Skip to content
nachomascort SEO · AI · Growth

Skills

Google Quality Audit

Audits your pages against Google's quality criteria: the four pillars from the Quality Rater Guidelines, the commodity content test and the quality signals from the leak. Give it a URL, a page type or a whole section and it returns a report with what fails and how to fix it.

The theory behind it is in the post Quality at Google: what it is, how it is measured and how long it takes to recover from a core update

  1. 01

    Scores the four pillars

    Effort, originality, talent or skill and accuracy, 0 to 4 each, always backed by evidence from the page.

  2. 02

    Runs the commodity test

    If you swap your brand for a competitor's and the page is still just as useful, it's commodity. It suggests an angle only you can cover.

  3. 03

    Looks for red flags

    Scaled content, filler, deceptive authorship or unreviewed AI, based on the spam policies and the guidelines.

  4. 04

    Zooms out to page type and site

    Which page types to improve, consolidate, move out or remove, and whether one is dragging the rest down.

  5. 05

    Maps it to the leak

    Every Google signal is labelled as documented, inference or hypothesis. No made-up weights.

  • Audit the quality of this URL: https://…
  • This is my city page template (pasting the text). Is it commodity content?
  • We dropped 20% in the last core update. These are our sections and how many pages each has. Where do I start?

I ran it on my own site twice

→

On 4 October I audited one page of each page type on nachomascort.com. I applied the recommendations that made sense to me, and on 5 October I ran the skill again, in a fresh session and without showing it the first report. The audited pages are in Spanish.

Page type4 Oct5 OctVerdict now
Home 11/16 11/16 Solid with gaps
Services 9/16 10/16 (+1) Solid with gaps
Case studies 9/16 10/16 (+1) Solid with gaps
Tools 11/16 11/16 Solid with gaps
New blog (2026) 12/16 13/16 (+1) Strong
SEO Hacks archive 5/16 6/16 (+1) At risk

What I changed between the two audits

  • SEO Hacks archive (in Spanish): I fixed advice that is wrong today (like blocking thin content with robots.txt), removed calls to action that no longer work and typos, and added a note to each post saying what still holds. See →
  • Softonic case study: years and job titles, "before ChatGPT" instead of "there were no commercial LLMs", and the core update and ccTLD story that used to live only in the post. See →
  • Sourced claims: the years of teaching add up and the media logos say "mentioned". See →
  • Consulting service: the former Google Search engineers who hired me are named (Cliqpod and Reverse Tech), the proof appears once, llms.txt is explained without selling it and the method is told through the Softonic example. See →
  • Quality post and bot checker: the Gary Illyes quote now links to its source (and I qualified it, because he said core updates work page by page) and the robots.txt 5xx case includes Google's timelines. See →

What I decided not to do

Both audits ask to take almost the whole SEO Hacks archive out of the index. I have not done it: they are my own posts from years ago, they still get links from seohacks.es, and I would rather fix them than hide them. That is why the archive is still "At risk". I also have not added charts to the case studies, because I have no data I can publish.

How to read the result

The increase is small, which is what you would expect: the fixes were about accuracy and cleanup, not new content. What really moves the score, according to the skill itself, is proof (charts, screenshots, original data) and new pieces that do not exist on another site.

Report from 5 October 2026

After this second report I fixed four details it pointed out: the "8 years" of teaching still on the training page, a boilerplate sentence in the service, "tarifas cerradas" versus "precio cerrado", and the llms.txt part of the post. It also has one error: it says the Google Search Central logo does not link to the mention, and it does.

Site verdict: Solid with gaps. The commercial pages and the new post hold up well in a core update because they rely on cases and data that only Nacho Mascort has. The risk is the SEO Hacks archive: 19 posts from 2015–2017, almost all commodity or outdated, which make up 44% of the Spanish URLs in the sitemap and 90% of the blog.

Method: I applied the Google Quality Audit skill (four pillars from 0 to 4, commodity test, red flags from the Search Quality Rater Guidelines of 11 Sept 2025 and signals from the May 2024 leak) to one page per page type. Source: HTML generated in dist/ (5 Oct 2026, version not yet published), read directly and judging only the main content, without navigation or forms. For the tool I reviewed the code of functions/api/bot-check.ts and tested the production endpoint (/api/bot-check) with nytimes.com and nachomascort.com. I ran the top 10 tests with web searches in the US; where I only saw titles and snippets I say so (“partly checked”). I have no traffic or Search Console data, so the recommendations on the archive do not take into account which posts bring visits.

Summary

Page typePage auditedScoreCommodity classVerdict
Home/11/16Non-commoditySolid with gaps
Services/servicios/consultoria-seo-geo/10/16Partly commoditySolid with gaps
Case studies/casos/softonic-0-a-20-millones/10/16Non-commoditySolid with gaps
Tools/herramientas/comprobador-bots/11/16Partly commoditySolid with gaps
New blog (2026)/blog/quality-google-core-updates/13/16Non-commodityStrong
SEO Hacks archive (2015–2017)/blog/que-es-thin-content/6/16CommodityAt risk

1. Home

Verdict: Solid with gaps (11/16, non-commodity). The page sells with its own names and figures, but all the proof ends in claims with no external evidence and several phrases are repeated almost word for word.

Purpose: introduce Nacho Mascort as an SEO, GEO and AI consultant and lead to a call or the form. YMYL: no (B2B marketing services; it does not affect health or personal finance).

PillarScoreEvidenceFix
Effort3/4Tailor-made content: four services, each with a concrete figure (“pasó de 200 a más de 3.000 creatividades por semana”, “contraté un equipo de 8 personas”), a teaching block with dates (UPF 2015–2024, Webpositer since 2018), a logo wall, a course with data (450+ students, 80+ videos).Add a piece that is hard to replicate: a testimonial with the name and title of a client or of a former Softonic/Reverse Tech colleague, or a real screenshot of a traffic chart.
Originality3/4Everything it offers is its own: Softonic 0 → 20M sessions, word.tips x2 with peaks of 2M visits/day, Reverse Tech 0 → $2M margin, savings of “unos 20.000 € al mes”. No competitor can copy this data.It would reach 4 with something others cite: a figure published here for the first time (e.g., the percentage of generated pages that survived the core updates at Softonic).
Talent or skill3/4Clear structure (who I am, why me, services, case studies, teaching, course, blog, contact), visible CTAs, figures at the top. Drawbacks: “Ex ingenieros de Google Search me contrataron” appears in the meta description, in “Por qué yo”, in the SEO service and again in the bio of every post; “Trabajas directamente conmigo, sin juniors ni subcontratas” appears twice. “Más de diez años enseñando” clashes with “lleva más de 8 años dando clase en másters” in the same scroll. The blog block shows only two posts (2026 and 2023), which exposes a three-year gap.Say the Google claim once and well (in “Por qué yo”), unify the years of teaching (since 2015 that is 11) and, until there are more new posts, show only the 2026 one or a tools block on the home page.
Accuracy2/4Specific figures, consistent across home, services and case studies, but none has a verifiable source: the case studies it links to have no charts, screenshots or external references either. “Me han mencionado: Google Search Central” does not link to the mention.Link each “Me han mencionado” logo to the specific piece and add one proof per main figure to the case studies (cropped Search Console or Analytics screenshot, link to a talk or to the company’s LinkedIn).

Commodity

  • Swap test: if you change the name for another consultant’s, the page loses the case studies, the logos, the teaching and the course. Passes.
  • Top 10 test: does not apply to a personal brand home page (the query is navigational, “Nacho Mascort”).
  • Only-you assets used: case studies with figures, client and media logos, teaching, own course, own tools.
  • Missing non-commodity angle: a quote from a named client with a result, or a real chart from one of the case studies in the “Resultados con nombre y cifra” block itself.

Red flags: none from the Lowest/Low lists. No ads or popups (clutterScore with no cause for concern). The “Company” field in the form is a hidden honeypot (aria-hidden), so it does not matter.


2. Services: SEO, AEO and GEO Consulting

Verdict: Solid with gaps (10/16, partly commodity). The social proof is unique, but the process, the deliverables and the FAQ are what any GEO consultant would publish.

Purpose: explain the SEO/GEO consulting service, filter the right client and lead to a call. YMYL: no.

PillarScoreEvidenceFix
Effort3/4Real fit and exclusion criteria (“Buscas SEO local para un único negocio. Ahí te irá mejor una agencia local”), own opinion on AI and SEO, the Softonic example (a catalog of 100,000 apps that covered a small part of demand), a “máximo de 4 clientes” limit, an honest warning about llms.txt.Add an anonymized example of a deliverable (a page from a real audit, a row from the 90-day roadmap).
Originality2/4What is its own: the case studies and the point of view. The four phases (diagnosis, fixes, content, measurement), the five deliverables and the FAQ “¿Qué diferencia hay entre SEO, AEO y GEO?” match what other GEO consultants publish (e.g., Alejandro Rioja’s article on what a GEO consultant does lists entity audit and structured data, citation baseline and a prioritized list).Replace the generic SEO/AEO/GEO FAQ with something only Nacho can tell: how he measured citations in ChatGPT/Perplexity at Reverse Tech, what changed and with what result.
Talent or skill3/4Very well organized: who it is for, point of view, how we work with weeks, deliverables, results, formats, FAQ, next step in three steps. It quickly answers “is this for me?”. Drawbacks: “No es otro servicio con un nombre de moda: es el mismo trabajo bien hecho” is exactly the slogan that gives away template text; “No tengo tarifas cerradas” and, further down, “propuesta con… precio cerrado” read as a contradiction even though they are not.Rewrite that sentence in plain language (“Lo mido también fuera de Google”) and clarify: “No tengo tarifa fija; cada propuesta lleva un precio cerrado”. Giving a rough price range would filter better.
Accuracy2/4The definitions and the note on llms.txt are correct. The business claims (“Las búsquedas informativas sencillas se las quedan los AI Overviews y los asistentes”) come with no data or source, and the case study figures depend on pages that do not prove them either.Support the thesis with a figure (own or from a cited study) and strengthen the proof in the case studies (see section 3).

Commodity

  • Swap test: it loses Softonic, word.tips, Cliqpod/Reverse Tech and the 4-client limit. The rest (phases, deliverables, FAQ) would stay the same under another name.
  • Top 10 test: partly checked. For “consultoría SEO GEO consultor freelance” the results are mostly Malt profiles and job listings, plus a generic article on what a GEO consultant does. None has case studies with figures, so the page stands out for its proof, not for its method.
  • Only-you assets used: case studies with figures, experience with former Google teams, own opinion, client limit.
  • Non-commodity angle: “Qué cambió cuando medimos las citas de Reverse Tech en ChatGPT y Perplexity durante X meses”, with the before and after table, as the central block of the page.

Red flags: none. One template-style sentence (see Talent) that does not reach filler (5.2.2).


3. Case studies: Softonic, from 0 to 20 million sessions a month

Verdict: Solid with gaps (10/16, non-commodity). It is a story only Nacho can tell, including the failure, but it is short, has no charts and tells the “how” in headlines.

Purpose: show with a real case the ability to design a large-scale content program and lead to contact. YMYL: no.

PillarScoreEvidenceFix
Effort3/4Context with dates and roles (2019–2024, SEO Specialist → Product Owner → SEO Manager), the problem, five levers, a “Lo que salió mal” section with the drop of “un 20 y pico por ciento” after consolidating languages and the recovery after moving to ccTLDs.Add the timeline with dates (when the consolidation, which core update, when the recovery) and a sessions chart, even without a Y axis.
Originality3/4First-hand experience nobody else can publish, with a non-obvious lesson (weak translations lowered the main domain’s rating).Share the figure that would make it citable: what percentage of the generated pages was indexed, how many were withdrawn and by what criterion.
Talent or skill2/4722 words. The only image is the Softonic logo. “Cómo lo hicimos” stays at headline level: “entrenamos nuestros propios modelos de generación” or “cada lote pasaba por flujos de generación y revisión” do not say which models, what review or how many people. “Páginas útiles” is not defined.Expand the “how” with an example page (before/after), the review flow in a diagram and the metric that decided whether a batch continued.
Accuracy2/4The figures are concrete (0 → 20M, 100k → 1M+, +40 % in 2024) but carry no proof or source; “un 20 y pico por ciento” is vague; the core update is not named.One cropped screenshot per main figure, name the core update and give the exact percentage of the drop.

Commodity

  • Swap test: passes. Without Nacho the page does not exist.
  • Top 10 test: does not apply (own case).
  • Only-you assets used: Softonic experience, internal figures, the story of the languages.
  • Pending non-commodity angle: it already is one; what is missing is evidence and process detail.

Red flags: none. One contextual caveat: the case describes AI-generated content at scale (100,000 → 1M+ pages). Told this briefly, a skeptical reader may read it as what section 4.6.5 calls scaled content. Explaining the quality control and the demand criterion in detail protects it. [Auditor judgment]


4. Tools: Bot access checker

Verdict: Solid with gaps (11/16, partly commodity). The tool works and the guide is precise, but almost identical checkers already exist.

Purpose: check whether Googlebot, Bingbot and AI bots can crawl a URL, and explain why they cannot. YMYL: no.

PillarScoreEvidenceFix
Effort3/4Own tool (a 488-line Cloudflare Pages Function): 15 bots with a real request plus 2 robots.txt-only tokens, robots.txt matching with the longest rule and wildcards, comparison with a browser, challenge detection (Cloudflare, Akamai, DataDome…), noindex in meta and X-Robots-Tag with bot prefix, sitemaps, “thin” response and SSRF protection. Tested in production: on nytimes.com it detects the DataDome 403 and the Disallow: / rules for each AI bot; on nachomascort.com, everything “ok” and the sitemap valid.Store (with permission and anonymized) aggregate results to publish original data.
Originality2/4There are almost identical tools. 365i’s (AI Crawler Checker) makes real requests with 14 bots, compares with the browser, checks noindex and warns about the IP verification limit; Lightsite makes requests with 8 bots and detects WAF; LLM Pulse only analyzes robots.txt. What is its own here: in Spanish, any URL (not just the home page), Google-style matching, sitemap and detection of trimmed responses.Publish a study with the tool: “Qué bots de IA bloquean las 500 webs más visitadas de España”, with the table. Nobody has that.
Talent or skill3/4Useful, well-written guide (four causes, search/user/training taxonomy, honest limits, FAQ with the exact Cloudflare dashboard path). The tool marks a fake Googlebot that is blocked as “warning” and not “blocked”, which is the correct reading.Link each “warning” result to the section of the guide that explains it.
Accuracy3/4The behavior on a 5xx robots.txt (12 hours, last copy for up to 30 days, then no restrictions if the site responds) matches Google’s documentation and links to it. The bot taxonomy for OpenAI, Anthropic and Perplexity is correct, as is the Google-Extended point. One detail in the code: the group is chosen with token.startsWith(agent), so a User-agent: Claude line would apply to ClaudeBot and to Claude-SearchBot, when RFC 9309 asks for an exact token match. [Auditor judgment]Change to exact (case-insensitive) matching and, if you want to imitate Google’s fallback for its own crawlers, treat it as a separate case.

Commodity

  • Swap test: with another brand, the Spanish-speaking user loses the guide in Spanish and some checks (sitemap, Google matching); the core they would find at 365i.
  • Top 10 test: checked by opening three results for “comprobar si bots de IA GPTBot ClaudeBot pueden acceder” (LLM Pulse, Lightsite, 365i). The guide and the tool are a bit ahead in technical detail, but the function is the same.
  • Only-you assets used: own tool, experience with Cloudflare and robots.txt.
  • Non-commodity angle: the study of AI bot blocking on Spanish sites done with the tool itself, updated every quarter.

Red flags: none.


5. New blog (2026): Quality at Google

Verdict: Strong (13/16, non-commodity). A first-hand account of Search Central Live Barcelona, with own photos, a real case and sources for every delicate claim.

Purpose: explain what quality means to Google, how it is applied and how long a recovery takes after a core update. YMYL: no.

PillarScoreEvidenceFix
Effort4/45.000 words, 17 images (photos of slides by the author and by Estela Franco, with credit), a recovery time table, Softonic case, 30 dated references (rater guidelines, Google docs, DOJ trial notes, patents, leak), lead magnet with own skill.Already at the maximum.
Originality3/4What is its own: presence in the room, photos of slides, “never (quality)” timings for sitemaps, indexing and structured data, the Softonic languages story and his reading on quality in AI Overviews. What is not its own: the review of the leak and of Q* has already been done by Mike King, SparkToro or SEJ, and there are event recaps (We Are ROAST, Marie Haynes).Publish the original data that would make it a source: results of running the skill on several large sites (anonymized), with which page types come out worst.
Talent or skill3/4TL;DR, index, H2s phrased as questions, tables, clearly separates what is documented from “mi lectura”. Drawbacks: the skill block appears twice in the body; many English terms left untranslated (“quality”, “page type”, “effort”); the length calls for more intermediate summaries.Keep a single skill block (at the end) and add a summary line at the start of the long sections.
Accuracy3/4Verifiable literal quotes, labels what is deduction (“Que ese campo sea la versión algorítmica del pilar de effort es una deducción mía”), warns that the “naranja” data was only seen on the slide. Details: the header says “3 de octubre de 2026” but the text tells of something done “el 4 de octubre de 2026” and there is no update date or dateModified; “El llms.txt, Google lo ignora” is more categorical than the cited guidance (“You don’t need to create… AI text files”).Show “Actualizado el…” and add dateModified; soften to “Google dice que no hace falta y no consta que lo use”.

Commodity

  • Swap test: passes. Without attendance at the event, the photos and the Softonic case, it would be one more leak summary.
  • Top 10 test: partly checked. For the query on recovery after a core update the results are generic guides (Surfer, W3era, Wordtracker) with no data from the event; none has Google’s timing table or the “never (quality)”.
  • Only-you assets used: own photos, event account, Softonic case, own skill.
  • Non-commodity angle to reach 4: the audit dataset proposed above.

Red flags: none. The double lead magnet block is a bit of noise, without reaching the clutter level that clutterScore penalizes.


6. SEO Hacks archive (2015–2017)

Page audited: ¿Qué considera Google que es el “Thin Content” en realidad?

Verdict: At risk (6/16, commodity). A 2015 post with the same list of thin content types that Google and the top 10 already give, a tone from that era and a 2026 note that corrects the one serious error.

Purpose: explain what Google understands by thin content and what to do with poor pages. YMYL: no.

PillarScoreEvidenceFix
Effort2/4Written by a person, with one practical tip of its own (“crear una página en la que englobar todas las noticias”) and a substantial October 2026 note on robots.txt versus noindex. No data, examples or own images.If kept, add a real example (before/after of consolidating short news items).
Originality1/4The four types (auto-generated, affiliate feeds, duplicate, doorway) are the list from Google’s manual action “Thin content with little or no added value” and the same ones SE Ranking gives in its article for “qué es thin content Google”.None that is cheap: what is original is already in the new quality post.
Talent or skill1/4Readable but in the register of 2015: “¡NO PENALIZA!”, “Pandazo”, “TIER 2 o incluso los TIER 3”, “recuerda hacerlo en una web que no sea tu Money Site”, “Un término del que se ha empezado a hablar recientemente” (from 2015), “manual de webmasters” with no link. It does not answer today’s searcher, who expects the relationship with Panda, the helpful content system and core updates.Consolidate (see below).
Accuracy2/4After the note, the essentials are correct (short content does not penalize; technical duplication is solved with canonical). Still outdated are the treatment of doorways as news and the absence of any reference to the current state (helpful content integrated into the core in 2024). [Auditor judgment]If it stays, a closing note linking to the quality post.

Commodity

  • Swap test: fails. Under another name, the reader loses nothing.
  • Top 10 test: checked with “qué es thin content Google” (IEBS, Cyberclick, SE Ranking, Semrush, Rank Math, SEOcrawl). All give the same definition and the same typology.
  • Only-you assets used: none, except the tip to group news items.
  • Non-commodity angle: “Cómo decidimos en Softonic qué páginas generadas retirar”, with the demand criterion and the indexing data. It fits better as a section of the quality post or of the Softonic case study than as a separate post.

Red flags: none from Lowest. It fits the description of 5.2.1 (low effort and little added value compared with what already exists). The date “Actualizado el 5 de octubre de 2026” is justified here by the note, but see the observation on dates in the rest of the archive.

Overview of the rest of the archive (19 posts, category: seohacks)

I reviewed the title, opening and notes of the 19 posts in src/content/blog/es/:

  • 18 of 19 have a 2026 note or a “Qué sigue vigente en 2026” section. They are real corrections and well done (UA → GA4, noindex in robots.txt since 2019, LSI, Penguin 4.0, SSL Flexible, etc.).
  • relacion-semantica-inversa has no note and its updatedDate is still 2016. It is a hypothesis with no data (“lanzando una hipótesis basada en la observación”) published without qualification.
  • Several posts teach techniques against the spam policies even though they now carry a warning: cloaking (a practical case of hacking + parasite SEO + cloaking), analizando-tecnica-seo-piel-de-cordero (“Analizando la técnica Black Hat”), link-building-alex-navarro (pyramids with PBN, “Tener una PBN en un segundo nivel te asegura que si la competencia te hace un spam report…”), imagenes-originales (the title promises “hacer creer a Google que tus imágenes son originales (cuando no lo son)”) and el-dia-a-dia-de-un-seo-en-gifs (PBN). On the site of someone who sells quality audits, this contradicts the message.
  • 2015 tone and format in almost all (“¡Hola, querido lector!”, “¡Entra Ya!” in the meta description of https-gratis) and no images in 15 of 19 because of the Internet Archive recovery.
  • All carry updatedDate: 2026-10-05 and the same dateModified. The notes are real changes, but 18 posts updated on the same day with a short note on top of a 2015 body looks like what Google advises against (changing the date without a substantial change). [Auditor judgment]
  • What is worth keeping: ddos (own experience, 120,000 visits lost), crawl-budget-optimizar-http-304 (technical and still useful), seo-tips-gary-illyes (a compilation with some historical value) and the two Quality Raters/E-A-T posts, whose topic the new post already covers better.

Page type verdict: At risk. None of it is spam, but as a whole it is commodity or outdated content with a minority of pieces of its own.


Page type and site view

Count from dist/sitemap-0.xml: 66 indexable URLs (43 in Spanish, 23 in English).

Page typeURLs (ES + EN)Fit with focusRecommendation
Home1 + 1TotalImprove (verifiable proof, less repetition)
Services4 + 4 (+ hubs)TotalImprove (own angle in process and FAQ, indicative price)
Case studies4 + 4 (+ hubs)TotalImprove (charts, dates, detail on the “how”)
Tools2 + 2 (+ hubs)HighImprove (publish own data with the tool)
Skills1 + 1 (+ hubs)HighKeep
New blog2 + 2HighKeep and publish more (there is only one 2026 post and one from 2023)
SEO Hacks archive19 + 0 (+ hub)Medium: it is SEO, but 2015 tactics and black hatConsolidate, noindex or remove most of it
Legal and contact3 + 3NeutralKeep

What weighs on the site:

  • Proportion. The archive is 19 of the 43 Spanish URLs (44%) and 19 of the 21 posts on the Spanish blog. For Google, most of the site’s “editorial content” is SEO Hacks. The quality post itself says it: weak pages drag down the good ones.
  • Focus. The topic (SEO) is the same, so siteFocusScore should not suffer much. What strays from the current core (SEO, GEO and AI consulting for large sites) is the WordPress tutorials, HTTPS with Cloudflare, GIFs and the black hat tactics. [Inference]
  • Brand coherence. The pieces on cloaking, PBN and “original images when they are not” go against the quality message the site sells.
  • English. The EN versions of home, services, case studies, tools and the two new posts are linked with hreflang. The archive has no EN version, so the English part of the site is cleaner than the Spanish one.

Recommendation for the archive (no traffic data; if any post brings visits or links, review it first):

ActionPosts
Improve and keepddos, crawl-budget-optimizar-http-304
Consolidate with a 301 to the quality postguia-quality-raters-google, credibilidad-vs-autoridad-eat, que-es-thin-content
Noindex (they stay as an archive for anyone arriving through a link)seo-tips-gary-illyes, guia-seo, 200-factores-seo-que-posicionan-en-google, analisis-pagina-web-seo-on-page, accesibilidad-web-seo-parte-1, contenido-latent-semantic-index, creando-blog-seo-wordpress, https-gratis, el-dia-a-dia-de-un-seo-en-gifs
Noindex or remove (black hat or hypothesis with no basis)cloaking, analizando-tecnica-seo-piel-de-cordero, link-building-alex-navarro, imagenes-originales, relacion-semantica-inversa

If you prefer not to touch indexing, the minimal alternative is to take the archive out of the sitemap and keep update dates only where the note changes the meaning of the post.


FindingSignalEvidence
Archive with little effort of its own versus the new post, with a lotcontentEffortDescription “LLM-based effort estimation for article pages” [Documented]; that it measures the same as the Effort pillar [Inference]
44% of the Spanish URLs are commodity posts from 2015–2017chardEncoded, tofu (site)“site quality predictor based on content” [Documented]; that the archive lowers that rating [Inference]
Archive posts well below the level of the rest of the siterhubarb“Site-URL delta signals based quality score” [Documented]; applying it to this case [Inference]
WordPress/HTTPS tutorials, GIFs and black hat far from the current coresiteFocusScore, siteRadiusDescriptions [Documented]; that these posts increase the radius [Inference]
A consultant’s personal websitesmallPersonalSite“Score of small personal site promotion” [Documented]; whether it applies to this site [Hypothesis]
No ads or popups; one repeated lead magnet in the new postclutterScoreDescription [Documented]; low risk here [Auditor judgment]
Archive posts with little demand and value could fall into low tiers or go unindexedscaledSelectionTierRankScore over tiers Base, Zeppelins, Landfills [Documented]; Base better and Landfills worse [Inference]
Quality computed per page but with site signalsQ* (DOJ trial, HJ Kim’s notes)“generally static across multiple queries” [Documented]

Priority actions

  1. Reduce the weight of the archive before publishing. Apply the table above: 301 the three posts that the quality post covers, noindex or remove the five black hat or baseless ones, and noindex the rest except ddos and crawl-budget. At a minimum, take them out of the sitemap and add a note to relacion-semantica-inversa.
  2. Review the archive dates. Keep updatedDate only where the note really changes the advice; on the rest, show “Nota añadida en 2026” without touching dateModified.
  3. Put proof in the case studies. A chart or screenshot per main figure, dates and the name of the core update at Softonic, and a “how” with examples. This raises the accuracy of home, services and case studies at once.
  4. Give the consulting service its own angle. Replace the generic SEO/AEO/GEO FAQ with a case of measuring AI citations, clarify the fixed price point and remove the sentence “No es otro servicio con un nombre de moda: es el mismo trabajo bien hecho”.
  5. Turn the tool into a source. Publish a study with it (AI bot blocking on Spanish sites) and fix the startsWith matching to an exact token match.
  6. Polish the quality post. A single skill block, dateModified and “Actualizado el…” consistent with the mention of 4 October, and soften the llms.txt point.
  7. Home. Unify the years of teaching, say the former Google engineers claim only once and link the media mentions.
  8. Editorial pace. With a single 2026 post, the blog depends on the archive. Two or three more non-commodity pieces (audit dataset, bot study, a measured GEO case) would change the proportion of the site.
See the original report from 4 October 2026

Site verdict: Solid with gaps. The commercial pages and the new post hold up well, but the SEO Hacks archive (19 posts from 2015–2017, indexable and in the sitemap) is the weakest page type and weighs heavily on such a small site.

One representative page audited per page type, judging only the main content (MC). Sources: HTML generated in dist/ (4 Oct 2026), a live check of nachomascort.com, and searches for the target queries. The score is a heuristic for prioritizing, not a Google grade.

Summary

Page typePage auditedScoreCommodity classVerdict
Home/11/16Non-commoditySolid with gaps
Services/servicios/consultoria-seo-geo/9/16Partly commoditySolid with gaps
Case studies/casos/softonic-0-a-20-millones/9/16Non-commoditySolid with gaps
Tools/herramientas/comprobador-bots/11/16Partly commoditySolid with gaps
New blog (2026)/blog/quality-google-core-updates/12/16Non-commoditySolid with gaps
SEO Hacks archive/blog/que-es-thin-content/5/16CommodityAt risk

No page reaches “Strong” (13+). The closest is the quality post; the furthest is the archive.


1. Home

Verdict: Solid with gaps (11/16, non-commodity). It is a personal page with a track record and original figures; the gap is that the strongest claims cannot be verified from the page.

Purpose: introduce Nacho Mascort and turn visits into calls or contacts. YMYL: no.

PillarScoreEvidenceFix
Effort3/4Detailed track record (UPF 2015–2024, Webpositer since 2018), 4 case studies with figures, a course with data (450+ students, 80+ videos). Handmade and tailored.Nothing urgent.
Originality3/4Everything is first person: Softonic, word.tips, Reverse Tech, the ad factory (200 → 3.000+ creatividades/semana). It does not exist on any other site.Nothing urgent.
Talent or skill3/4Clear structure: promise, social proof, services, case studies, teaching, contact. It quickly answers “why you?”.Nothing urgent.
Accuracy2/4“Me contrataron ex ingenieros de Google Search” is the main hook and has no name or link. “Me han citado: Google Search Central” appears with a logo, when on /sobre-mi/ it is a “Mención” in a 2022 community recap. “Más de ocho años enseñando” contradicts the page’s own line “UPF 2015–2024” (that is over 10).Link to or specify the former engineers claim (even as “project X, with permission”), link each “Me han citado” logo to the specific piece, and correct the years of teaching.

Commodity

  • Swap test: if you swap the name for another consultant’s, the page loses everything. Non-commodity.
  • Top 10 test: not checked (brand page, navigational query).
  • Assets only you have: case studies with figures, track record, own course, tools.
  • Angle to strengthen: link each figure to the case study that proves it (it already does this in part).

Red flags: none found. No popups or interstitials; the contact form sits at the end.


2. Services

Verdict: Solid with gaps (9/16, partly commodity). The proof (Softonic, word.tips, maximum 4 clients) is original, but the method and deliverables could be signed by any consultancy, and the same proof is repeated three times.

Purpose: sell SEO/AEO/GEO consulting to large sites. YMYL: no.

PillarScoreEvidenceFix
Effort2/4Clear human work: “Para quién es / No es para ti”, the 4-client limit and the pricing explanation. But there is nothing first-hand in the method: no example audit and no screenshot of AI measurement.Add an anonymized excerpt from an audit or a real 90-day roadmap.
Originality2/4“Diagnóstico → Arreglos → Contenido → Medición” and the five deliverables are the standard outline of any consulting page. The only original line is “llevé el catálogo de Softonic de 100.000 a más de 1 millón de páginas útiles”.Explain the method with a concrete case: what you found in the crawl of X and what changed.
Talent or skill3/4Well structured, useful FAQ, the price is explained without hiding it. It loses some points because the phrase “Softonic, 1.000 millones de usuarios / word.tips, 2 millones de visitas diarias” appears in the intro, in Resultados and in the FAQ.State the proof once and use the space for something new.
Accuracy2/4The SEO claims are generally correct but vague (“en las webs grandes es donde más tráfico se pierde”). “llms.txt” appears as a deliverable of the GEO strategy without explaining that no major search engine has confirmed it uses it [Auditor judgment]. The “ex ingenieros de Google” hook again has no source.Explain what llms.txt does and does not do, or remove it from the deliverables. Specify the hook.

Commodity

  • Swap test: if you remove the Softonic/word.tips paragraphs, the rest works for any other consultancy. Partly commodity.
  • Top 10 test: partly checked (only titles and snippets for “consultoría SEO GEO consultor freelance”: service pages from Malt and from consultants with the same audit + GEO strategy outline).
  • Assets only you have, used: Softonic and word.tips case studies, 4-client limit.
  • Non-commodity angle: “Qué encuentro en la primera semana de una auditoría a una web de +1M de URLs”: the 3–5 problems that repeat most across your clients, with anonymized data.

Red flags: none from Lowest/Low. Repetition of the social proof, minor (5.2.2, filler).


3. Case studies

Verdict: Solid with gaps (9/16, non-commodity). The site’s most valuable asset is told in 450 words, with no dates, no chart and no explanation of the quality controls of a content program generated with in-house models.

Purpose: show results to build commercial trust. YMYL: no.

PillarScoreEvidenceFix
Effort2/4Four figures and five “Cómo lo hicimos” blocks of one line each. No chart, screenshot, timeline or example of a generated page.Add the sessions curve (even indexed to 100), a timeline and a before/after of one page.
Originality3/4It is a case only you can tell: 0 → 20M sessions/month, catalog from 100k → 1M+ pages, in-house models.Nothing to add here; the problem is depth, not originality.
Talent or skill2/4Correct and readable, but it stays at headline level: “flujos de generación y revisión que mantenían el nivel editorial” does not explain how. It lacks what a reader would most want: what went wrong.Explain what quality controls existed, what percentage was reviewed and what was discarded.
Accuracy2/4Figures without a date or period: “0 → 20M en 24 meses” does not say which 24 months (the Softonic stage is 2019–2024 according to /sobre-mi/). “Cuando arrancamos aún no había LLM comerciales” depends on the start year: the GPT-3 API was commercial from 2020 [Auditor judgment]. The “+40 % en 2024” is not connected to the program.Add years. Qualify it as “before ChatGPT” instead of “there were no commercial LLMs”. Explain where the +40% comes from.

Commodity

  • Swap test: if you change the name, the case study loses everything. Non-commodity.
  • Top 10 test: not checked (there is no generic target query).
  • Assets only you have, used: Softonic’s own figures.
  • Missing non-commodity angle: the story you already tell in the quality post (translations dragged the site down in a core update, move to ccTLD and recovery) is not in the case study.

Red flags: none on the page. Reading risk: the case study describes, without visible controls, the pattern the guidelines associate with scaled content (4.6.5): generating hundreds of thousands of pages with models. It is not a flag on this page, but on a site that positions itself on quality, it makes sense to show the quality control.


4. Tools

Verdict: Solid with gaps (11/16, partly commodity). The tool works and the guide is practical, but several free tools do something similar.

Purpose: check whether search and AI bots can access a URL and capture leads. YMYL: no.

PillarScoreEvidenceFix
Effort3/4Working tool (tested live: /api/bot-check returns robots.txt, sitemaps and a verdict per bot for 15 bots, with CDN challenge detection). Guide with 4 causes + FAQ.Nothing urgent.
Originality2/4Search turns up LLM Pulse, SEOCrawl, Surmado, IsVisible or Lightsite with the same purpose. What sets it apart is comparing the browser’s response with each bot’s and the specific Cloudflare dashboard paths.Publish original data: run the tool on the 500 Spanish sites with the most traffic and report how many block each bot without knowing it.
Talent or skill3/4It explains the search / user / training split well and is honest about its limits (“confírmalo en los logs”).Nothing urgent.
Accuracy3/4Correct and specific: Google-Extended versus Googlebot for AI Overviews, robots.txt 5xx versus 404. The 5xx point is simplified (Google applies time windows and uses the cached copy) [Auditor judgment].Add the nuance about 5xx time windows and a link to Google’s robots.txt documentation.

Commodity

  • Swap test: with another brand the page would lose little; the tool is the value and others partly replicate it. Partly commodity.
  • Top 10 test: partly checked (titles and snippets for “comprobar si bots de IA GPTBot ClaudeBot pueden acceder”: at least 5 similar tools, several only read robots.txt).
  • Assets only you have, used: own tool, Cloudflare experience.
  • Non-commodity angle: the data study above, published as a post and linked from the tool.

Red flags: none found.


5. New blog (2026)

Verdict: Solid with gaps (12/16, non-commodity). It is the best page on the site: a first-hand account, labeled sources and an original case. It falls one point short of “Strong” because that original case is told without figures or dates.

Purpose: explain what quality means to Google based on the Search Central Live Deep Dive in Barcelona and lead readers to the audit skill. YMYL: no.

PillarScoreEvidenceFix
Effort3/4Attendance at the event, own photos of the slides, 28 references, timing tables, a distinction between what is documented and what is inferred.To reach 4: original data (e.g., the evolution of “Rastreada: actualmente sin indexar” for a client).
Originality3/4The only unique parts: the Softonic case (translations that lower site quality, move to ccTLD, recovery in the next core) and the retrieval slides with quality as a URL attribute. The rest (leak, patents, DOJ Q*) is already on iPullRank, SEJ and similar.Give more space to the Softonic case and less to the leak summary.
Talent or skill3/4TL;DR, tables, literal quotes and an action list at the end. It is 4.300 words; the twiddlers section drifts away from the reader’s question.Cut twiddlers to one paragraph or move it to a separate post.
Accuracy3/4Careful: “es una deducción mía”, “tómalos como lo que Google enseñó en la sala”, “nadie lo ha documentado”. Gaps: “una caída de un 20 y pico por ciento” with no date or core update; “Gary Illyes dejó claro que los core updates miran el site” with no written source. It links to the 2015 thin content post as “ya lo decía”, and that post recommends blocking with robots.txt (see section 6).Date the Softonic case, cite Gary’s source (We Are ROAST recap or another) and change or update the link to the thin content post.

Commodity

  • Swap test: with another brand, the account and the Softonic case are lost. Non-commodity.
  • Top 10 test: partly checked (the event ran 30 Sep to 2 Oct 2026; there are few written accounts and search only returns event announcements, not equivalent content).
  • Assets only you have, used: attendance, photos, Softonic case, own skill.
  • Non-commodity angle to strengthen: the Softonic case with figures and dates.

Red flags: none found. There is a lead magnet block, but it does not cover the MC.


6. SEO Hacks archive (2015–2017)

Verdict: At risk (5/16, commodity). A 2015 post that says the same as the current top 10, with advice that is incorrect today, missing images and calls to action that no longer work. And it is the most numerous page type on the blog.

Purpose: explain what Google considers thin content. YMYL: no.

Page audited: /blog/que-es-thin-content/ (published 16 Nov 2015, updated 8 Feb 2016; 1.167 words, at the archive’s median). It carries the notice “Artículo recuperado de SEO Hacks… parte de la información puede estar obsoleta”. It is indexable and in the sitemap.

PillarScoreEvidenceFix
Effort2/4Written by a person, with one original tip (“crear una página en la que englobar todas las noticias”). The recovered version has unreviewed leftovers: “deja tu comentario aquí abajo” (there are no comments), “suscríbete al blog”, “Ya estoy preparando un nuevo post sobre…”.If kept, actually review it: remove dead CTAs and update it.
Originality1/4The four types (auto-generated, affiliate feeds, duplicate, doorways) come from Google’s documentation and appear across the whole current top 10 (IEB School, Cyberclick, SE Ranking, Semrush…).Only worth it if rewritten with original cases (Softonic, word.tips).
Talent or skill1/4Typos (“Feeds de afilición”, “feed da afiliación”, “españa”), a long intro (“Después de meter el rollo…”), lost images, a video linked instead of explained.Rewrite or retire.
Accuracy1/4It recommends “capar el acceso a esa categoría al bot de Google a través del Robots.txt” for thin content: blocking does not remove URLs from the index and also stops Google from seeing a noindex [Auditor judgment, consistent with Google’s robots.txt documentation]. It dates the doorways update “en verano” when Google announced it in March 2015 [Auditor judgment]. It frames duplicates as “Pandazo” without distinguishing technical duplication from copying [Auditor judgment]. It recommends hiring cheap writing on foreign platforms.Fix or retire. At a minimum, remove the robots.txt advice.

Commodity

  • Swap test: with another brand nothing is lost. Commodity.
  • Top 10 test: partly checked (titles and snippets for “qué es thin content Google”: the same definitions and types everywhere).
  • Assets only you have, used: none relevant today.
  • Non-commodity angle: “Thin content en un catálogo de 1M de páginas: qué consolidamos y qué borramos en Softonic”, with data.

Red flags: 5.2.1 (low effort and originality, Low) for this page in its current state. It is not scaled content abuse (4.6.5): it is real human content, from a real author, with an archive notice.

What the rest of the archive says (by title and opening of each post, not audited one by one): posts with first-hand experience (ddos: “perder 120.000 visitas por un ataque DDoS”; link-building-alex-navarro; crawl-budget-optimizar-http-304; cloaking) coexist with commodity or obsolete pieces (200-factores-seo-que-posicionan-en-google, an infographic based on English-language roundups; guia-seo “Posiciona cualquier web en 2017”, unfinished at 10 installments; https-gratis; el-dia-a-dia-de-un-seo-en-gifs; seo-tips-gary-illyes, a compilation of tweets; creando-blog-seo-wordpress, which the author himself describes in another post as “demasiado básico… hay mil y un blogs que explican cómo hacerlo”).


Page type and site view

Size: the sitemap has 62 URLs. 19 are SEO Hacks archive posts (31%). On the Spanish blog, 19 of 21 posts are from 2015–2017 (the other later one is from 2023). There is no archive in English, so there are no weak translations: the EN versions are full equivalents with hreflang.

Topical focus: the whole site is SEO, GEO, AI and growth. No page type strays from the core topic, so siteFocusScore and siteRadius do not seem to be a problem [Inference]. The problem is not focus but average quality: on a 62-URL site, a third of unreviewed 2015 content weighs far more than on one with 50.000.

Drag: the site is small and personal, with strong case studies and tools. If Google judges quality at site level (as the core updates documentation indicates, “the site as a whole”), the archive is the only thing that can visibly lower the average. seohacks.es redirects with a 301 to the new URLs, so the archive inherits the old links; that explains why it was migrated, but does not justify indexing the weak posts.

Recommendation by page type:

Page typePages (ES/EN)RecommendationWhy
Home1 / 1ImproveSource the main claims.
Services4 / 4 (+ index)ImproveRemove repetition and add one element only you have per service.
Case studies4 / 4 (+ index)Improve (priority)It is the best asset on the site and is told in 450 words.
Tools2 / 2 (+ index)ImproveThey work; an original data angle is missing.
New blog2 / 2Improve (light)Already the best on the site; date and quantify the case.
SEO Hacks archive19 / 0Remove or noindex most; consolidate the 3–4 with first-hand experienceMostly commodity or obsolete and 31% of the sitemap.

For the archive specifically: noindex and out of the sitemap (or 410) for the commodity or obsolete posts. Review and update those with first-hand experience (ddos, link building, crawl budget 304, cloaking) and merge the two 2016 posts on Quality Raters and E-A-T (guia-quality-raters-google, credibilidad-vs-autoridad-eat) into the new quality post, with a 301. Moving the archive to another domain does not pay off: seohacks.es was already consolidated here and it would undo that work.

FindingSignalEvidence
A third of the sitemap is unreviewed 2015–2017 contentchardEncoded, tofu (site)Description “site quality predictor based on content” [Documented]; that the archive lowers that prediction [Inference]
Archive posts well below the quality of the rest of the siterhubarb“Site-URL delta signals based quality score” [Documented]; applying it to this case [Inference]
Archive with little visible effort versus the new postscontentEffort“LLM-based effort estimation for article pages” [Documented]; that it measures the effort pillar [Inference]
Small personal single-author sitesmallPersonalSite“Score of small personal site promotion” [Documented]; that this site benefits [Hypothesis]
No popups or interstitialsclutterScoreDescription [Documented]; no findings here
Whole site on the same topicsiteFocusScore, siteRadiusDescriptions [Documented]; low risk here [Inference]

Priority actions

  1. Triage the SEO Hacks archive. Noindex and out of the sitemap (or 410) for the commodity or obsolete posts; update the 3–4 with first-hand experience; consolidate the Quality Raters and E-A-T posts into the new post. Change the quality post’s link to the thin content post, or update that post and remove the robots.txt advice.
  2. Rewrite the case studies, starting with Softonic. Years, timeline, sessions chart, quality controls of the AI program, what was reviewed and what was discarded, and the story of the core update and the ccTLD move that you already tell in the post. Correct “aún no había LLM comerciales”.
  3. Source the commercial claims and remove repetition in services. Specify or link the “ex ingenieros de Google Search” claim, link the “Me han citado” logos, correct “más de ocho años” of teaching, explain or remove llms.txt as a deliverable and add an anonymized audit excerpt to each service.
  4. Publish a study with data from the bot checker (e.g., the Spanish sites with the most traffic) to turn the tool into a citable source.

Want the same report for your site?

Tell me about your project

Tell me what you want to achieve and I'll tell you honestly whether I can help.

What happens next

  1. Reply within 24 business hours.
  2. 30-minute call.
  3. Fixed proposal with scope and price.

Prefer to talk? Book a 30-min call
Or email me at [email protected]

I only use your data to reply to you. Privacy policy.