How to read this report
Every statement about Northgauge carries exactly one evidence label. In a client report, most technical findings would be [M] Measured, with the page, date and method recorded. In this demonstration nothing was measured, so company facts are labelled [S] Scenario.
| Label | Meaning | Use in this demonstration |
|---|---|---|
| [S] Scenario assumption | Invented for the fictional business | All facts about Northgauge, its website and competitors |
| [H] Hypothesis | Plausible opportunity that needs validation | Opportunities, buyer expectations, all German terms |
| [E] Estimate | Approximate effort or planning figure | Effort sizes and roadmap timing |
| [M] Measured | Value from an actual documented measurement | Not used. Nothing was measured. |
| [V] Verified | Fact supported by an identifiable source | Only general guidance from the sources in chapter 14, never Northgauge facts |
Effort sizes [E]: S = under 1 working day; M = 1–5 working days; L = more than 5 working days. These are planning estimates.
Executive summary
- Overall readiness
- Not yet ready for Germany, but fixable in about one quarter. [S][H]
Northgauge has a clear product and a plausible German audience, but its website cannot yet serve German buyers. The two blockers are technical: a crawl block on the product pages and no language architecture. The main commercial gap is the absence of German-relevant trust signals. None of these needs a rebuild. The order of work matters more than the volume.
SEO opportunities [H]
- Remove the
/product/crawl block first (SEO-01). - Decide the
/de/structure and hreflang before translating (SEO-02). - Build pages around how German maintenance managers describe the problem (SEO-05).
GEO opportunities [H]
- One consistent, factual product description everywhere (GEO-01).
- Publish verifiable evidence for claims (GEO-03).
- Make a deliberate AI-crawler decision (GEO-05).
Localization risks [H]
- Literal translation of UK-specific content and GBP pricing (LOC-03).
- No terminology or formality decisions (LOC-04, LOC-05).
- Publishing German copy without competent review (LOC-07).
Five highest-priority recommendations
| # | Recommendation | Rationale | Effort [E] |
|---|---|---|---|
| 1 | Remove the /product/ disallow rule in robots.txt | Product pages may not be crawled at all | S |
| 2 | Decide the German site architecture (/de/ plus hreflang) | Prevents costly rework after translation | S decision / M build |
| 3 | Write a German terminology and tone sheet, reviewed by a qualified German reviewer | Consistency and credibility on every German page | M |
| 4 | Add German-relevant trust content: EU data-hosting facts, security, local support hours | Likely decision criteria for German buyers [H] | M |
| 5 | Localize five priority pages in the order set out in chapter 08 | Concentrates a limited budget on the pages closest to revenue | L |
Scope and methodology
Scope
- Up to 10 agreed English-language pages on one domain (listed in chapter 04).
- One target market: Germany, German (de-DE).
- Four areas: technical and on-page SEO, German search intent, AI search readiness, localization readiness.
Method in a real audit
| Area | Method and free tools | Evidence recorded |
|---|---|---|
| Technical SEO | robots.txt, XML sitemap and page source; free crawler within its limits; PageSpeed Insights; Rich Results Test | URL, date, tool, exact observation |
| German search intent | Manual searches with German language and region settings: autocomplete, "People also ask"; Google Trends (relative only); competitor German pages | Settings, date, observed suggestions |
| AI search | 10–15 buyer-intent prompts in approved assistants, fresh conversation per prompt | Prompt, date, tool, mentions, visible sources |
| Localization | Page-by-page review of content, formats, trust signals and transferability | Page, observation, reviewer status |
Not included
Implementation; German copywriting or translation; paid-tool data such as search volumes, backlinks or rank tracking; legal, tax or compliance advice; markets other than Germany. No rankings, traffic, AI citations or commercial outcomes are guaranteed or implied.
In this demonstration: none of these methods was executed. Chapter 13 sets out what a real audit would record instead.
Client scenario [S]
Every fact in this chapter is a scenario assumption invented for the demonstration.
| Company | Northgauge Ltd (fictional), Leeds, UK; about 40 employees |
|---|---|
| Product | Preventive-maintenance and asset-inspection software (a CMMS) for mid-sized manufacturing plants: mobile inspection checklists, maintenance scheduling, spare-parts tracking, audit-ready reports |
| Customers today | UK and Irish manufacturers with 100–1,000 employees |
| Sales model | Website demo requests and a 14-day free trial, then sales-assisted closing |
| Why Germany | Large mid-sized manufacturing sector; inbound demo requests from German-speaking companies with no German content to send them; board goal to test Germany within 12 months |
| German buyer | Head of maintenance (Leiter Instandhaltung, unreviewed) or plant manager; IT and data-protection staff influence the decision |
| Constraints | No German website, staff or customers yet; small marketing team; translation budget for about five pages in phase one |
Pages in scope
| # | Page | URL |
|---|---|---|
| P1 | Homepage | / |
| P2 | Product overview | /product |
| P3 | Preventive maintenance feature | /product/preventive-maintenance |
| P4 | Mobile inspections feature | /product/inspections |
| P5 | Pricing | /pricing |
| P6 | Manufacturing use case | /industries/manufacturing |
| P7 | Customer story (fictional customer) | /customers/brightwater-foods |
| P8 | Security and data protection | /security |
| P9 | Demo request | /demo |
| P10 | Blog post: "What is a CMMS?" | /blog/what-is-a-cmms |
Readiness by area
Qualitative ratings based on the scenario. They are judgements, not scores.
| Area | Rating | Basis | Findings |
|---|---|---|---|
| Crawlability and indexing | /product/ disallowed [S] | SEO-01 | |
| Language architecture | No German structure or hreflang [S] | SEO-02, LOC-01 | |
| On-page SEO | Generic titles, vague headings [S] | SEO-03, SEO-04, SEO-06 | |
| German search-intent coverage | No German content; gaps identified [H] | SEO-05 | |
| Trust and conversion | No EU hosting facts; one UK case study [S] | SEO-09, SEO-10 | |
| AI search readiness | Inconsistent descriptions; no crawler policy [S][H] | GEO-01 to GEO-06 | |
| Localization readiness | GBP-only pricing; text in screenshots [S] | LOC-02 to LOC-07 |
SEO findings
Each finding separates what was observed, what it implies and what we recommend.
SEO-01
Product pages blocked from crawling
- Observation
robots.txtcontainsDisallow: /product/. [S]- Implication
- Search engines cannot crawl the pages that best explain the product, in English now and in German later.
- Recommendation
- Remove the rule, check that no other rule blocks key pages, and request a re-crawl in Search Console.
- Measure by
- URL Inspection shows the pages as crawlable and indexed.
SEO-02
No language architecture for German
- Observation
- No German pages, language switcher or hreflang annotations. [S]
- Implication
- An unplanned structure leads to competing language versions and rework. Google notes that keywords in URLs have "hardly any effect" on rankings, so structure should be chosen for clarity. [V: Source 1]
- Recommendation
- A
/de/subfolder with hreflang (en-GB,de-DE,x-default) and a crawlable switcher that keeps users on the equivalent page. Check hreflang details against Google's localized-versions documentation during implementation. [H] - Measure by
- Reciprocal hreflang on all German pages; URL Inspection shows the expected canonical for each version.
SEO-03
Generic title tags and meta descriptions
- Observation
- Feature pages use "Features | Northgauge"; meta descriptions repeat the homepage text. [S]
- Implication
- Google recommends titles that are "unique to the page, clear and concise" and meta descriptions that are "short, unique to one particular page". [V: Source 1]
- Recommendation
- Unique titles and descriptions in English now; German versions written from the reviewed terminology sheet (LOC-04), not by literal translation.
- Measure by
- Crawl export shows no duplicates; with Search Console access, compare click-through rate over time (a trend, not a guaranteed result).
SEO-04
Headings do not state the problem solved
- Observation
- P3's H1 is "Stay ahead"; P4's is "Inspections, reimagined". [S]
- Implication
- Vague headings make the page topic harder to understand. Google treats headings as a reader aid and says order is not critical for Search [V: Source 1]; the issue here is clarity [H].
- Recommendation
- H1s that name the task, for example "Preventive maintenance scheduling for manufacturing plants".
- Measure by
- Every in-scope page has one descriptive H1.
SEO-05
No content for German search intent
- Observation
- No German content exists. [S] German buyers may search with terms such as Instandhaltungssoftware or Wartungsplaner rather than "CMMS". [H; German unreviewed]
- Implication
- Pages matched to how buyers search are more likely to be found and to convert than literal translations. [H]
- Recommendation
- Build the German page plan from the search-intent map (chapter 09) after validating each theme.
- Measure by
- Each theme validated; later, Search Console impressions for German queries on German pages.
SEO-06
Weak internal linking to commercial pages
- Observation
- The blog post (P10) and use-case page (P6) do not link to pricing or demo pages. [S]
- Implication
- Google mainly discovers pages through links. [V: Source 1] Readers also lose the path to the next step.
- Recommendation
- Contextual links from P6 and P10 to P2, P5 and P9; mirror the structure on German pages.
- Measure by
- Crawl shows each commercial page linked from all relevant in-scope pages.
SEO-07
No structured data
- Observation
- No structured data on any page. [S]
- Implication
- Valid structured data can make pages "eligible" for special search features; eligibility is not a guarantee of display. [V: Source 1]
- Recommendation
- Organization markup site-wide; evaluate SoftwareApplication markup for the product page; validate with the Rich Results Test.
- Measure by
- Rich Results Test passes without errors.
SEO-08
Page experience not yet measured
- Observation
- No performance data exists in this scenario. [S]
- Implication
- Without measurement there is no basis for a performance finding, so none is made.
- Recommendation
- In a real audit, test all 10 pages in PageSpeed Insights (mobile and desktop), record lab data with dates, and use field data only where available.
- Measure by
- This finding is the measurement step itself.
SEO-09
Trust signals do not address German buyer concerns
- Observation
- The security page mentions "UK GDPR" but not hosting location; the only customer story is a UK food manufacturer. [S]
- Implication
- German B2B buyers may ask where data is hosted and who uses the product in comparable markets. [H] Google's guidance favours "helpful, reliable, people-first" content. [V: Source 1]
- Recommendation
- State verified facts only: hosting location, certifications actually held, support languages and hours. Add an EU-continental customer story when one exists. Never invent certifications or references.
- Measure by
- Facts confirmed by the client in writing; qualitative feedback from German prospects.
SEO-10
Conversion path not ready for German leads
- Observation
- The demo form requires a UK phone format and shows GBP pricing; there is no German-speaking contact option. [S]
- Implication
- A German buyer who reaches the demo page may leave if the form rejects their details. [H]
- Recommendation
- Accept international phone formats; show EUR pricing on German pages once approved; state the language of demos and support honestly.
- Measure by
- Test submissions with German details; with analytics access, demo requests from German pages.
AI search (GEO) readiness
GEO here means making Northgauge easy to understand, verify and describe accurately when buyers use AI assistants. No AI system can be made to mention or cite a brand, and none is promised.
| Principle | What it means | Basis |
|---|---|---|
| Consistent description | The same core sentence about who, what and for whom across pages and profiles | General practice; [H] that it improves AI representation |
| Accessible content | Key facts in HTML text, not only in images or PDFs | General practice |
| Verifiable claims | Claims backed by customer stories, documentation or third-party references | Aligns with "helpful, reliable" content [V: Source 1] |
| Crawler access decisions | robots.txt choices affect which AI crawlers may access pages | [V: Source 2] for OpenAI crawlers |
| Dated testing | Visibility reported as a dated sample, never a general truth | Locentri method |
GEO-01
Inconsistent product description
- Observation
- The homepage says "maintenance platform", the product page "asset management suite", the blog "CMMS". [S]
- Implication
- A single, consistent description makes it easier for buyers and AI assistants to categorize the product correctly. [H]
- Recommendation
- Agree one English definition and, after review, one German definition; use them in introductions, metadata and company profiles.
GEO-02
Key facts locked in images
- Observation
- Feature comparisons and pricing tiers appear as images. [S]
- Recommendation
- Put essential facts in HTML text and tables; keep images as support.
GEO-03
Thin evidence for claims
- Observation
- Claims such as "cut downtime" have no supporting customer evidence or method. [S]
- Recommendation
- Support each claim with a sourced customer story or remove it. Never publish invented results.
GEO-04
No expert or reference content
- Observation
- No named authors or maintenance expertise on the blog. [S]
- Implication
- Practical, expert-written guides for maintenance managers could earn references from industry sources. [H]
- Recommendation
- Publish expert content in English and, later, reviewed German versions.
GEO-05
No decision on AI crawler access
- Observation
- robots.txt has no AI-crawler rules. [S]
- Context
- OpenAI documents separate crawlers. Sites opted out of OAI-SearchBot "will not be shown in ChatGPT search answers"; disallowing GPTBot "indicates a site's content should not be used in training generative AI foundation models"; for user-initiated ChatGPT-User visits, "robots.txt rules may not apply". [V: Source 2]
- Recommendation
- Decide deliberately, for example allowing search-oriented crawlers while deciding separately on training crawlers. Check other providers' crawler documentation individually. [H]
- Measure by
- robots.txt reflects the documented decision.
GEO-06
No German-language presence for AI answers
- Observation
- No German content about Northgauge exists. [S]
- Implication
- AI answers to German-language questions may draw on German-language sources, which Northgauge does not yet have. [H]
- Recommendation
- Localized priority pages (chapter 08) and, later, German presence on relevant directories or review sites where Northgauge qualifies.
AI search test design
No AI tests were run for this demonstration. There is no real brand to test, so no outputs, mentions, citations or test dates are reported.
- Prompt set: 10–15 buyer-intent prompts, mostly in German, across buying stages. Illustrative English equivalents: "What is the best maintenance software for a mid-sized manufacturing plant in Germany?"; "Compare CMMS tools that host data in the EU." German wording prepared with, or checked by, a competent German reviewer.
- Tools: the AI assistants agreed in advance, on approved tiers and accounts.
- Conditions: date, tool, account state, region and language settings recorded; a fresh conversation per prompt.
- Reporting: "mentioned in N of M prompts on [date] in [tool]", never "invisible in AI search". Answers change by date, account, region and wording.
| Field | Entry format |
|---|---|
| Prompt ID and exact prompt | G-03 / [exact text] |
| Language | de |
| Date and time | [YYYY-MM-DD hh:mm] |
| Tool and interface | [name, web or app, signed in?] |
| Brand mentioned / website cited | [yes / no] / [yes / no, URL] |
| Competitors mentioned | [names as shown] |
| Sources visible in the answer | [URLs as shown] |
| Accuracy of brand description | [accurate / partly / inaccurate, with note] |
| Reproducibility note | [repeat result, if repeated] |
Localization readiness
Page selection and order [H]
| Order | Page | Reason |
|---|---|---|
| 1 | P1 Homepage | Entry point for all German traffic; carries the core description |
| 2 | P2 Product overview | Explains the product once; other pages link to it |
| 3 | P8 Security and data protection | Likely decision criterion for German buyers and IT [H] |
| 4 | P9 Demo request | The conversion point must work in German |
| 5 | P3 Preventive maintenance | Strongest match to the likely primary search theme [H] |
| Later | P5 Pricing | After EUR pricing and contract terms are decided |
| Later | P6, P4, P7, P10 | After first German leads show which themes matter |
LOC-01
No German URL and language metadata
- Observation
- No
/de/structure,lang="de"attribute or hreflang. [S] - Recommendation
- See SEO-02; set the page language attribute on German pages. Illustrative pattern, to be validated during implementation (slug
produktunreviewed):
<link rel="alternate" hreflang="en-GB" href="https://northgauge.example/product" /> <link rel="alternate" hreflang="de-DE" href="https://northgauge.example/de/produkt" /> <link rel="alternate" hreflang="x-default" href="https://northgauge.example/product" />
LOC-02
Text inside screenshots and an English-only interface
- Observation
- Screenshots show English interface text; the app itself is English-only. [S]
- Implication
- A German page showing an English-only product may set wrong expectations. [H]
- Recommendation
- Decide honestly: localize the interface, or state "Interface currently available in English" on German pages.
LOC-03
UK-specific content and formats
- Observation
- GBP prices, UK date formats, UK customer examples and references to UK regulations. [S]
- Recommendation
- Adapt rather than translate: EUR once approved, German date and number formats, UK-only references removed. Do not add German regulatory claims without qualified advice.
LOC-04
No terminology sheet
- Observation
- Product terms vary in English; there is no German glossary. [S]
- Recommendation
- Create a reviewed glossary before any German page is written. Illustrative starting points:
| English term | Candidate German term(s) | Status |
|---|---|---|
| Preventive maintenance | vorbeugende Instandhaltung, präventive Instandhaltung | Illustrative, unreviewed |
| Maintenance software / CMMS | Instandhaltungssoftware, CMMS | Illustrative, unreviewed |
| Inspection checklist | Prüfcheckliste, Inspektionscheckliste | Illustrative, unreviewed |
| Spare parts | Ersatzteile | Illustrative, unreviewed |
| Book a demo | To be written by the reviewer | Not drafted |
LOC-05
Formality and tone undecided
- Observation
- English copy is informal ("you'll love it"). [S]
- Implication
- German B2B software buyers in manufacturing may expect formal address (Sie) and a factual tone. [H]
- Recommendation
- The German reviewer decides formality and tone and records the decision in the terminology sheet.
LOC-06
Legal and company information pages
- Observation
- The site has a UK company information page and a privacy policy. [S]
- Note
- German websites commonly show a legal notice (Impressum). Whether Northgauge needs one, and what it must contain, is a legal question outside this audit. Refer to qualified legal advice. [H]
LOC-07
Human review requirement
- Requirement
- No German customer-facing content is published or treated as final until a demonstrably competent German reviewer familiar with the German market (de-DE) has reviewed it. AI-generated German text is a draft only.
- In this report
- No German marketing copy has been written. All German terms are illustrative and unreviewed.
German search-intent map
No search volumes, keyword difficulty, rankings or traffic forecasts are included. German themes are unreviewed hypotheses [H] to be validated.
| Buying stage / problem | German theme (unreviewed) | Recommended asset | Intent | Business purpose | Validation method |
|---|---|---|---|---|---|
| Awareness: unplanned downtime | ungeplante Stillstände vermeiden | German guide on preventive maintenance | Informational | Early-stage visibility | Autocomplete, "People also ask", competitor content |
| Solution search: category | Instandhaltungssoftware | P2 German product page | Commercial | Main category page | Autocomplete; Google Trends (Germany, relative only); competitor titles |
| Solution search: scheduling | Wartungsplanung Software | P3 German feature page | Commercial | Feature-led entry | As above |
| Solution search: inspections | digitale Checklisten Instandhaltung | P4 German feature page (later) | Commercial | Secondary feature | As above |
| Evaluation: data protection | CMMS DSGVO, Hosting in der EU | P8 German security page | Commercial / trust | Remove a likely objection | Sales-call questions; competitor security pages |
| Evaluation: comparison | CMMS Vergleich | Comparison page (later, verified facts only) | Comparison | Shortlist inclusion | Review of comparison formats in results |
| Decision: price | Instandhaltungssoftware Kosten | P5 German pricing page | Transactional | Qualify budget | Sales feedback; results review |
| Decision: demo | Instandhaltungssoftware Demo | P9 German demo page | Transactional | Conversion | Demo-request tracking (client analytics) |
Every theme must be validated in a real audit and reviewed by a competent German reviewer before it shapes page content.
Competitor landscape
Archetypes, not real companies. No competitor research was conducted. Every cell is a scenario assumption [S].
| A: established German CMMS vendor | B: global enterprise asset-management suite | C: mobile-first inspection app (UK/US) | |
|---|---|---|---|
| Positioning | Local, long-standing, German-language support | Broad platform for large enterprises | Simple, fast inspections |
| German content | Full German site | German site, often translated from English | Limited or none |
| Trust signals | German customer references; German hosting | Analyst reports; large logos | App-store ratings; free plan |
| Likely weakness | Older interface; longer implementation | Complex and costly for mid-sized plants | Narrow scope; weak maintenance planning |
| Northgauge opportunity [H] | Faster setup and modern mobile use, if supported by evidence | Right-sized for mid-sized plants | Broader maintenance coverage beyond inspections |
In a real audit, Locentri would review up to three named competitors' public German pages and record the URL and date of every observation.
Prioritized action plan
"Audit" means the recommendation is delivered in the audit; implementation is separate work by the client or a separate engagement. Success indicators show that work was done; they are not promises of rankings, traffic, AI citations or revenue.
Now: first 30 days
| Action | Rationale | Dependency | Effort [E] | Success indicator | Covered by |
|---|---|---|---|---|---|
Remove /product/ crawl block (SEO-01) | Product pages may be invisible | Web team access | S | Pages crawlable in URL Inspection | Audit recommends; client implements |
Decide /de/ architecture and hreflang (SEO-02) | Avoid rework | Leadership decision | S | Written decision | Audit recommends |
| Arrange German reviewer; terminology and tone sheet (LOC-04, 05, 07) | Consistency and quality | Reviewer availability | M | Reviewed sheet exists | Separate work |
| Agree one product description (GEO-01) | Clarity for buyers and AI assistants | Marketing sign-off | S | Used on all key pages | Audit recommends |
| Decide AI crawler policy (GEO-05) | Deliberate visibility choice | Leadership decision | S | Documented in robots.txt | Audit recommends |
| Measure page experience (SEO-08) | Replace assumptions with data | None | S | Dated PageSpeed records | Part of a real audit |
Next: before German launch
| Action | Rationale | Dependency | Effort [E] | Success indicator | Covered by |
|---|---|---|---|---|---|
Build /de/ structure, switcher, hreflang (SEO-02, LOC-01) | Technical base | Architecture decision | M | Valid, reciprocal hreflang | Separate implementation |
| Validate search themes (chapter 09) | Base pages on real demand signals | Reviewer; public search tools | M | Validated theme list | Separate work |
| Localize the first five pages (chapter 08) | Revenue-closest pages first | Reviewer; terminology sheet | L | Reviewed pages live | Separate engagement |
| Add verified trust content (SEO-09) | Address likely objections | Client confirms facts | M | Facts published and sourced | Client implements |
| Fix titles, H1s, internal links (SEO-03, 04, 06) | Clarity and discoverability | None | S | Crawl shows no duplicates | Client implements |
| Internationalize the demo form (SEO-10) | Avoid losing German leads | Web team | S–M | Test submissions succeed | Client implements |
Later: after launch
| Action | Rationale | Dependency | Effort [E] | Success indicator | Covered by |
|---|---|---|---|---|---|
| Structured data (SEO-07) | Eligibility for search features | Development time | S | Rich Results Test passes | Client implements |
| German expert content and references (GEO-04, 06) | Long-term authority | Reviewer; expertise | M–L | Published, reviewed guides | Separate engagement |
| Repeat AI visibility test | Track change over time | Baseline log | S | Comparable dated log | Possible follow-up |
| Localize remaining pages | Expand coverage | First German leads | L | Reviewed pages live | Separate engagement |
30-day roadmap
The five first-month priorities, sequenced. The week allocation is a planning estimate [E] and depends on team and reviewer availability.
- Week 1Fix the crawl block and confirm indexingRemove the
/product/rule; confirm in Search Console that product pages can be indexed (SEO-01). - Week 1Run a baseline AI visibility testExecute the prompt set in chapter 07 and save the dated log as the reference point.
- Weeks 1–2Agree architecture and page orderDecide
/de/, the hreflang approach and the German page order (SEO-02, chapter 08). - Weeks 2–4Collect the facts German buyers will ask aboutHosting location, security certifications actually held, support hours, contract terms (SEO-09).
- Weeks 2–4Arrange review and terminologyEngage a demonstrably competent German reviewer and agree the terminology and formality sheet (LOC-04, LOC-05, LOC-07).
From demonstration to real audit
In a client report, [S] never appears. Every finding is [M], [V], [E] or [H], supported by the following evidence.
| Illustrative element here | Evidence a real audit requires |
|---|---|
| Scenario website facts | Dated observations of the live site: URL, date, method; exports kept in the findings register |
| robots.txt and crawl findings | Live robots.txt, crawl export within free-tool limits, Search Console data if access is granted |
| Titles, headings, links | Crawl export or page-source checks with dates |
| Page experience | PageSpeed Insights results per page, with dates and device noted |
| German search themes | Recorded search observations with settings and dates; Google Trends comparisons; review by a competent German reviewer |
| AI visibility | Executed prompt log with exact prompts, dates, tools and visible sources |
| Competitor comparison | Up to three named competitors' public German pages, with URLs and dates |
| Trust and product facts | Written confirmation from the client; no unverified claims |
| German terminology | Review by a demonstrably competent German reviewer |
Sources
General guidance only. These sources say nothing about Northgauge.
- Google Search Central, "SEO Starter Guide: The Basics" (page last updated 2025-12-10 UTC according to the page), accessed 2026-10-10. Google SEO Starter Guide
- OpenAI, "Overview of OpenAI Crawlers", accessed 2026-10-10. OpenAI crawler documentation