Google alone is no longer enough: customers now search on ChatGPT, Gemini, Reddit and review sites, while most companies stick to old playbooks.
• Seen offers embedded work with the client's team across strategy, content and authority, measuring success by pipeline.
• They personally build the strategy, while a small partner team executes and scales it—a contrast with agencies that hand clients off after onboarding.

For brands planning a campaign

For a SaaS buyer, this post sets the engagement shape: embedded, retainer-style work, with strategy owned by the principal and success measured by pipeline rather than rankings. Brands expecting a standalone deliverable or a junior team should note the mismatch before starting a conversation.

The advice is specific enough to act on the same day.

Maltseva summarizes Kevin Indig's analysis of 18,000-plus verified citations across 1.2 million ChatGPT responses and turns it into writing rules.

• The findings: 44% of citations come from the first 30% of a page; attention is top-heavy, with a drop-off and a small recovery at the summary; definitive statements ('X is Y') outperform exploratory phrasing; question-style headings outperform abstract ones because models treat headings like user prompts; cited content contains roughly 20% proper nouns, making named products more effective than generic categories; balanced sentiment beats hype or dryness; and the cited sentence within a paragraph is often the middle one, which carries the most complete idea.
• Recommended actions: lead with a direct definition, put the strongest claim in the first 20-30%, turn headings into real questions and answer them immediately, add concrete entities, cut hedging and close with a real summary.

For brands planning a campaign

This is a practical content brief checklist for teams creating pages designed to surface in AI answers.

• It also shows the standard Maltseva would apply to a partner's drafts.
• Brands whose editorial style relies on a narrative build-up may need to restructure.

Maltseva describes turning down a $12,000 engagement that would have taken two weeks, citing two rules.

• First, they stick to SEO and GEO: the client wanted an audit covering paid search, paid social and demand generation.
• They say they can do that work and have done it while running marketing teams at SaaS companies, but no longer take it on.
• Second, they don't take one-time projects.
• An audit is 5% of the job, they argue, while execution is the other 95%; without an in-house team or resources to execute, the engagement is a bad fit for both sides.
• Their conclusion: knowing who isn't a fit attracts better clients.

For brands planning a campaign

This post shows a clear mismatch for briefs covering a one-off audit or a broad multi-channel review. Bring a plan for who will execute the recommendations: Maltseva says that is a condition for accepting work.

It's the best single window into how Maltseva approaches keyword work now.

Maltseva argues that keyword tools miss real demand: a buyer asking ChatGPT, 'what can I use to book meetings for my cybersecurity SaaS without hiring another SDR' shows zero volume in Ahrefs but appears verbatim in a company's Gong call transcripts. The post summarizes an LLM Mastery podcast episode with Jabez Reuben and covers several topics: turning down a paid third-party mentions project from an enterprise client with a good budget; Aura taking almost two years to reach number one for 'identity theft protection' and why they would still start with the hardest page; why 90 out of 100 Reddit mentions sold by agencies get deleted within a month; why listicles on a company's own site are losing visibility while those on third-party sites gain it; and what they tell founders who ask for a guaranteed ChatGPT ranking in three months.

For brands planning a campaign

The transcript-first method is a concrete research deliverable brands can request or replicate: mine sales calls for buyer phrasing before building a content plan. The post also makes clear that Maltseva won't guarantee AI rankings, a useful expectation to set before procurement.

Maltseva recommends asking founders to name recent connections at other SaaS businesses in the same space but not in competition, such as integration partners in an adjacent category with the same audience.

• After checking a partner's site for relevance, the founder makes an introduction and Maltseva pitches an article exchange: each side writes a relevant piece for the other's site, targeting the angles where visibility in AI answers is wanted.
• They contrast this with cold outreach and claim nine out of ten warm pitches result in published articles.

For brands planning a campaign

This is a low-cost route to third-party 'best tools for X' listicles, relying on a founder's network rather than budget. Brands with strong partner integrations but no founder available to make introductions will need a different approach.

Maltseva warns against the 'hand-crafted buyer's guide' playbook that followed Google's shift toward demonstrated first-hand experience.

• The example, credited to Gaetano Nino DiNardi for spotting it, is a set of 'I tested the X software' guides that ticked every box: first-person language, helpful-content signals, an unbiased tone and links to data studies.
• They describe those pages as tanking because readers and algorithms both recognize manufactured authenticity.
• Maltseva argues that little original, differentiated, first-hand B2B content exists; almost all of it is ghostwritten for SEO or AI visibility, a pattern that becomes impossible to miss when produced at scale or programmatically.
• Their point: following every best practice can still fail when the environment shifts.

For brands planning a campaign

Briefs calling for high-volume, review-style content in a first-person voice run straight into this objection. The stronger approach is to publish fewer pages based on genuine hands-on testing, with the author's actual work behind them.

After auditing what is sold under the GEO umbrella, Maltseva lists three examples: an outreach automation startup placing brand mentions on private blog networks and charging clients roughly fifteen times the real cost for links that hurt them; a Reddit agency buying aged accounts and mass-posting mentions for every client from the same account, paid per mention, most removed within thirty days; and an agency founder with no SEO background teaching the industry that SEO is dead.

• The counter-position: brand mentions only work on relevant, authoritative sites, because a private blog network mention signals spam to a language model; one genuine answer in a high-signal Reddit thread beats fifty mentions that get flagged; and GEO without SEO foundations is guesswork.
• The closing line is that the bar for calling yourself a specialist has never been lower.

For brands planning a campaign

Send this to procurement teams vetting AI search vendors: it also states which methods Maltseva won't use. Brands already running PBN or mass Reddit placement should expect the relationship to begin with that disagreement.

Maltseva spent four months finding a hero concept for Seen’s website, rejecting the standard title, subtitle, call to action and stock illustration.

• The solution was to drop logos and testimonials and instead show actual results from ChatGPT, Claude, Google and Perplexity for the commercial searches clients want to be visible for, with an invitation to verify them directly in the tools.
• The argument: seeing different clients in different niches appear where it matters says more than a long case study.

For brands planning a campaign

A brand commissioning visibility work can ask for this proof format: screenshots of real AI answers for named queries, rather than a written case study. It also sets a bar for the brand’s own marketing pages.

They launched Seen, an AI SEO agency for SaaS companies, after 10+ years of in-house SaaS SEO, advisory work and fractional CMO roles. They describe the model as embedding with client teams to build visibility through strategy, content and authority, measured by pipeline rather than rankings.

For brands planning a campaign

For a B2B SaaS brand seeking citations in AI answers, Maltseva is a direct fit for a strategy-and-execution engagement, not a one-off audit. Expect classic SEO groundwork as a prerequisite for LLM visibility work.

Examples: Source 1 · Source 2 · Source 3

The argument: AI rewards extractability over dwell time, and high-visibility content reads like a structured briefing.

Extractability-first writing: answer-first structure, question headings, entity density and definitive language.

The citation figures come from Kevin Indig's study, which Maltseva summarises.

For brands planning a campaign

A brand could commission a content audit or rewrite against these rules, or use them to brief internal writers and freelancers without further research.

Examples: Source 1

Their reasoning: keyword research turns up high-volume generic terms, while a buyer's specific ChatGPT query may show zero volume in Ahrefs—even though the same sentence appeared in the client's Gong transcripts the previous week.

Mining sales-call transcripts for target queries instead of starting with keyword-tool volume.

For brands planning a campaign

A partnership should plan to open up sales conversations, not just analytics dashboards, and build targeting from buyer language rather than keyword lists.

Examples: Source 1

They then pitch an article exchange, with each side writing a relevant piece for the other. Maltseva claims roughly nine out of ten warm pitches lead to published articles, and argues third-party listicles are gaining visibility while listicles on a brand's own site are losing it.

Map the founder's network, check the partner site's relevance, then arrange a reciprocal article exchange targeting the angles where LLM visibility is wanted.

Warm introductions drive the tactic; Maltseva says cold outreach to the same tools converts far worse.

For brands planning a campaign

This is a low-cost route into "best tools for X" roundups, but it depends on the founder's network and genuine relevance. Brands with a strong partner ecosystem can supply most of the raw material.

Examples: Source 1 · Source 2

Their counterarguments: mentions work only on relevant, authoritative sites; one genuine answer in a high-signal Reddit thread beats dozens of removed spam posts; and GEO without SEO foundations is guesswork.

Maltseva sells GEO and SEO services, so their criticism targets specific practices, not the category.

For brands planning a campaign

Useful as a second opinion or due-diligence voice on existing GEO or LLM-visibility spend. Brands hiring Maltseva should expect blunt public commentary on work they consider low quality.

Examples: Source 1 · Source 2

They say those pages are now declining because readers and algorithms recognise the invented expertise, and because almost all B2B first-hand-experience content is ghostwritten at scale.

For brands planning a campaign

Content claiming a brand tested a product when it did not carries growing risk. Campaigns based on documented, real product use—with an accountable author—fit this position better.

Examples: Source 1

They take on work only within their expertise and when clients have the in-house capacity to execute, arguing that an audit accounts for about 5% of the job and execution the remaining 95%.

For brands planning a campaign

Expect a retainer-style engagement with execution attached, not a standalone audit. Maltseva may decline brands without an internal team to act on recommendations.

Examples: Source 1

They argue that a result visitors can check in their own AI tool says more than a long case study. The shift behind the rebuild—search across Google, AI assistants, Reddit and review sites—is the same one the agency sells against.

For brands planning a campaign

Brands can test Maltseva’s claim cheaply by running the cited searches in their own AI tools. The approach favors demonstrable visibility over narrative case studies in co-marketing assets.

Examples: Source 1 · Source 2

Later posts speak from inside the business: rejecting a $12K audit because it fell outside SEO and GEO and was a one-time project, rebuilding the Seen website to show live ChatGPT, Claude, Google, and Perplexity results for client searches, and describing how the team gets clients into third-party "best tools for X" listicles. The posts treat SEO and GEO as the agency's whole scope and pipeline as its success measure.

View example

The post shows commercial work connected to Aura; ask whether it was an agency client, an in-house role, or advisory work.

View example

View example

Dyulgerova treats branded search — people typing a brand name alongside a keyword — as the trust signal search engines reward. That overlaps with Maltseva's position that LLM visibility rests on reputation built across many platforms, not on one page.

Dyulgerova's posts are short arguments drawn from work across dozens of SaaS clients.

• They often focus on team behaviour, such as how slow review cycles stall SEO.
• Maltseva writes from inside the agency they founded, with longer tactical playbooks and pointed criticism of named vendors.

Read original · Original 2

Nisar frames AI as a reputation engine that rewards clear category positioning, credible proof and specific outcomes — close to Maltseva's argument that generic category language leaves a company invisible in AI answers.

Nisar's worked examples focus on ecommerce product pages, with frameworks such as entity-attribute-value, intent mapping and schema, sometimes packaged in a detailed PDF collected through comments. Maltseva works on B2B SaaS, sells strategy and execution together, and is more likely to say what a buyer should not purchase.

Read original · Original 2

Senapati's method — find threads Google already ranks, read the thread, then contribute one genuinely useful comment and let upvotes bring people to your profile — aligns with Maltseva's claim that one real answer beats mass-posted mentions that get removed.

Senapati tests ideas with small businesses and ecommerce, publishing a two-week case study with lead and purchase counts and inviting peers to compare their own observations. Maltseva writes from a SaaS agency perspective and takes firm positions on which widely sold practices are worthless.

Read original · Original 2

Their argument is that branded search—people typing the company name alongside a category or competitor term—is the trust signal that precedes organic growth, and that it is built by everything outside SEO: word of mouth, Reddit threads, YouTube comparisons, podcasts, newsletters and LinkedIn posts from a founder.

• For a B2B SaaS company whose pages rank reasonably but whose pipeline is flat, and whose sales calls are not recorded or whose founder will not make partner introductions, that diagnostic frame is closer to the problem than a page-restructuring engagement.
• Dyulgerova also writes directly about the organisational cause of stalled SEO: across more than 50 SaaS businesses, the pattern is slow execution—ten departments reviewing one blog post, months of planning, a single post taking six months to go live.
• A company that suspects its real constraint is internal approval speed will recognise that description immediately, and it is a useful outside opinion to put in front of a founder who keeps reopening decisions.
• The tradeoff: these posts offer diagnosis, not delivery.
• Dyulgerova describes what moves rankings and what slows teams down, but the posts do not describe an agency, a retainer, an embedding arrangement or a scope of work.
• Ask what execution support is available before hiring on this basis; the posts make a case for analysis, not an execution partner.
• There is also no AI-answer-specific method here—no query set, extractability rules or measurement of ChatGPT or Perplexity output.
• If the goal is to be named inside AI answers and someone has to rewrite the pages, this is a complementary voice rather than a replacement.

Examples: Source 1 · Source 2

The posts frame AI visibility as a positioning and data-structure problem rather than a writing problem: the question to answer is why an AI would recommend this company over its competitors, and the answer comes from solution specificity, consistent category terminology and complete product data.

• The worked example is ecommerce—a query like “best coffee machines under $300 with a grinder that fits a small kitchen” rather than “best coffee machine”—with Amazon and Reddit surfacing because their data and use cases are specific, while generic 500-word feature pages stay invisible.
• The second post sets out a Product Page Optimization framework covering entity-attribute-value structure, query-intent mapping, schema, review and user-generated content strategy, internal linking, product variants, discontinued products and conversational commerce readiness.
• For a SaaS or ecommerce company whose catalogue is large, inconsistent or full of near-duplicate pages, that checklist addresses a real gap that Maltseva’s buyer-language and page-structure work assumes is already in order.
• The tradeoff: the centre of gravity is ecommerce and product catalogues, not a B2B SaaS retainer.
• The framework comes as a visual mind map and a PDF released on request through a comment—a lead-generation format rather than an engagement.
• The posts show no embedding model, sales-call research or partner-introduction play.
• Ask whether the work can extend to a multi-month, in-team strategy engagement; otherwise, treat it as specialist catalogue and entity work to be scoped separately.

Examples: Source 1 · Source 2

The case study describes a two-week project with a small jewellery business in Malaysia: after relatively simple changes to how the site’s information was structured and connected, Senapati observed 10 leads, 5 purchases and further conversations, with some customers saying they found the business through ChatGPT.

• The second post is a Reddit playbook that costs nothing but time: use site:reddit.com searches to find threads Google already ranks, read the thread before commenting, write one genuinely useful answer rather than a pitch, and let profile visits and inbound questions follow.
• For a founder-led business or a small team that cannot commit to a multi-month retainer, and for whom Reddit is an untapped channel where competitors are absent, this is a low-cost way to test whether conversational discovery produces qualified enquiries before spending on a full programme.
• The measurement instinct is also worth copying: Senapati explicitly asks peers whether they are tracking AI-generated traffic and whether it behaves differently from search traffic, the right question to settle early.
• The tradeoff: the evidence base is thin and self-reported.
• One two-week case study in a consumer retail category, with no control and no independent verification, does not establish what will happen for a B2B SaaS company with a long sales cycle.
• The post ends by asking other practitioners to compare observations, so it reads as an experiment shared for discussion, not a proven method.
• The Reddit approach also depends on patience and genuine contribution, which is exactly what most brands abandon; it does not scale instantly.
• A company that needs a strategy owner, a query set built from sales calls and reciprocal article placements should stay with a retainer engagement and use these tactics as additions.

Examples: Source 1 · Source 2