Most business directories no longer move rankings, but a few listings still shape what search engines believe

What a citation was supposed to do

In local search, a citation is any mention of a business's name, address and telephone number on a website other than its own. A directory entry is a citation, and so is a chamber of commerce page, a supplier's stockist list, a professional body's member roll and an entry on a statutory register. For most of the 2010s the working theory was simple. Search engines were assembling a picture of each business from fragments scattered across the web, so the more fragments existed and the more consistently they agreed, the better that business would do.

The theory had the advantage of being actionable. A small business could not easily acquire editorial links or hundreds of reviews, but anyone could fill in a form on two hundred directories, and an industry grew up doing that for a monthly fee.

The size of the market explains why the work was sold by volume. The Office for National Statistics counted 2.73 million VAT or PAYE registered businesses in the UK operating from 3.2 million sites, recording the position as it stood on 14 March 2025, with companies making up 76.7 per cent of them and sole proprietors a further 13.9 per cent ^[15]. Most of those sites are one location trading in one town, and the cheapest thing anybody could sell them was presence.

What Google says about the information it holds on a business

Google's public account of local ranking is short and has not changed much. Its Business Profile help page states that "Local results are mainly based on relevance, distance and popularity" ^[1]. Relevance is "how well a Business Profile matches what someone is searching for", distance is how far the business sits from the searcher, and the third factor, which the page heads Prominence, is "how well-known a business is" ^[1].

The sentence that matters here is the one that follows. Prominence, Google says, is "also based on info like how many websites link to your business and how many reviews you have" ^[1]. Links and reviews are named. Directory listings, citations and NAP consistency are not named anywhere on that page, which also states flatly that "there's no way to request or pay for a better local ranking on Google" ^[1].

That is not the same as saying published records are irrelevant, because a separate Google page describes where profile information comes from. It is "compiled from a variety of sources", listed as business owners who claim their listing, "licensed data from third parties", "users who contribute factual information", "publicly available information" including "crawled web content", and Google's own interactions with a place ^[2]. Licensed third-party data and crawled content are, in practice, directories and other published records. The page does not say how Google resolves a conflict between them ^[2].

The honest reading is narrower than either side of the usual argument. A consistent published record is one of the inputs to what Google believes a business is, and not one of the factors Google names when explaining why one business outranks another.

The representation guidelines reinforce the first half of that and say nothing about the second. A profile name "should reflect your business's real-world name", a business should be represented "as it's consistently represented and recognized in the real world", and breaches can lead to "changes to your information or, in some cases, removal of your business information from Google" ^[3]. Consistency, in Google's framing, is a truthfulness obligation rather than a ranking tactic.

What the measured evidence shows, and how old it is

The figure most often produced in support of citation building comes from one BrightLocal study. It examined anonymous data for 122,125 local businesses across 26 industries, identified 18,580 local search terms and looked at the top ten businesses for each, and reported that businesses ranked in the local top three held an average of 85 citations, rising to 86 for the business in first place ^[8].

Two things about that number are usually left out when it is quoted. The first is that the study's own authors said it plainly: "correlation doesn't equal causation", and businesses at the top of local results tend to be performing well on everything else too, so a high citation count may be a symptom of an established business rather than a cause of its position ^[8]. The second is the date. It was published in August 2018 ^[8]. No study of comparable scale appears to have replaced it.

The practitioner position has shifted accordingly, including among the people who sell the tools. Darren Shaw of Whitespark, whose own business includes citation building, argues that citations do still matter, and in the same post writes that they are "just not the golden ticket that they used to be" and that "those days are done, that concept of citation building is over" ^[9]. His recommendation is narrow: correct the data on the sites humans actually look at, add industry-specific and local sites, cover the leading directories, then stop ^[9].

There is also evidence that the underlying problem was real, even where the prescribed cure was oversold. A study by the listings company Uberall, reported in 2019, analysed 73,000 business location listings and counted 978,305 errors in opening hours, which it characterised as almost half of all listings, together with 710,113 incorrect website addresses, 510,010 wrong location names and 421,048 wrong street details ^[14]. Those are not ranking problems. Those are customers standing outside a closed door.

What happened to the directories themselves

The commercial decline of the general directory is easier to date than its search value. Yell announced in September 2017 that it would stop printing the Yellow Pages, with the final edition to be distributed in Brighton in 2019, at which point the group was still publishing 104 print editions and its UK print advertising revenue had fallen 30 per cent to £43 million in the year to 31 March 2017 while digital revenue rose 5 per cent to £193 million ^[13]. The chief executive, Richard Hanscott, said the company was "proud of the transformation we've made from print to digital" ^[13].

What replaced the printed directory was not a digital directory of comparable reach. It was a single dominant map, surrounded by a handful of specialists. BrightLocal's 2026 consumer survey, based on responses from 1,002 US adults, found Google still the standout source consumers use for reviews although its share had fallen from 83 per cent in 2025 to 71 per cent, reported that use of generative AI tools for local recommendations had risen from 6 per cent to 45 per cent, and recorded Apple Maps use growing from 14 per cent to 27 per cent ^[10].

That survey covers American consumers rather than British ones, which limits how far it can be read across to a clinic in Bradford. The structural point stands on its own. The platforms gaining ground are not directories in the old sense at all. They are maps, social platforms, review specialists and assistants.

When a directory listing becomes a liability

There is one firm line in Google's documentation on this subject, and it runs through the middle of the old citation-building industry. Google's spam policies for web search list "low-quality directory or bookmark site links" among the practices it treats as link spam, alongside "buying or selling links for ranking purposes" and "excessive link exchanges" ^[4]. Paid placements are acceptable, the policies say, where the link carries a rel attribute marking it as nofollow or sponsored ^[4].

This does not make every directory listing risky. A statutory register, a professional body's member list and a local newspaper's business page are none of them low-quality directory links. What the policy does mean is that a service submitting a business to several hundred unvetted sites is manufacturing precisely the pattern Google's own policies name.

The more common harm has nothing to do with penalties. A listing created once, carrying a telephone number that changed three years later, keeps sending callers to a dead line. A duplicate profile created by a submission service splits reviews between two records, so neither looks as established as the business is.

Google treats accuracy as an obligation and reserves the right to remove or suspend profiles that breach its guidelines ^[3]. A stale listing elsewhere is therefore at best noise, and at worst a contradiction somebody will eventually have to resolve. Every listing is a small ongoing liability, which is a reason to own few of them.

The listings that are not optional

Underneath the marketing argument sits a category of listing that has nothing to do with search at all, and that a UK business in a regulated trade cannot choose to skip.

Anyone carrying out gas work must be on the Gas Safe Register. The government's regulated professions service states that registration "is a legal requirement for anyone carrying out gas work in the United Kingdom", describes the register as "the official list of gas engineers in the UK, Isle of Man, Jersey and Guernsey", and places the requirement under the Gas Safety (Installation and Use) Regulations 1998, with the Health and Safety Executive as the responsible regulator ^[18].

In regulated healthcare, the Care Quality Commission states that service providers must register with it, and that it is "the legal entity carrying on the regulated activity that must register" rather than the premises where care happens ^[16]. In financial services, the Financial Conduct Authority describes its Financial Services Register as "a public record of firms, individuals and other bodies that are, or have been, authorised by us or the PRA", and notes that "nearly all financial service activities must be authorised or registered by us" ^[17].

Two further records are free, public and durable. Companies House "incorporate and dissolve limited companies" and "register company information and make it available to the public", covering more than 5 million registered limited companies ^[19]. TrustMark describes itself as "the Government Endorsed Quality Scheme" and tells homeowners to "make sure the tradesperson is registered with TrustMark for the specific trade you need" ^[20].

These entries have three properties that no commercial directory has. They are authoritative, because a regulator or a government-endorsed body maintains them rather than selling them. They are consulted by customers who are checking a business rather than browsing for one, which is a later and more valuable moment in a decision. And they let any independent system confirm that a business exists and is permitted to do what it advertises.

What the AI assistants appear to read

The newest version of the directories argument is that listings are how a business gets itself into AI answers. The evidence available is a good deal thinner than the claim.

Ahrefs examined the domains cited most often across three systems, drawing on roughly 76.7 million AI Overviews, 957,000 ChatGPT prompts and 953,500 Perplexity prompts recorded in June 2025 ^[11]. Wikipedia led in all three on mention share, which Ahrefs defines as how often a model chooses a domain, at 16.3 per cent in ChatGPT, 12.5 per cent in Perplexity and 8.4 per cent in AI Overviews ^[11]. YouTube was prominent in Perplexity at 16.1 per cent and in AI Overviews at 9.5 per cent but absent from ChatGPT's top ten, while Reddit and Quora featured mainly in AI Overviews, at 7.4 per cent and 3.6 per cent ^[11].

Semrush's account of the same question is qualitative rather than quantitative. It reports that review sites and aggregators are favoured for evaluative queries and forums where firsthand experience is wanted, and, citing a Search Engine Land study, that "ChatGPT often cites Wikipedia, Reddit, and sites like Forbes", that "Google often cites places like Facebook and Yelp" and that "Perplexity often cites Reddit, LinkedIn, and G2" ^[12]. For local businesses it names Yelp and TripAdvisor, and gives no figure for how often third-party sources are cited in place of a brand's own site ^[12].

Not one of those lists is a general business directory. Where a listings platform does appear it is a review platform with a real audience of its own, which is a different proposition from a submission site that exists to hold listings. Google's guidance on its own AI features is blunt about eligibility: "There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary", a page needs only to be "indexed and eligible to be shown in Google Search with a snippet", and "you don't need to create new machine readable files, AI text files, or markup to appear in these features", while "indexing and serving isn't guaranteed" ^[7].

The part structured data does, and the part it does not

If the object is to help a search engine identify a business rather than rank it, structured data on the business's own site does that work more honestly than an off-site listing, and Google documents what it is for.

Organization markup on a home page "can help Google better understand your organization's administrative details", and the url property "helps Google uniquely identify your organization" ^[5]. The sameAs property is defined as "the URL of a page on another website with additional information about your organization", with social media and review profiles given as examples and multiple values permitted ^[5]. Some properties, Google notes, "are used behind the scenes to disambiguate your organization from other organizations" ^[5].

That is the mechanism the citation industry was reaching for, expressed properly. Rather than hoping an engine notices that forty directories agree, the business states on its own site which off-site records are in fact it, and a regulator entry and a Companies House record are the right things to point at.

The limits are documented as clearly as the uses. Local business markup requires the name, "the name of the business", and the address, "the physical location of the business", and Google says "the more properties you provide, the higher quality the result is to users" ^[6]. Both pages warn that "Google does not guarantee that features that consume structured data will show up in search results" ^[5][6]. Structured data describes a business. It does not promote one.

What it means for a local business

For a single-site business in a UK town the position is narrower and cheaper than the old orthodoxy, and the work is worth doing for reasons that have little to do with ranking.

1. Correct the places humans actually use before thinking about search at all. Wrong opening hours on a widely read platform cost real enquiries, and the measured error rates were concentrated in exactly those fields ^[14].
2. Claim and complete the Google Business Profile using the real-world name and the actual location, because Google says businesses with complete and accurate information are more likely to show up in local search results ^[1][3].
3. Make sure every register the business is legally required to appear on is present and current, whether that is the Gas Safe Register, CQC registration for a regulated healthcare service, or the Financial Services Register ^[18][16][17].
4. Add the free authoritative records: an accurate Companies House entry, the relevant professional body, and a scheme such as TrustMark where it applies to the trade ^[19][20].
5. Point at those records from the website using Organization markup and sameAs, so the site states which off-site entries belong to it rather than leaving a search engine to infer it ^[5].
6. Decline bulk submission services and recurring listings fees. Google's spam policies name low-quality directory links specifically, and the headline evidence for volume is a correlation study from 2018 ^[4][8].

It also leaves out any promise of a ranking improvement, because none can honestly be made. The reasons to do the work are that customers find correct information, that prospective clients can verify the business independently, and that any system describing it has consistent material to work from. On Google's own account it is links and reviews that bear on local ranking, not a count of listings ^[1].

For a regulated business there is a stronger reason still. In healthcare, financial services and the gas trades the register entry is what a cautious customer checks before committing. A correct and findable entry reassures. A missing one raises exactly the question the business does not want raised.

What remains uncertain

The largest gap in this subject is the absence of recent measurement. The citation study everyone still cites is from 2018 and was explicit that it showed correlation rather than cause ^[8]. The listings accuracy study is from 2019 ^[14]. Nothing at comparable scale has been published since, so both the claim that citations still matter and the claim that they no longer do rest on inference.

Second, Google does not say how it weighs its sources against one another. Its documentation confirms that licensed third-party data and crawled web content both feed profile information, but not what happens when either disagrees with the owner's own claim, nor how quickly a correction travels ^[2]. A business fixing a wrong address cannot tell whether it must also correct an upstream record.

Third, the AI picture is early and moving. The domain-level citation data covers a single month in 2025 across three systems ^[11], the qualitative account published in April 2026 declines to quantify how often third-party sources are used instead of a brand's own site ^[12], and Google states both that no special markup is needed and that inclusion is not guaranteed ^[7]. Whether assistants come to lean more heavily on public registers, or less, is not settled.

Fourth, the consumer behaviour evidence is American. BrightLocal's figures come from 1,002 US adults, and the rise it reports in generative AI tools used for local recommendations, from 6 per cent to 45 per cent, has no equally recent UK equivalent ^[10]. Sector differences here, particularly in regulated trades where a register carries legal weight rather than marketing weight, could make the British pattern meaningfully different.

What can be said with reasonable confidence is the asymmetry. The cost of a wrong listing on a platform customers use is immediate and measurable. The benefit of a correct listing on a platform nobody reads is, on the published evidence, unproven. That asymmetry, rather than any theory about ranking signals, is the argument for keeping a few records accurate, making the statutory ones impeccable, and leaving the rest alone.