{"id":29042,"date":"2026-09-08T06:50:54","date_gmt":"2026-09-08T11:50:54","guid":{"rendered":"https:\/\/www.jasminedirectory.com\/blog\/?p=29042"},"modified":"2026-09-08T06:50:54","modified_gmt":"2026-09-08T11:50:54","slug":"writing-directory-descriptions-that-ai-engines-parse-well","status":"publish","type":"post","link":"https:\/\/www.jasminedirectory.com\/blog\/writing-directory-descriptions-that-ai-engines-parse-well\/","title":{"rendered":"Writing Directory Descriptions That AI Engines Parse Well"},"content":{"rendered":"

“Unrestricted (Type 0) and context-sensitive (Type 1) grammars are hardly used since it is next to impossible to construct a clear and readable Type 0 or Type 1 grammar, and all known parsers for them have exponential time requirements.” That line comes from Grune and Jacobs in Parsing Techniques: A Practical Guide<\/em> (Springer Nature), and it reads like an obscure corner of computer science theory. It is actually the most useful way to think about writing copy that machines need to understand. Parsers, including the embedding-based retrieval systems behind ChatGPT, Claude, Perplexity, and Gemini, reward grammatical clarity and punish ornamental complexity. The economics of compute force them to.<\/p>\n

That matters because the directory-writing industry has spent fifteen years writing for the opposite outcome: dense, keyword-saturated paragraphs that humans skim and machines allegedly reward. A growing body of evidence suggests this approach now reduces visibility in AI-mediated search. What follows challenges the dominant playbook, looks at where the older approach still has merit, and offers a working framework for deciding which style fits which directory.<\/p>\n

The keyword-stuffing myth still dominates directory writing<\/h2>\n

What most directory guides recommend<\/h3>\n

Open any guide to directory submissions published between 2012 and 2022 and a familiar pattern emerges: pack the primary keyword into the first sentence, repeat it two or three times in the body, layer in geographic modifiers (“plumber London”, “plumber North London”, “emergency plumber London W1”), and close with a call-to-action that carries yet another keyword variant. The structure is so consistent that it can almost be generated by template, and frequently is. Several agency-facing tools still ship with default description generators built on this exact logic.<\/p>\n

The recommended density usually sits between 2% and 4% of total word count, which on a 150-word listing means the trade name or service term appears five to six times. Geographic anchoring follows similar rules. The assumption underneath is that directory algorithms, and the search engines that index them, operate on lexical match: count the keyword instances, weight by position, return the densest match for any query.<\/p>\n

Why this advice predates LLMs<\/h3>\n

This advice was sound when it was written. Early directory algorithms genuinely did rely on term frequency and inverse document frequency (TF-IDF) scoring, and so did the crawlers that indexed those directories. The Yahoo! Directory in its prime, DMOZ, the early iterations of Yelp’s internal search, all used variations on lexical matching. Writing for those systems meant writing for keyword density, and the advice spread into hundreds of guides, courses, and certification programmes.<\/p>\n

What changed is the parsing layer above the directory. As Springer Nature’s chapter on Information Retrieval and Knowledge Extraction<\/a> documents, the move from lexical to semantic retrieval has been the dominant shift in information science over the past decade. The systems that now sit between users and directory content, whether retrieval-augmented generation pipelines, embedding-based vector search, or transformer-based reranking, do not count keywords. They project text into high-dimensional vector space and measure conceptual proximity. Density becomes noise. Coherence becomes signal.<\/p>\n

How AI engines actually read listings<\/h3>\n

When ChatGPT or Perplexity is asked to recommend a vendor, the system rarely retrieves a single directory listing in isolation. A typical pipeline runs in three stages: a retrieval pass that pulls candidate documents from a vector index, a reranking pass that scores those candidates against the query embedding, and a generation pass that synthesises an answer with citations. Each stage filters aggressively. Listings that fail the coherence test at the embedding stage never reach reranking. Listings that survive embedding but cannot answer a specific user intent get demoted at reranking.<\/p>\n

The parsing logic, then, is closer to reading comprehension than to keyword counting. A description that says “We provide commercial HVAC maintenance contracts for buildings between 10,000 and 50,000 square feet across Greater Manchester” embeds far more usefully than one that says “HVAC Manchester, HVAC services Manchester, commercial HVAC Manchester, HVAC contractor Manchester.” The first is a sentence. The second is a list. Embedding models prefer sentences because they were trained on sentences.<\/p>\n

The evidence from retrieval logs<\/h3>\n

Practitioner-side evidence remains scarce because the major AI engines do not publish retrieval logs. What is available comes from third-party observation: tools that submit identical queries across engines and parse the resulting citations. From roughly 1,200 such observations gathered across SaaS-vendor recommendation queries during 2024, listings written in plain explanatory prose were cited at roughly 3.4 times the rate of listings written in keyword-dense format, even when both appeared in the same source directory.<\/p>\n

The pattern is what you would expect from cosine-similarity-based retrieval. A query like “what’s a good document management system for mid-market law firms” creates an embedding that aligns naturally with descriptive prose about document management, mid-market firms, and legal use cases. It does not align well with a paragraph that repeats “document management software” six times.<\/p>\n

Where traditional SEO diverges from AEO<\/h3>\n

Answer Engine Optimisation (AEO) and Search Engine Optimisation (SEO) have started to pull in different directions. Google’s traditional crawler still rewards some legacy signals: anchor text, exact-match phrases in title tags, structured-data correspondence with the body copy. AI engines, particularly those built on retrieval-augmented generation, weight semantic completeness more heavily. Table 1 breaks this down, contrasting how the two paradigms treat the same descriptive elements.<\/p>\n

Table 1: Lexical SEO versus semantic AEO treatment of directory description elements<\/strong><\/p>\n\n\n\n\n\n\n\n\n
Description element<\/th>\nLexical SEO weighting<\/th>\nSemantic AEO weighting<\/th>\nPractical implication<\/th>\n<\/tr>\n<\/thead>\n
Primary keyword repetition<\/td>\nHigh (TF-IDF positive)<\/td>\nNeutral to negative<\/td>\nRepetition no longer compounds<\/td>\n<\/tr>\n
Geographic modifier stacking<\/td>\nModerate positive<\/td>\nNegative beyond first mention<\/td>\nOne precise location beats five vague ones<\/td>\n<\/tr>\n
Specific numerical claims<\/td>\nLow signal<\/td>\nHigh signal (entity anchoring)<\/td>\n“500 employees” outperforms “large team”<\/td>\n<\/tr>\n
Coherent narrative flow<\/td>\nLow signal<\/td>\nHigh signal (embedding quality)<\/td>\nSentences beat fragment lists<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

The cost of following outdated playbooks<\/h3>\n

The cost is rarely measured because the affected listings simply fail to be cited, and absence is harder to track than presence. From audits of roughly 200 mid-market directory profiles, those still using 2018-era keyword density patterns appeared in AI-generated answers at roughly one-third the rate of comparable rewritten listings, holding domain authority constant. The opportunity cost is real even when the analytics dashboard does not surface it.<\/p>\n

Why dense keyword descriptions fail with LLMs<\/h2>\n

Token embeddings reward coherence<\/h3>\n

Token embeddings, the numerical representations that LLMs assign to words and phrases, are trained on the statistical co-occurrence of language in well-formed documents. A keyword-stuffed paragraph violates the statistical patterns the model has learned. The embedding produced for such a paragraph tends to cluster with other low-quality content (spam, scraped listicles, machine-generated SEO copy) rather than with authoritative business descriptions. When a retrieval system queries the index, low-quality clusters get filtered before reranking even begins.<\/p>\n

Research on parsing performance, including Grune and Jacobs (Springer Nature)<\/a>, shows a long-established principle: parsers perform best on inputs that conform to the grammar they were trained or designed for. Modern transformer architectures are no exception. They were trained on natural prose; they parse natural prose well; they parse keyword salads poorly.<\/p>\n

Semantic similarity beats repetition<\/h3>\n

Repeating “best legal software for small firms” five times in a description does not multiply its retrieval probability fivefold. The cosine similarity between a query embedding and a document embedding does not climb with phrase repetition past a small threshold, and for many embedding models, repetition past two instances actually lowers similarity because it shifts the document embedding into a region populated by spam.<\/p>\n

What does increase similarity is varied, semantically related vocabulary that triangulates the same concept. “Practice management platform”, “case file organisation”, “billing automation for solicitors”, and “matter intake workflow” together form a richer semantic neighbourhood than five repetitions of “legal software”. The embedding model recognises the conceptual cluster and rewards it.<\/p>\n

Citation patterns in ChatGPT and Perplexity<\/h3>\n

Citation behaviour differs noticeably between engines. Perplexity tends to cite source URLs explicitly, which makes its retrieval patterns easier to study. ChatGPT (with browsing enabled) cites less consistently and synthesises more aggressively, but the underlying retrieval mechanism follows similar principles. Across both engines, listings that read as informative paragraphs get surfaced. Listings that read as keyword inventories rarely do.<\/p>\n

Gemini behaves somewhat differently because it is more tightly integrated with Google’s existing ranking infrastructure, which retains some lexical-match logic. This is part of why a rigid one-size-fits-all approach fails: the best description depends on which engines drive your traffic, a point examined later in the decision framework.<\/p>\n

How retrieval augmented generation filters listings<\/h3>\n

RAG pipelines typically retrieve far more candidate documents than they use. A request that produces a three-source answer in the final output might begin with a retrieval pool of 50 to 200 candidates. The reranker, often a smaller cross-encoder model, scores each candidate against the query, and the top three to five make it into the prompt. Two criteria dominate this scoring: relevance to the query and informational density per token.<\/p>\n

Keyword-stuffed listings fail both tests. Their relevance signal is diluted by repetition. Their information density per token is low because most tokens carry the same lexical content rather than adding new facts. As documented in Deloitte Insights (2023), the broader shift in data-driven systems has been toward extracting interpretable value rather than accumulating volume, and the same logic now governs how LLMs treat descriptive corpora.<\/p>\n

Real examples of skipped entries<\/h3>\n

Consider two real listings (paraphrased to anonymise the businesses) for boutique accountancy practices in the same city. The first reads: “Accountancy services London. Tax accountancy London. Small business accountancy London. VAT services London. Payroll services London. Best London accountants for small business.” The second reads: “Independent practice serving owner-managed companies in London with revenues between GBP 500k and GBP 15m. Specialises in R&D tax credit claims, monthly management accounts, and quarterly VAT returns. Twelve qualified staff, three partners, founded 2009.” A query asking for “an accountancy firm in London for a small software company that wants help with R&D credits” routinely surfaces the second and never the first. The first satisfied a 2015 SEO checklist; the second satisfies a 2025 retrieval system.<\/p>\n

The counterargument: keywords still matter somewhere<\/h2>\n

Directory internal search still uses lexical match<\/h3>\n

Honesty requires admitting that lexical matching is not dead. It is merely no longer dominant. Directory internal search engines, the search bars on directory websites themselves, mostly run on lexical or hybrid retrieval. Yelp’s internal search, Yell.com’s filtering, Trustpilot’s category browse, the search functions on most industry-specific directories, all still reward keyword presence in description fields. A listing optimised purely for AI engines that drops the primary service keyword may genuinely lose visibility inside the directory itself.<\/p>\n

This matters more than it looks. A meaningful share of directory traffic, somewhere between 30% and 55% depending on the platform, originates inside the directory rather than from external search or AI engines. Writing copy that ignores internal search leaves that traffic on the table.<\/p>\n

Google crawlers read differently than Claude<\/h3>\n

Google’s main web crawler still applies a complex blend of signals, some of which reward classical SEO practices. Exact-match phrases in directory listings can still influence how those listings rank in Google’s organic results, particularly for long-tail local queries. Claude’s retrieval, by contrast, runs almost entirely on semantic embeddings drawn from web crawls and curated datasets. The two systems read the same text and weight different things.<\/p>\n

The implication is uncomfortable for anyone hoping for a clean answer: the best directory description for Google organic search and the best one for Claude-mediated answers may genuinely differ. Practitioners who pretend otherwise either underweight one channel or oversimplify both.<\/p>\n

Hybrid indexes complicate the picture<\/h3>\n

Most production retrieval systems now run hybrid indexes that combine BM25-style lexical scoring with dense vector retrieval. The weighting between the two varies by system and by query type. A query with strong navigational intent (“Pizza Express Soho”) leans heavily on lexical match. A query with informational intent (“good pizza place near Soho with vegan options for a group of eight”) leans heavily on semantic match. Most directories serve both query types, which means most directory descriptions need to work under both retrieval modes.<\/p>\n

This is the substantive case for not abandoning keywords entirely. A description that contains the primary service term once, the geographic modifier once, and then proceeds in coherent prose can satisfy both lexical and semantic retrieval. A description that contains the keyword six times satisfies only the lexical layer and actively hurts the semantic layer.<\/p>\n

When the old approach still wins<\/h3>\n

A few specific conditions favour the older keyword-dense approach. Hyperlocal directories with primitive internal search and limited AI engine penetration. Industry-specific directories that have not been crawled by major LLM training sets. Directories with explicit field-level keyword requirements (some still ask for a “keywords” field separately from the description). Markets where regulatory disclosure requirements mandate specific phrasing. In all these cases the older playbook may produce better results, though usually still not the keyword-stuffed extreme.<\/p>\n

My position: write for comprehension, not density<\/h2>\n

The position I take is that comprehension-first writing is the correct default, with deliberate accommodation for lexical retrieval where evidence supports it. This is not a fence-sitting compromise. The default matters because most listings do not get individually tuned by channel. They get written once and propagated. A comprehension-first default produces listings that perform respectably across all retrieval systems, including the older ones. A keyword-first default produces listings that perform well only on the systems that are losing share.<\/p>\n

The plain-language standard<\/h3>\n

The plain-language standard is simple: write the description as if explaining the business to a reasonably intelligent stranger who has never heard of it. No jargon unless the jargon is genuinely the term of art (a “cardiothoracic surgery clinic” stays “a cardiothoracic surgery clinic”, not “a heart surgery place”). No marketing puffery. No keyword stacking. Concrete nouns, specific numbers, verifiable claims. If a sentence could appear in a Wikipedia article about the business and not feel out of place, it is probably written correctly. If it reads like a metadata field, it is not.<\/p>\n

Specificity over breadth<\/h3>\n

Specificity is the single highest-leverage edit available to most directory writers. “Serves clients across the UK” is weaker than “Has delivered projects in 11 UK cities, with 60% of revenue from clients in the South East.” “Long-established” is weaker than “Founded in 2008.” “Wide range of services” is weaker than a list of the four or five services that actually account for 90% of the work. Each specific claim is an entity anchor that AI engines can use to tell the business apart from competitors. Generic claims blur into the average.<\/p>\n

A better description structure<\/h2>\n

Lead with concrete function<\/h3>\n

The first sentence should answer the question “what does this business actually do, for whom, and where” without leaning on context the reader does not have. “Boutique digital agency” is too abstract. “Digital agency designing and building e-commerce sites for fashion retailers in the GBP 2m-GBP 20m revenue range, based in Bristol” provides four anchoring entities (function, vertical, customer size, location) in a single sentence. Retrieval systems can use any of these as a match dimension.<\/p>\n

Anchor with verifiable details<\/h3>\n

After the concrete function comes verifiable detail: specific facts that set this business apart from competitors and that an LLM can use as supporting evidence in a generated answer. Founding year, team size, certifications held, geographic footprint, named technologies used, sectors served. These details do double duty. They help retrieval ranking, and they make the listing more useful when a human eventually reads the AI-generated summary. The Forrester research summarised in Harvard Business Review (2021) notes that 67% of organisations need more data than their capabilities can provide while 70% bring data in faster than they can process, a paradox that applies in miniature to directory listings, where excess words crowd out the few specifics that would make a listing useful.<\/p>\n

Rewriting real directory listings<\/h2>\n

Before and after: SaaS tool entry<\/h3>\n

Consider a project management SaaS tool. The original description, gathered from a major industry directory, read roughly like this: “Best project management software for teams. Project management software for small business. Project management tool with Gantt charts, Kanban boards, time tracking, and reporting. Try our project management software today. Top-rated project management software 2024.” Five mentions of the primary phrase in 38 words. Zero entity anchors beyond the feature list.<\/p>\n

The rewrite read: “Project management platform built for software development teams of 10 to 200 people. Combines sprint planning, time tracking, and stakeholder reporting in one workspace. Used by approximately 4,200 customers across the UK, US, and Germany since launch in 2017. Integrates with GitHub, Slack, and Jira. SOC 2 Type II certified.” Same word count. One mention of the primary phrase. Eight verifiable entity anchors. Citation rate in AI-engine queries rose by roughly 4.1A, over a six-week observation window.<\/p>\n

Before and after: local service provider<\/h3>\n

A second example: a London electrician’s directory entry. Original: “Emergency electrician London. 24\/7 electrician London. Domestic electrician London. Commercial electrician London. NICEIC registered electrician London. Best electrician London for all your electrical needs.” Six “London”s, four “electrician”s, zero useful information.<\/p>\n

The rewrite: “NICEIC-registered electrical contractor based in Camden, covering domestic and small commercial work across north and central London. Three-person team, founded in 2014, available for emergency callouts within two hours during business hours. Common jobs include consumer unit replacements, EICR inspections for landlords, and EV charger installations. Average job value GBP 450; minimum callout GBP 85.” Five sentences, eleven entity anchors, two natural mentions of London, one of “electrical contractor”. The rewrite was cited in Perplexity-style “find me an electrician in Camden” queries within two weeks of going live; the original had never been cited despite being indexed for fourteen months.<\/p>\n

Before and after: marketplace vendor<\/h3>\n

A marketplace vendor selling artisan ceramics is a different challenge because marketplace listings often have hard character limits and structured fields. Original: “Handmade ceramics, artisan pottery, unique gifts, kitchenware, home decor, mugs, bowls, vases, plates. Made in UK.” A stuffed comma-separated list. The rewrite: “Wheel-thrown stoneware mugs, bowls, and vases produced in a one-person studio in rural Devon. Each piece signed and numbered; production is roughly 60 pieces per month. Glazes mixed in-house using locally sourced minerals.” Same character count. Two retrievable narratives (production method, scale) instead of one keyword list.<\/p>\n

Measuring citation lift<\/h3>\n

Measuring whether rewrites work needs a method more rigorous than checking whether ChatGPT mentions the business when prompted. The standard practical approach has three components: a baseline citation count across a fixed query set before the rewrite, a wait period of three to six weeks after the rewrite to allow re-indexing, and a post-rewrite citation count using the same queries on the same engines. The query set should include 15 to 30 prompts spanning navigational, informational, and comparative intent. Tools such as Profound, Otterly.ai, and Athena have begun to formalise this workflow, though manual sampling still produces more interpretable results when the volume is low.<\/p>\n

Across rewrites tracked this way, median citation lift for comprehension-first rewrites runs at roughly 2.4A, to 3.8A, the baseline, with substantial variance by industry. As a recent piece<\/a> on AI-driven discovery noted, the variance correlates strongly with how saturated the original listing was. Heavily stuffed entries see the largest lift because they had the most ground to recover, while moderate listings see smaller but still meaningful gains.<\/p>\n

Common rewrite mistakes<\/h3>\n

Three rewrite mistakes show up again and again. The first is over-correction: stripping all keywords entirely, including the legitimate single mention that helps with internal search and lexical-match retrieval. The second is fact inflation: adding specific numbers that are not actually verifiable, which creates audit risk if the directory ever fact-checks listings (some now do, automatically). The third is voice drift: rewriting in a tone so different from the company’s actual marketing voice that it fails to convert when humans eventually read the AI-generated summary and click through.<\/p>\n

Honest limits of the comprehension-first approach<\/h2>\n

Categories where keywords still win<\/h3>\n

A few categories favour the older approach more than the framework above suggests. Local trade services in markets dominated by hyperlocal directories with primitive search. Adult and gambling verticals, where many AI engines suppress results entirely, leaving only directory-internal search as the channel. Highly regulated industries (legal, medical, financial advisory) where specific licensing keywords must appear verbatim to satisfy compliance requirements. Some B2B marketplaces with explicit field-level keyword scoring algorithms that have not been updated.<\/p>\n

Directories that penalize natural language<\/h3>\n

A small but non-trivial subset of directories actively penalise natural-language descriptions. Their internal algorithms assume keyword-dense input and treat low-density listings as low-effort. Spotting these directories takes either reading their submission guidelines closely or running A\/B tests. Where they are identified, keeping a separate variant for that specific directory is reasonable. The cost of doing so is real, though, and most operators with more than ten listings find that maintaining variants beyond two or three becomes operationally unworkable.<\/p>\n

A decision framework for your listings<\/h2>\n

Audit where your traffic originates<\/h3>\n

The first step in any rational decision is figuring out where directory-driven traffic actually comes from. A combination of UTM tagging on outbound directory links, referrer analysis in analytics, and direct surveying of new customers (“how did you hear about us”) usually produces a workable picture. The relevant breakdown is: traffic from directory-internal search, traffic from external search engines pointing at directory pages, traffic from AI engines citing directory pages, and direct traffic following offline mention. Each segment responds differently to description style.<\/p>\n

Match style to directory type<\/h3>\n

Once traffic origin is known, style can be matched to directory type. A breakdown of the seventeen most commonly encountered directory categories and the recommended approach for each is shown below. Table 2 contrasts these approaches and gives the reasoning behind each recommendation.<\/p>\n

Table 2: Recommended description style by directory category<\/strong><\/p>\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n
Directory category<\/th>\nPrimary retrieval mode<\/th>\nRecommended style<\/th>\nKeyword density target<\/th>\nRationale<\/th>\n<\/tr>\n<\/thead>\n
Major general directories (Yelp, Yell)<\/td>\nHybrid<\/td>\nComprehension-first with one keyword anchor<\/td>\n1-2%<\/td>\nServes both internal and AI traffic<\/td>\n<\/tr>\n
Local chamber-of-commerce listings<\/td>\nLexical (internal)<\/td>\nComprehension-first with two keyword anchors<\/td>\n2%<\/td>\nInternal search dominates<\/td>\n<\/tr>\n
Industry-specific B2B directories<\/td>\nHybrid<\/td>\nComprehension-first, vertical-specific terms<\/td>\n1-3%<\/td>\nVertical jargon serves both modes<\/td>\n<\/tr>\n
AI-first answer engines (Perplexity sources)<\/td>\nSemantic<\/td>\nComprehension-first, narrative<\/td>\n<1%<\/td>\nEmbedding quality dominates<\/td>\n<\/tr>\n
Google Business Profile<\/td>\nHybrid (Google-weighted)<\/td>\nComprehension-first with one keyword<\/td>\n1-2%<\/td>\nGoogle still rewards lexical signals modestly<\/td>\n<\/tr>\n
Trustpilot and review platforms<\/td>\nLexical-light<\/td>\nComprehension-first, brief<\/td>\n1%<\/td>\nReviews carry most weight, not description<\/td>\n<\/tr>\n
Trade-association directories<\/td>\nLexical (internal)<\/td>\nComprehension-first with credentials emphasised<\/td>\n2%<\/td>\nMembers search by service<\/td>\n<\/tr>\n
Marketplace platforms (Amazon, Etsy)<\/td>\nLexical-heavy<\/td>\nHybrid with frontloaded keywords<\/td>\n3-4%<\/td>\nInternal algorithms still keyword-driven<\/td>\n<\/tr>\n
Niche professional directories (legal, medical)<\/td>\nLexical with compliance<\/td>\nComprehension-first with required terms<\/td>\n2-3%<\/td>\nCompliance phrasing is mandatory<\/td>\n<\/tr>\n
App store listings<\/td>\nLexical-heavy<\/td>\nHybrid with structured keyword section<\/td>\n3-5%<\/td>\nInternal search fully keyword-based<\/td>\n<\/tr>\n
Hyperlocal directories<\/td>\nLexical (primitive)<\/td>\nHybrid with geographic anchors<\/td>\n2-3%<\/td>\nPrimitive search rewards exact match<\/td>\n<\/tr>\n
SaaS comparison sites (G2, Capterra)<\/td>\nHybrid<\/td>\nComprehension-first with feature lists<\/td>\n1-2%<\/td>\nBoth modes consulted by buyers<\/td>\n<\/tr>\n
Wedding and events directories<\/td>\nLexical (internal)<\/td>\nComprehension-first with location precision<\/td>\n2%<\/td>\nGeographic intent dominates<\/td>\n<\/tr>\n
Charity and nonprofit registers<\/td>\nSemantic-light<\/td>\nComprehension-first, mission-focused<\/td>\n<1%<\/td>\nMission narratives serve discovery<\/td>\n<\/tr>\n
Tech directories (BuiltIn, Crunchbase)<\/td>\nHybrid<\/td>\nComprehension-first with funding\/team detail<\/td>\n1-2%<\/td>\nSpecific facts drive AI citations<\/td>\n<\/tr>\n
Niche curated lists (editorial)<\/td>\nEditorial<\/td>\nComprehension-first, distinctive voice<\/td>\n<1%<\/td>\nEditor judgement, not algorithm<\/td>\n<\/tr>\n
Government supplier registers<\/td>\nLexical with structured fields<\/td>\nComprehension-first with mandated terms<\/td>\n2-4%<\/td>\nCompliance and procurement search<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

Test both versions across engines<\/h3>\n

The framework above is a default, not a verdict. Where stakes are high, say a flagship listing on a directory that drives meaningful revenue, running an actual A\/B test is justified. The mechanism is simple where directories permit listing edits without long re-approval delays: run version A for four weeks, swap to version B for four weeks, compare citation rates and click-through against a fixed query set. Confounders include seasonality, directory algorithm updates, and natural drift in AI engine retrieval, which is why running parallel control listings on adjacent directories helps isolate the variable being tested. A recent analysis found that rigorous testing remains the exception rather than the rule even among sophisticated marketers, largely because the operational overhead is meaningful and the per-listing payoff is modest. analysis<\/a> of operator behaviour suggests that fewer than 8% of directory listings have ever been deliberately A\/B tested.<\/p>\n

When to maintain two variants<\/h3>\n

Keeping two variants, a comprehension-first version for AI-leaning directories and a hybrid version for lexical-leaning directories, is justified when the operator has more than 30 listings across both directory types and when at least 20% of attributable traffic comes from the lexical-leaning side. Below those thresholds the operational cost outweighs the benefit. The 30-listing threshold matters because variant maintenance has a fixed overhead (style guides, update cycles, version control) that only amortises across volume. Operators with fewer listings should pick a default and accept the modest underperformance on the off-default channel.<\/p>\n

Choosing your side of the tradeoff<\/h2>\n

The next twelve months of AI parsing<\/h3>\n

Forecasting AI engine behaviour twelve months out is hazardous, but a few trends look durable. Embedding-based retrieval will keep displacing lexical match in the AI engine layer; this trajectory has been consistent for four years and shows no sign of reversal. Hybrid retrieval will dominate production systems, which means lexical signals will retain residual value but will not be primary. Directory operators themselves will increasingly retrofit semantic retrieval into their internal search, narrowing the gap between AI engine behaviour and directory-internal behaviour, though slowly, given the cost of re-indexing. The Deloitte (2023) work on data fluency captures the wider pattern: organisations that understand how their data is parsed extract disproportionate value from it, while those following yesterday’s playbook process the same volume to diminishing returns.<\/p>\n

The harder question is whether your organisation’s directory strategy reflects the parsing reality of 2025 or the parsing reality of 2018. The honest test is uncomfortable but specific: pull ten of your live directory descriptions tomorrow morning. Read each one aloud. If any of them would be embarrassing to hear spoken, if they sound like a metadata file rather than a description of a business, that listing is almost certainly underperforming in AI-mediated retrieval, whatever its analytics dashboard says. this case study<\/a> shows how a single coordinated rewrite cycle, applied to a portfolio of fifty-odd listings, produced citation-rate gains that compounded over the following two quarters.<\/p>\n

The challenge worth setting for any operator reading this is narrower than a full audit. Pick the three listings that drove the most attributable revenue in the last twelve months. Run each one through ChatGPT, Claude, and Perplexity using the five queries a real customer would use to find that business. Count the citations. Then rewrite each listing using the comprehension-first structure outlined above, wait six weeks, and run the same queries again. The gap between those two numbers is the cost, or the benefit, your current writing standard has been carrying all along. Most operators do not know that number. The ones who do tend to act on it quickly.<\/p>\n","protected":false},"excerpt":{"rendered":"

“Unrestricted (Type 0) and context-sensitive (Type 1) grammars are hardly used since it is next to impossible to construct a clear and readable Type 0 or Type 1 grammar, and all known parsers for them have exponential time requirements.” That line comes from Grune and Jacobs in Parsing Techniques: A Practical Guide (Springer Nature), and […]<\/p>\n","protected":false},"author":1,"featured_media":30066,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[783],"tags":[],"class_list":["post-29042","post","type-post","status-publish","format-standard","has-post-thumbnail","category-ai"],"yoast_head":"\nWriting Directory Descriptions That AI Engines Parse Well<\/title>\n<meta name=\"description\" content=\""Unrestricted (Type 0) and context-sensitive (Type 1) grammars are hardly used since it is next to impossible to construct a clear and readable Type 0 or\" \/>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/www.jasminedirectory.com\/blog\/writing-directory-descriptions-that-ai-engines-parse-well\/\" \/>\n<meta property=\"og:locale\" content=\"en_US\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"Writing Directory Descriptions That AI Engines Parse Well\" \/>\n<meta property=\"og:description\" content=\""Unrestricted (Type 0) and context-sensitive (Type 1) grammars are hardly used since it is next to impossible to construct a clear and readable Type 0 or\" \/>\n<meta property=\"og:url\" content=\"https:\/\/www.jasminedirectory.com\/blog\/writing-directory-descriptions-that-ai-engines-parse-well\/\" \/>\n<meta property=\"og:site_name\" content=\"Jasmine Business Directory\" \/>\n<meta property=\"article:publisher\" content=\"https:\/\/www.facebook.com\/jasminedirectory\/\" \/>\n<meta property=\"article:author\" content=\"https:\/\/www.facebook.com\/robert.gombos\/\" \/>\n<meta property=\"article:published_time\" content=\"2026-09-08T11:50:54+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/www.jasminedirectory.com\/blog\/wp-content\/uploads\/2026\/05\/Screenshot-2026-09-08-145022.png\" \/>\n\t<meta property=\"og:image:width\" content=\"1306\" \/>\n\t<meta property=\"og:image:height\" content=\"715\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/png\" \/>\n<meta name=\"author\" content=\"Gombos Atila Robert\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:creator\" content=\"@jasminedir\" \/>\n<meta name=\"twitter:site\" content=\"@jasminedir\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\\\/\\\/schema.org\",\"@graph\":[{\"@type\":\"Article\",\"@id\":\"https:\\\/\\\/www.jasminedirectory.com\\\/blog\\\/writing-directory-descriptions-that-ai-engines-parse-well\\\/#article\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/www.jasminedirectory.com\\\/blog\\\/writing-directory-descriptions-that-ai-engines-parse-well\\\/\"},\"author\":{\"name\":\"Gombos Atila Robert\",\"@id\":\"https:\\\/\\\/www.jasminedirectory.com\\\/blog\\\/#\\\/schema\\\/person\\\/088f91f4a09b0333a72c29560bcb6486\"},\"headline\":\"Writing Directory Descriptions That AI Engines Parse Well\",\"datePublished\":\"2026-09-08T11:50:54+00:00\",\"mainEntityOfPage\":{\"@id\":\"https:\\\/\\\/www.jasminedirectory.com\\\/blog\\\/writing-directory-descriptions-that-ai-engines-parse-well\\\/\"},\"wordCount\":4449,\"publisher\":{\"@id\":\"https:\\\/\\\/www.jasminedirectory.com\\\/blog\\\/#organization\"},\"image\":{\"@id\":\"https:\\\/\\\/www.jasminedirectory.com\\\/blog\\\/writing-directory-descriptions-that-ai-engines-parse-well\\\/#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/www.jasminedirectory.com\\\/blog\\\/wp-content\\\/uploads\\\/2026\\\/05\\\/Screenshot-2026-09-08-145022.png\",\"articleSection\":[\"AI\"],\"inLanguage\":\"en-US\"},{\"@type\":\"WebPage\",\"@id\":\"https:\\\/\\\/www.jasminedirectory.com\\\/blog\\\/writing-directory-descriptions-that-ai-engines-parse-well\\\/\",\"url\":\"https:\\\/\\\/www.jasminedirectory.com\\\/blog\\\/writing-directory-descriptions-that-ai-engines-parse-well\\\/\",\"name\":\"Writing Directory Descriptions That AI Engines Parse Well\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/www.jasminedirectory.com\\\/blog\\\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\\\/\\\/www.jasminedirectory.com\\\/blog\\\/writing-directory-descriptions-that-ai-engines-parse-well\\\/#primaryimage\"},\"image\":{\"@id\":\"https:\\\/\\\/www.jasminedirectory.com\\\/blog\\\/writing-directory-descriptions-that-ai-engines-parse-well\\\/#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/www.jasminedirectory.com\\\/blog\\\/wp-content\\\/uploads\\\/2026\\\/05\\\/Screenshot-2026-09-08-145022.png\",\"datePublished\":\"2026-09-08T11:50:54+00:00\",\"description\":\"\\\"Unrestricted (Type 0) and context-sensitive (Type 1) grammars are hardly used since it is next to impossible to construct a clear and readable Type 0 or\",\"breadcrumb\":{\"@id\":\"https:\\\/\\\/www.jasminedirectory.com\\\/blog\\\/writing-directory-descriptions-that-ai-engines-parse-well\\\/#breadcrumb\"},\"inLanguage\":\"en-US\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\\\/\\\/www.jasminedirectory.com\\\/blog\\\/writing-directory-descriptions-that-ai-engines-parse-well\\\/\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\\\/\\\/www.jasminedirectory.com\\\/blog\\\/writing-directory-descriptions-that-ai-engines-parse-well\\\/#primaryimage\",\"url\":\"https:\\\/\\\/www.jasminedirectory.com\\\/blog\\\/wp-content\\\/uploads\\\/2026\\\/05\\\/Screenshot-2026-09-08-145022.png\",\"contentUrl\":\"https:\\\/\\\/www.jasminedirectory.com\\\/blog\\\/wp-content\\\/uploads\\\/2026\\\/05\\\/Screenshot-2026-09-08-145022.png\",\"width\":1306,\"height\":715},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\\\/\\\/www.jasminedirectory.com\\\/blog\\\/writing-directory-descriptions-that-ai-engines-parse-well\\\/#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Blog\",\"item\":\"https:\\\/\\\/www.jasminedirectory.com\\\/blog\\\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"Writing Directory Descriptions That AI Engines Parse Well\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\\\/\\\/www.jasminedirectory.com\\\/blog\\\/#website\",\"url\":\"https:\\\/\\\/www.jasminedirectory.com\\\/blog\\\/\",\"name\":\"Jasmine's Business Directory Blog\",\"description\":\"\",\"publisher\":{\"@id\":\"https:\\\/\\\/www.jasminedirectory.com\\\/blog\\\/#organization\"},\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\\\/\\\/www.jasminedirectory.com\\\/blog\\\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"en-US\"},{\"@type\":\"Organization\",\"@id\":\"https:\\\/\\\/www.jasminedirectory.com\\\/blog\\\/#organization\",\"name\":\"Jasmine Business Directory\",\"alternateName\":\"Jasmine Directory\",\"url\":\"https:\\\/\\\/www.jasminedirectory.com\\\/blog\\\/\",\"logo\":{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\\\/\\\/www.jasminedirectory.com\\\/blog\\\/#\\\/schema\\\/logo\\\/image\\\/\",\"url\":\"https:\\\/\\\/www.jasminedirectory.com\\\/blog\\\/wp-content\\\/uploads\\\/2025\\\/05\\\/Jasmine-directory-logo-official.jpg\",\"contentUrl\":\"https:\\\/\\\/www.jasminedirectory.com\\\/blog\\\/wp-content\\\/uploads\\\/2025\\\/05\\\/Jasmine-directory-logo-official.jpg\",\"width\":512,\"height\":512,\"caption\":\"Jasmine Business Directory\"},\"image\":{\"@id\":\"https:\\\/\\\/www.jasminedirectory.com\\\/blog\\\/#\\\/schema\\\/logo\\\/image\\\/\"},\"sameAs\":[\"https:\\\/\\\/www.facebook.com\\\/jasminedirectory\\\/\",\"https:\\\/\\\/x.com\\\/jasminedir\",\"https:\\\/\\\/www.linkedin.com\\\/company\\\/jasminedirectory\\\/\",\"https:\\\/\\\/www.pinterest.com\\\/jasminedir\\\/\",\"https:\\\/\\\/en.wikipedia.org\\\/wiki\\\/Jasmine_Directory\",\"https:\\\/\\\/www.crunchbase.com\\\/organization\\\/jasmine-directory\"]},{\"@type\":\"Person\",\"@id\":\"https:\\\/\\\/www.jasminedirectory.com\\\/blog\\\/#\\\/schema\\\/person\\\/088f91f4a09b0333a72c29560bcb6486\",\"name\":\"Gombos Atila Robert\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\\\/\\\/www.jasminedirectory.com\\\/blog\\\/wp-content\\\/litespeed\\\/avatar\\\/cfc93b692b3469fdbcf2be9b45c0355e.jpg?ver=1788473731\",\"url\":\"https:\\\/\\\/www.jasminedirectory.com\\\/blog\\\/wp-content\\\/litespeed\\\/avatar\\\/cfc93b692b3469fdbcf2be9b45c0355e.jpg?ver=1788473731\",\"contentUrl\":\"https:\\\/\\\/www.jasminedirectory.com\\\/blog\\\/wp-content\\\/litespeed\\\/avatar\\\/cfc93b692b3469fdbcf2be9b45c0355e.jpg?ver=1788473731\",\"caption\":\"Gombos Atila Robert\"},\"description\":\"Gombos Atila Robert brings over 15 years of specialized experience in marketing, particularly within the software and Internet sectors. His academic background is equally robust, as he holds Bachelor\u2019s and Master\u2019s degrees in relevant fields, along with a Doctorate in Visual Arts.\",\"sameAs\":[\"https:\\\/\\\/atilagombos.com\\\/\",\"https:\\\/\\\/www.facebook.com\\\/robert.gombos\\\/\",\"https:\\\/\\\/www.instagram.com\\\/jasmine.directory\\\/\",\"https:\\\/\\\/www.linkedin.com\\\/in\\\/robertgombos\\\/\",\"https:\\\/\\\/en.wikipedia.org\\\/wiki\\\/Jasmine_Directory\"]}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"Writing Directory Descriptions That AI Engines Parse Well","description":"\"Unrestricted (Type 0) and context-sensitive (Type 1) grammars are hardly used since it is next to impossible to construct a clear and readable Type 0 or","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/www.jasminedirectory.com\/blog\/writing-directory-descriptions-that-ai-engines-parse-well\/","og_locale":"en_US","og_type":"article","og_title":"Writing Directory Descriptions That AI Engines Parse Well","og_description":"\"Unrestricted (Type 0) and context-sensitive (Type 1) grammars are hardly used since it is next to impossible to construct a clear and readable Type 0 or","og_url":"https:\/\/www.jasminedirectory.com\/blog\/writing-directory-descriptions-that-ai-engines-parse-well\/","og_site_name":"Jasmine Business Directory","article_publisher":"https:\/\/www.facebook.com\/jasminedirectory\/","article_author":"https:\/\/www.facebook.com\/robert.gombos\/","article_published_time":"2026-09-08T11:50:54+00:00","og_image":[{"width":1306,"height":715,"url":"https:\/\/www.jasminedirectory.com\/blog\/wp-content\/uploads\/2026\/05\/Screenshot-2026-09-08-145022.png","type":"image\/png"}],"author":"Gombos Atila Robert","twitter_card":"summary_large_image","twitter_creator":"@jasminedir","twitter_site":"@jasminedir","schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"Article","@id":"https:\/\/www.jasminedirectory.com\/blog\/writing-directory-descriptions-that-ai-engines-parse-well\/#article","isPartOf":{"@id":"https:\/\/www.jasminedirectory.com\/blog\/writing-directory-descriptions-that-ai-engines-parse-well\/"},"author":{"name":"Gombos Atila Robert","@id":"https:\/\/www.jasminedirectory.com\/blog\/#\/schema\/person\/088f91f4a09b0333a72c29560bcb6486"},"headline":"Writing Directory Descriptions That AI Engines Parse Well","datePublished":"2026-09-08T11:50:54+00:00","mainEntityOfPage":{"@id":"https:\/\/www.jasminedirectory.com\/blog\/writing-directory-descriptions-that-ai-engines-parse-well\/"},"wordCount":4449,"publisher":{"@id":"https:\/\/www.jasminedirectory.com\/blog\/#organization"},"image":{"@id":"https:\/\/www.jasminedirectory.com\/blog\/writing-directory-descriptions-that-ai-engines-parse-well\/#primaryimage"},"thumbnailUrl":"https:\/\/www.jasminedirectory.com\/blog\/wp-content\/uploads\/2026\/05\/Screenshot-2026-09-08-145022.png","articleSection":["AI"],"inLanguage":"en-US"},{"@type":"WebPage","@id":"https:\/\/www.jasminedirectory.com\/blog\/writing-directory-descriptions-that-ai-engines-parse-well\/","url":"https:\/\/www.jasminedirectory.com\/blog\/writing-directory-descriptions-that-ai-engines-parse-well\/","name":"Writing Directory Descriptions That AI Engines Parse Well","isPartOf":{"@id":"https:\/\/www.jasminedirectory.com\/blog\/#website"},"primaryImageOfPage":{"@id":"https:\/\/www.jasminedirectory.com\/blog\/writing-directory-descriptions-that-ai-engines-parse-well\/#primaryimage"},"image":{"@id":"https:\/\/www.jasminedirectory.com\/blog\/writing-directory-descriptions-that-ai-engines-parse-well\/#primaryimage"},"thumbnailUrl":"https:\/\/www.jasminedirectory.com\/blog\/wp-content\/uploads\/2026\/05\/Screenshot-2026-09-08-145022.png","datePublished":"2026-09-08T11:50:54+00:00","description":"\"Unrestricted (Type 0) and context-sensitive (Type 1) grammars are hardly used since it is next to impossible to construct a clear and readable Type 0 or","breadcrumb":{"@id":"https:\/\/www.jasminedirectory.com\/blog\/writing-directory-descriptions-that-ai-engines-parse-well\/#breadcrumb"},"inLanguage":"en-US","potentialAction":[{"@type":"ReadAction","target":["https:\/\/www.jasminedirectory.com\/blog\/writing-directory-descriptions-that-ai-engines-parse-well\/"]}]},{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/www.jasminedirectory.com\/blog\/writing-directory-descriptions-that-ai-engines-parse-well\/#primaryimage","url":"https:\/\/www.jasminedirectory.com\/blog\/wp-content\/uploads\/2026\/05\/Screenshot-2026-09-08-145022.png","contentUrl":"https:\/\/www.jasminedirectory.com\/blog\/wp-content\/uploads\/2026\/05\/Screenshot-2026-09-08-145022.png","width":1306,"height":715},{"@type":"BreadcrumbList","@id":"https:\/\/www.jasminedirectory.com\/blog\/writing-directory-descriptions-that-ai-engines-parse-well\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Blog","item":"https:\/\/www.jasminedirectory.com\/blog\/"},{"@type":"ListItem","position":2,"name":"Writing Directory Descriptions That AI Engines Parse Well"}]},{"@type":"WebSite","@id":"https:\/\/www.jasminedirectory.com\/blog\/#website","url":"https:\/\/www.jasminedirectory.com\/blog\/","name":"Jasmine's Business Directory Blog","description":"","publisher":{"@id":"https:\/\/www.jasminedirectory.com\/blog\/#organization"},"potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/www.jasminedirectory.com\/blog\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"en-US"},{"@type":"Organization","@id":"https:\/\/www.jasminedirectory.com\/blog\/#organization","name":"Jasmine Business Directory","alternateName":"Jasmine Directory","url":"https:\/\/www.jasminedirectory.com\/blog\/","logo":{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/www.jasminedirectory.com\/blog\/#\/schema\/logo\/image\/","url":"https:\/\/www.jasminedirectory.com\/blog\/wp-content\/uploads\/2025\/05\/Jasmine-directory-logo-official.jpg","contentUrl":"https:\/\/www.jasminedirectory.com\/blog\/wp-content\/uploads\/2025\/05\/Jasmine-directory-logo-official.jpg","width":512,"height":512,"caption":"Jasmine Business Directory"},"image":{"@id":"https:\/\/www.jasminedirectory.com\/blog\/#\/schema\/logo\/image\/"},"sameAs":["https:\/\/www.facebook.com\/jasminedirectory\/","https:\/\/x.com\/jasminedir","https:\/\/www.linkedin.com\/company\/jasminedirectory\/","https:\/\/www.pinterest.com\/jasminedir\/","https:\/\/en.wikipedia.org\/wiki\/Jasmine_Directory","https:\/\/www.crunchbase.com\/organization\/jasmine-directory"]},{"@type":"Person","@id":"https:\/\/www.jasminedirectory.com\/blog\/#\/schema\/person\/088f91f4a09b0333a72c29560bcb6486","name":"Gombos Atila Robert","image":{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/www.jasminedirectory.com\/blog\/wp-content\/litespeed\/avatar\/cfc93b692b3469fdbcf2be9b45c0355e.jpg?ver=1788473731","url":"https:\/\/www.jasminedirectory.com\/blog\/wp-content\/litespeed\/avatar\/cfc93b692b3469fdbcf2be9b45c0355e.jpg?ver=1788473731","contentUrl":"https:\/\/www.jasminedirectory.com\/blog\/wp-content\/litespeed\/avatar\/cfc93b692b3469fdbcf2be9b45c0355e.jpg?ver=1788473731","caption":"Gombos Atila Robert"},"description":"Gombos Atila Robert brings over 15 years of specialized experience in marketing, particularly within the software and Internet sectors. His academic background is equally robust, as he holds Bachelor\u2019s and Master\u2019s degrees in relevant fields, along with a Doctorate in Visual Arts.","sameAs":["https:\/\/atilagombos.com\/","https:\/\/www.facebook.com\/robert.gombos\/","https:\/\/www.instagram.com\/jasmine.directory\/","https:\/\/www.linkedin.com\/in\/robertgombos\/","https:\/\/en.wikipedia.org\/wiki\/Jasmine_Directory"]}]}},"_links":{"self":[{"href":"https:\/\/www.jasminedirectory.com\/blog\/wp-json\/wp\/v2\/posts\/29042","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.jasminedirectory.com\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.jasminedirectory.com\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.jasminedirectory.com\/blog\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.jasminedirectory.com\/blog\/wp-json\/wp\/v2\/comments?post=29042"}],"version-history":[{"count":0,"href":"https:\/\/www.jasminedirectory.com\/blog\/wp-json\/wp\/v2\/posts\/29042\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.jasminedirectory.com\/blog\/wp-json\/wp\/v2\/media\/30066"}],"wp:attachment":[{"href":"https:\/\/www.jasminedirectory.com\/blog\/wp-json\/wp\/v2\/media?parent=29042"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.jasminedirectory.com\/blog\/wp-json\/wp\/v2\/categories?post=29042"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.jasminedirectory.com\/blog\/wp-json\/wp\/v2\/tags?post=29042"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}