Local Citations vs Directory Listings: 2026 Comparison

When the original local search ecosystem took shape in the mid-2000s, the distinction between a “citation” and a “directory listing” barely registered as a meaningful consideration. A practitioner submitted a business name, address, and telephone number to whichever sites accepted submissions — Yellow Pages online, Yahoo Local, Citysearch, Superpages — and that was that. The two terms were, for most operational purposes, synonyms. The inflection point arrived around 2014, when Google’s Pigeon update tightened the relationship between local pack rankings and traditional organic signals, and Mike Blumenthal and David Mihm’s annual surveys of local search ranking factors started consistently splitting “citation signals” out as a distinct ranking category from “link signals.” That methodological choice — separating mentions of NAP data from earned links and from human-curated directory placements — propagated through the industry’s vocabulary, and by the late 2010s most agencies were treating the two as conceptually distinct levers that happened to overlap in practice.

By 2026, the divergence has hardened. Citations are now mostly a hygiene factor managed through aggregator pipelines, while curated directory listings have re-emerged as a discoverability and authority play in their own right, partly because generative search systems treat them differently. The walkthrough that follows is drawn from a recent engagement with a multi-location dental group in the English Midlands, supplemented with details from two analogous accounts (a regional veterinary chain and an independent optometry practice) where the same patterns surfaced. Names and a few numbers have been altered, but the decision tree is the one that was actually used, and the metrics fall within the ranges the engagements produced.

The Client Scenario: Multi-Location Dental Group

The client operated seven practices across a roughly forty-mile corridor, ranging from a flagship cosmetic-and-implant clinic in a city centre to two semi-rural NHS-mixed sites that had been acquired from retiring sole practitioners between 2021 and 2024. Annual marketing budget across the group sat at £84,000, of which roughly £22,000 had historically been allocated to “local SEO” — a line item that, on inspection, meant a monthly retainer with a freelancer who ran an unspecified citation-building tool and produced a quarterly screenshot deck. New patient acquisition had plateaued in late 2025, the principal dentist had grown sceptical of the retainer, and the brief was straightforward: rebuild the local visibility programme from first principles, with a defensible split between citation hygiene and directory placements, and prove the result in patient enquiries within ninety days.

Initial Audit Findings Across 7 Offices

The audit ran for nine working days and covered four data layers: the Google Business Profile for each location, the citation footprint on the top fifty data sources for UK healthcare businesses, the inbound link profile filtered for directory and association sources, and the actual map pack and organic rankings for thirty-six commercially relevant queries per location.

The citation picture was worse than the brief had implied but better than worst-case. Across the seven locations, 312 distinct citation instances were identified. Of those, 41% contained at least one NAP inconsistency — usually a stale telephone number from before the group consolidated its phone system in 2023, occasionally an address that referenced a former unit number, and in eleven cases a previous trading name that had been retired four years earlier. Two locations had duplicate Google Business Profiles, both in the form of an unclaimed legacy profile created by a patient or a third-party scraper, sitting alongside the verified profile. One of those duplicates was actually outranking the verified profile for a high-volume implant query, a finding that on its own justified the audit.

The directory listings picture was different in kind. The group held memberships with the British Dental Association, the Care Quality Commission’s public register (which is not optional but matters for trust signals), and two implant-specific manufacturer locator tools. Beyond that, almost nothing. There were no placements in regional business compendia, no chamber of commerce profiles, no presence in the curated health-services indexes that had grown in prominence as generative answer engines began sourcing entity descriptions from human-edited collections. The principal had assumed, reasonably, that the citation tool was covering this ground. It was not. The tool was pushing NAP data to aggregator pipelines and a few dozen low-authority sites, which is what citation tools do, and which is not the same activity as securing editorial placements in vetted indexes.

The ranking data closed the loop. The flagship clinic ranked in the local pack for thirteen of its thirty-six target queries; the two acquired rural sites ranked for four and three respectively. Pages two and three of the SERP were thinly populated with the group’s locations, and the share of voice in AI Overviews — measured manually across a sample of forty queries — sat at roughly 6%, with the dominant cited sources being two regional health portals the group had no relationship with.

Why Citations and Directories Diverged in 2026

The planning question the audit forced was why the two activities had drifted apart, and the answer matters because it dictates the budget split. Three forces were in play.

The first is technical. Citation aggregation pipelines — the wholesale data feeds that propagate NAP information to dozens of downstream consumers — have become both more reliable and more commoditised. Cleaning a business’s NAP across the major aggregators is now an annual hygiene task rather than a year-round project, and the marginal ranking benefit of citation volume beyond the core data layer has compressed substantially. This is consistent with the broader pattern that Forrester has documented in adjacent measurement disciplines, where standardised data infrastructures shift practitioner attention from data acquisition to data governance once a threshold of completeness is reached.

The second is algorithmic. Generative answer systems — Google’s AI Overviews, Bing’s Copilot mode, and the perplexity-class crawlers — preferentially surface a small number of high-authority, editorially curated sources when constructing local answers. According to eMarketer, AI systems disproportionately favour brand-owned and editorially vetted sources over scraped aggregator data, which has direct implications for which third-party properties matter for visibility. A listing on a curated index that the answer engine treats as a trusted source can produce more downstream visibility than fifty citations on auto-generated business pages.

The third is behavioural. Patient research patterns, at least in the dental vertical, have become more bifurcated. There is a fast lane — voice queries, “near me” mobile searches, map-pack clicks — where conventional citation hygiene supports the local pack mechanics. And there is a slow lane — comparative research for higher-ticket procedures like implants, orthodontics, and cosmetic dentistry — where prospects spend twenty to forty minutes across multiple sessions, often visiting curated directories, professional association pages, and review aggregators before booking a consultation. The two lanes reward different investments.

Treating citations and directories as a single line item, then, conflates a hygiene problem with a discoverability problem, and almost guarantees that one will be over-resourced while the other is starved. The audit findings made this concrete: the group had been spending against a hygiene budget on a discoverability problem they had not diagnosed.

Mapping the Decision Tree

With the audit findings on the table, the planning phase had four sequential decisions. Each was a fork rather than a slider, in the sense that the answer at each step constrained the options at the next. The framing here borrows from the kind of decision-staging that Harvard Business Review (2022) has documented in its retrospective on durable management frameworks: separate the irreversible decisions from the reversible ones and resolve the irreversible ones first.

Defining Citation Quality Thresholds

The first fork concerned what counts as a citation worth fixing or building. The 2018-vintage answer — fix everything, build everywhere — is no longer defensible, because the marginal cost of pursuing low-authority citations now exceeds the marginal benefit, and because the aggregator pipelines deliver most of the long tail without manual intervention.

The thresholds applied were concrete. A citation was considered worth manual remediation if it met at least three of the following five criteria: domain authority above 35 by either of the major third-party metrics; visible indexation of the business profile page in Google’s site: operator; an editable interface that allowed direct correction rather than requiring email-based requests with five-day turnarounds; presence in the citation footprint of at least two of the group’s three nearest competitors; and any signal that AI answer engines had crawled or surfaced content from the source in the past ninety days. The last criterion was new to the 2026 playbook and proved discriminating: roughly a quarter of the sources that passed the first four criteria failed the fifth, and a small number of sources that scored modestly on traditional authority metrics ranked highly on AI surfacing and were therefore prioritised.

Citations that failed the threshold were not deleted, since deletion is rarely possible and sometimes counter-productive, but were also not actively maintained. The aggregator pipelines would refresh them on their normal cadence, and any inconsistency in the long tail was tolerated as the cost of not fighting battles that no longer mattered.

Filtering Directories Worth Pursuing

The second fork concerned which directories to pursue actively. The criteria here were different, because a directory listing is closer to an editorial placement than to a data record, and the relevant questions are about audience, editorial standards, and the property’s own visibility within its niche.

Four filters were applied. The first was editorial vetting: did a human review submissions, and was there evidence of rejected applications? Properties that accepted any submission for a fee were deprioritised, since their utility as a trust signal had collapsed. The second was topical fit: a generalist regional index was less valuable than a healthcare-specific one with smaller traffic but tighter relevance, because the entity associations the listing creates are stronger. The third was the property’s own search visibility for queries the group cared about: if the directory itself ranked on page one for “cosmetic dentist [city]”, a placement on it captured a slice of that traffic directly, independent of any indirect ranking benefit. The fourth was longevity: properties that had existed for at least five years and showed evidence of editorial maintenance were favoured over newer indexes whose business model was unproven.

For readers wanting the methodology in fuller form, an in-depth piece on the topic of editorial directory evaluation walks through how each of these filters can be operationalised with publicly available data; the framework used in the dental engagement was a simplified version of that approach, adapted to the time constraints of a ninety-day brief.

Applying the four filters to a candidate list of 84 directories produced a shortlist of 23, which was further narrowed to 14 after a sense-check on submission requirements (some required a paid membership the client did not want to commit to in the first quarter) and on whether the listing format would accept multi-location entities cleanly rather than forcing the group into seven separate single-location submissions with no parent-entity relationship.

Budget Allocation Between Both Channels

The third fork was the budget split. The historic allocation had been, implicitly, 100% citation activity, because the freelancer’s tool only did one thing. The proposed reallocation, after the audit, ran roughly 35% to citation remediation and ongoing hygiene, 50% to directory placements and editorial outreach, and 15% to measurement infrastructure that the prior arrangement had lacked entirely. The total budget was held constant at £22,000 for the ninety-day window, with a pre-agreed re-baselining at the end of the quarter.

The reasoning behind the split deserves some attention, because the temptation when an audit reveals a neglected channel is to over-correct. The 50% directory allocation was not a bet that directories would outperform citations indefinitely; it was a recognition that the group had a one-time backlog of editorial placements to clear, after which the steady-state ratio would shift back toward parity. The measurement infrastructure spend — call tracking with location-level numbers, a properly configured Google Business Profile insights export pipeline, and a manual share-of-voice tracker for AI Overviews — was non-negotiable, because without it the ninety-day proof point would collapse into the same screenshot-deck reporting the principal had already rejected.

Deloitte Insights has noted, in its broader work on enterprise governance, that organisations frequently over-invest in execution capacity and under-invest in measurement maturity, with the result that programme outcomes cannot be attributed cleanly to interventions. The 15% measurement allocation was a deliberate response to that pattern, scaled to the small size of the engagement.

Choosing Manual Submission vs Aggregators

The fourth fork was tactical: for each citation or directory action, was the right execution path a manual submission, an aggregator-driven push, or a hybrid? The answer depended on the property type and the desired outcome.

For the core data layer — the four major UK-relevant aggregators — the choice was clearly aggregator-driven, because the cost of manual maintenance across hundreds of downstream consumers exceeded the cost of an annual aggregator subscription several times over, and because aggregator pipelines were the only practical way to maintain consistency at scale. For the curated directories on the shortlist, the choice was equally clearly manual, because each placement involved an editorial review, often a custom description, sometimes a category negotiation, and a relationship that the aggregator pipelines could not transact. For the middle tier — vertical citation sources that were neither core aggregator destinations nor curated directories — a hybrid approach was used: an aggregator push for the initial NAP record, followed by a manual enhancement pass to add categories, photos, and longer descriptions where the property allowed.

This three-way split is, in practice, where most of the tactical sophistication sits. The group’s previous freelancer had used aggregator-only execution for everything, which delivered NAP consistency but left enrichment fields empty across hundreds of profiles. Empty enrichment fields are a missed opportunity rather than an active harm, but at scale they represent meaningful foregone visibility.

Execution Over 90 Days

The execution phase was structured into three roughly equal blocks, with explicit handoff criteria between them. The intent was to front-load the remediation work that any subsequent activity depended on, then run the directory outreach in the middle block while the citation changes propagated, and use the final block to measure and adjust.

Week 1-3: NAP Cleanup Sprint

The first three weeks were almost entirely about citation hygiene, because every subsequent activity assumed clean NAP data. The team started with the duplicate Google Business Profile problem, since unresolved duplicates can suppress the legitimate profile and any improvements made to it. Both duplicates were resolved through Google’s standard merge-or-remove flow, which took eight and twelve days respectively — a reminder that the platform’s response times have not improved meaningfully in the past several years and that any project plan needs to budget for them.

In parallel, the audit’s list of 312 citations was filtered against the quality thresholds defined in planning. 187 instances qualified for active remediation. Of those, 154 were addressable through the four core aggregators, which were updated in a single pass on day four. The remaining 33 required manual intervention: 21 through self-service editing interfaces, 12 through email-based correction requests. The email-based requests had a 67% completion rate by the end of week three, with the remainder either pending or refused on the grounds that the source required documentary proof of the corrected information, which was supplied in the follow-up.

The flagship clinic’s profile received a more comprehensive enrichment pass: rewritten service descriptions for each of its sixteen listed services, eighteen new photos taken on a single afternoon shoot organised for the purpose, the addition of attributes that the previous freelancer had not surfaced (online appointment booking, wheelchair accessibility, language capabilities of the clinical team), and the configuration of the products module to feature the three highest-margin treatment categories. The other six locations received a lighter version of the same treatment, calibrated to their patient mix.

One detail worth noting: the cleanup sprint surfaced a structural problem with the group’s website that no amount of citation work could compensate for. The location pages had been built from a single template with the practice name and address dynamically inserted, but the template had a bug that produced inconsistent schema markup across the seven pages. Two locations were missing LocalBusiness schema entirely; three had it but with a malformed address property. Fixing the template was outside the original brief, but the principal authorised the development work — about £1,400 with the group’s web developer — once the dependency was explained. This is the kind of finding that distinguishes an audit from a checklist: the citation hygiene work is necessary but not sufficient if the entity’s own website is sending inconsistent signals.

Week 4-8: Niche Directory Outreach

Weeks four through eight focused on the fourteen directory placements on the shortlist. The pacing here was deliberate: rather than batch-submitting to all fourteen in a single week and waiting passively for outcomes, the team submitted in tranches of three or four per week, allowing time to respond to editorial queries, supply additional documentation, and adjust the submission template based on the patterns of acceptance and rejection.

The submission template was the largest investment in this block. Each directory accepts a slightly different format — character limits on descriptions, required versus optional fields, category taxonomies that vary in granularity — and producing a single canonical entity description that could be adapted to each was the precondition for efficient submission. The canonical description was 480 words long, written in the third person, and structured to support extraction by both human editors and automated entity-recognition systems. It included the group’s founding date, the principal’s clinical credentials, the locations served, the procedure categories offered, the professional affiliations, and three specific differentiators — same-day implant capability, a digitally-led orthodontic workflow, and a sedation-supported anxiety pathway — that distinguished the group from generalist practices in the catchment.

Twelve of the fourteen submissions were accepted within the eight-week window. One was rejected on the grounds that the directory had recently narrowed its scope to specialist practices in a sub-discipline the group did not formally claim, which was a fair editorial decision. One was held in queue and ultimately accepted in week eleven, outside the planned block but within the engagement.

The work in this block was substantively different from the citation work in the prior block. Citation remediation is industrial: standardised data, repetitive actions, throughput-dominated. Directory outreach is craft: each submission is a small editorial negotiation, and the cumulative effort of twelve placements consumed roughly seventy hours of senior practitioner time over the five weeks. The cost-per-placement was high in absolute terms but defensible when measured against the visibility outcomes that emerged in the third block.

One of the placements deserves a specific mention because it illustrates the principle that filter-quality matters more than placement-volume. A regional health-services index with modest traffic accepted the flagship clinic’s submission after a back-and-forth that included a phone conversation with the editor, the supply of two patient testimonials in a format compatible with the index’s review schema, and a category negotiation that ultimately placed the clinic in two categories rather than the one originally requested. Within thirty days of acceptance, that single placement was driving more referral traffic to the cosmetic-dentistry landing page than the previous twelve months of aggregated traffic from low-authority citations. The delta was not because the directory had unusually high traffic in absolute terms — it did not — but because its audience was unusually well-qualified and its own search visibility for cosmetic dentistry queries in the region was strong. an in-depth piece that this kind of disproportionate return from a small number of editorially vetted placements is increasingly the norm in mature local search markets, which matches the engagement’s own findings.

Week 9-13: Tracking Map Pack Movement

The final block was for measurement and adjustment. By week nine, the citation cleanup had fully propagated through the aggregator pipelines, the directory placements were live, and the website schema fixes had been crawled and re-indexed. The intent of the block was not to add new activity but to read the results carefully and decide what to carry into the next quarter.

Map pack tracking ran daily across the original thirty-six commercially relevant queries per location, with the rank check performed from within each location’s geographic catchment using a proxy network configured for that purpose. The aggregate movement is summarised in the results section, but the more interesting pattern was at the per-query level. Queries that combined a procedure term with a city or neighbourhood — “dental implants Lichfield”, “Invisalign Sutton Coldfield” — showed faster and larger movement than generic queries like “dentist near me”, which is consistent with the broader pattern that specific intent queries have a flatter local-pack ranking surface and respond more rapidly to entity-level signal improvements.

Referral traffic from directories was tracked through a combination of UTM-tagged links (where the directory permitted them), referrer logs (where it did not), and call tracking with location-specific numbers used in the directory listings only. The call tracking arrangement was particularly informative because it isolated calls originating from directory listings from calls originating from the Google Business Profile or organic search, and allowed a clean attribution of consultation bookings to the channel that generated the initial enquiry.

One adjustment was made within the block. A second flagship-tier directory submission was prepared and submitted in week ten after the data from the first directory placement made it clear that the cosmetic-dentistry category was responding more strongly than expected. This kind of mid-flight adjustment is only possible with the measurement infrastructure in place, and it justifies the 15% budget allocation to measurement that some clients resist.

Results and Transferable Lessons

The results across the ninety-day window were stronger than the brief required but not uniformly so. The pattern was instructive: the channels and locations that were over-performing reinforced the planning logic, and the underperforming pockets revealed where the playbook needed adjustment.

Performance Metrics by Channel

The headline numbers, before drilling into the breakdowns: across the seven locations, total new patient enquiries in the ninety-day window grew by 38% relative to the prior ninety days, with the strongest growth at the flagship clinic (61%) and the weakest at the two acquired rural sites (19% and 14%). Map pack visibility, measured as the share of target queries on which a location appeared in the local pack, grew from a weighted average of 31% to 54%. AI Overview citation share, measured manually across the same forty-query sample as the audit, grew from approximately 6% to approximately 19%. Cost per acquired patient, calculated against the £22,000 marketing spend for the window and the 174 net new patients attributed to local channels, came in at £126.

A breakdown is provided in Table 1. The figures are aggregated across the seven locations and rounded for presentation, but the underlying numbers were tracked at the location level and the patterns held across the group, with the magnitude varying by location size and competitive intensity.

Table 1: Channel performance summary across the 90-day window

A few observations are warranted on the table itself, because the figures invite over-interpretation. The cost per enquiry differential between citation remediation and directory placements does not mean that citations are intrinsically cheaper to convert; it reflects the fact that the citation work was largely remediation of an existing footprint rather than greenfield construction, and that the 90-day window captured the first wave of directory effects without capturing the full tail. The directory cost per enquiry will fall in subsequent quarters as the placements continue producing without further investment, while the citation channel will require ongoing hygiene spend to hold its position.

Local Pack Ranking Shifts

The local pack movement was concentrated in the queries where competitive intensity was moderate rather than at the extremes. For queries with very low competition — niche procedure terms in smaller catchments — the locations had been ranking before the engagement and continued to rank, with marginal position improvements. For queries with very high competition — generic single-word terms in the city centre — the local pack remained dominated by entrenched incumbents, and the movement was modest. The bulk of the gain came from the middle band: specific procedure-plus-location queries in moderately competitive markets, where the combination of cleaned-up citations, schema fixes, and new directory placements moved the locations from page two to page one and into the local pack.

The flagship clinic’s gains were not evenly distributed across its query set. Implant-related queries showed the strongest movement, with eight of twelve target queries entering the local pack from prior off-pack positions; cosmetic dentistry queries showed strong but slightly slower movement; orthodontics queries showed the weakest movement, partly because the competitive set in that category included two specialist orthodontic practices whose entity signals on the directory layer were stronger than the group’s. That last finding became the brief for the subsequent quarter: a targeted push on orthodontics-specific directory placements and editorial features.

The rural locations’ weaker performance is worth dwelling on, because it tempers the optimistic reading of the headline numbers. Both acquired sites had been operating under their previous trading names until late 2024, and their entity histories on the major aggregators contained legacy data that the cleanup sprint corrected but could not fully overwrite in the propagation window. Local search systems appear to weight entity history, and a recently re-named practice carries less authority in the rankings than a long-established one with the same NAP. This is a constraint the playbook cannot remove; it can only be waited out, and the projection on current trajectories is that the rural sites will close most of the ranking gap with the flagship over the subsequent two to three quarters as the entity history matures.

Referral Traffic from Directories

Referral traffic from the twelve directory placements totalled 4,318 sessions across the engagement window, with extremely uneven distribution: the top three placements accounted for 71% of the traffic, the next four accounted for 22%, and the bottom five accounted for the residual 7%. The conversion rate from session to enquiry varied by directory in ways that were intelligible: directories with stronger editorial vetting and tighter topical focus produced sessions that converted at three to four times the rate of generalist regional indexes, even when the absolute session counts from the generalists were higher.

This pattern — high concentration of value in a small number of placements — is the most important transferable finding from the engagement, and it has implications for how the next ninety-day cycle should be planned. The right response is not to abandon the lower-performing placements, since their cost was sunk and their incremental maintenance burden is low, but to allocate the next cycle’s directory budget to deepening the relationship with the top-performing properties (sponsored content, editorial features, premium category placements) rather than to broadening the placement footprint at the bottom of the distribution.

The conversion-rate differential also has implications for measurement design. Tracking referral sessions in aggregate would have produced a misleading picture in which the top-of-the-distribution and bottom-of-the-distribution traffic were lumped together and the average masked the variance. The measurement infrastructure budget paid for itself on this finding alone, because without per-placement tracking the planning for the subsequent quarter would have been built on an aggregate average rather than the actual distribution.

Cost Per Acquired Patient

The £126 blended cost per acquired patient sits within the range that the group’s principal had quoted as acceptable based on the lifetime value of an average patient in the practice. Lifetime value calculations for dental practices are sensitive to the procedure mix, with implant patients carrying multi-thousand-pound lifetime values and routine NHS-mixed patients carrying substantially lower values. The £126 figure therefore conceals a wide internal range: cosmetic and implant enquiries from the directory channel had an effective cost per acquired patient of approximately £180 (because the conversion rate from enquiry to booked consultation was lower for high-ticket procedures), while routine new-patient registrations from the citation and Google Business Profile channels had an effective cost per acquired patient closer to £85.

The differential is not a problem; it is a portfolio. High-ticket acquisitions are intrinsically more expensive to source per unit, but their downstream value is also higher, and the group’s overall return on the marketing spend was positive across both ends of the distribution. The composite cost per patient was used in client reporting because it is the figure the principal cared about for budget purposes, but the per-segment figures were the basis of the planning conversation for the subsequent quarter.

For practitioners reading this with their own client portfolios in mind, the central caveat is that £126 is a result specific to this engagement and should not be applied to other contexts. It is the value that the specific budget, the specific competitive context, the specific patient mix, and the specific ninety-day window produced. A different practice in a different market with a different starting position would produce a different number, and any figure imported from another engagement should be treated as a sanity check rather than a target. As Forrester‘s media citation guidance underlines in a related context, comparative figures need their methodology footnotes attached or they invite misuse.

Adapting the Playbook to Tighter Budgets

The playbook described above was sized to a £22,000 ninety-day budget, which is at the upper end of what a multi-location independent practice would typically commit to local search work in a single quarter. The interesting question for most practitioners is how it scales down, because the majority of clients operate under tighter constraints. Three reduced-budget scenarios are worth working through, because each reveals different priorities.

The first scenario is a £6,000 ninety-day budget for a single-location practice in the same vertical. The original allocation cannot be uniformly compressed; the measurement spend is a fixed cost rather than a percentage, and the curated directory placements have a minimum effective volume below which the channel cannot demonstrate a result within the window. The reallocation that made sense in analogous engagements was: roughly 40% to citation remediation (sufficient to clean the core aggregator layer and ten to fifteen manual corrections, though not the long tail), 45% to four or five carefully chosen directory placements (rather than fourteen, with the selection biased toward the highest-quality candidates from the shortlist), and 15% to a stripped-down measurement setup (call tracking on the primary number only, manual share-of-voice tracking on a smaller query sample). The expected result is a smaller absolute uplift but a similar proportional one, and the cost per acquired patient is roughly comparable because both the numerator and the denominator scale together.

The second scenario is a £3,000 ninety-day budget. At this level, the citation versus directory choice becomes binary rather than balanced, because the budget cannot support both channels at minimum effective volumes. The framework for choosing between them is the audit finding: if the practice has a substantial citation hygiene problem (more than 30% of citations contain inconsistencies, or there are unresolved Google Business Profile duplicates), the budget should go to citation cleanup, because the hygiene problem is a ceiling on every other activity’s effectiveness. If the citation footprint is already clean — which is increasingly common for practices that have been working with a competent agency for several years — the budget should go to two or three directory placements, because the marginal return on additional citation work is low and the directory channel offers the bigger uplift opportunity. The temptation to split the budget evenly should be resisted, because at this level a half-funded citation programme and a half-funded directory programme produces no meaningful result on either side.

The third scenario is the timeline-constrained variant: a £22,000 budget but a thirty-day window rather than ninety. This is a harder constraint than the budget reductions, because the citation propagation cycles and the directory editorial review cycles cannot be compressed. A thirty-day brief should not promise ranking movement, because the data simply will not have settled in the window. What it can promise is foundation-level deliverables — citation cleanup completed, directory submissions made, measurement infrastructure live — with the expectation that ranking and traffic outcomes materialise in the subsequent thirty to sixty days. Practitioners who accept thirty-day briefs with ranking commitments are setting themselves up to fail, and the right response to such a brief is to renegotiate the timeline rather than to compress the methodology.

A fourth scenario, addressed more briefly because it is industry-specific rather than budget-specific, concerns vertical adaptation. The dental playbook above transferred reasonably cleanly to the veterinary engagement that informed parts of the framework, with the main adjustment being the directory shortlist, which was rebuilt from scratch around veterinary-specific properties. It transferred less cleanly to the optometry engagement, where the dominant directory landscape was less developed and the citation hygiene work returned a larger share of the result. The implication is that the citation-versus-directory split is not a fixed ratio but a vertical-dependent variable, and any agency template that prescribes a single ratio across clients is hiding a planning step rather than performing one.

One more constraint worth flagging is geographic. The dental engagement was in a market where the curated directory landscape is reasonably well-developed and editorially active. In markets where it is not — particularly smaller English regions and parts of the UK outside England — the directory shortlist will be shorter and the achievable budget allocation to directories correspondingly smaller. In those cases the adjustment is to redirect the directory budget to other forms of editorial placement: regional press features, association sponsorships with online listings, and partnerships with non-competing local businesses that operate listings as part of their member-services offering. The underlying logic is the same — earned third-party visibility from sources the answer engines treat as authoritative — but the tactical execution has to follow the local supply of opportunities.

The transferable principles, abstracted from the engagement and the variants, come down to four. First, citations and directories are different activities that have drifted apart and should be planned and budgeted separately. Second, quality thresholds matter more than volumes on both channels, and the right response to a finite budget is to raise the threshold rather than to spread the spend thinner. Third, measurement infrastructure is a fixed cost that scales poorly downward, and below a certain budget level the choice is between funding measurement properly or running blind, with no meaningful middle option. Fourth, the audit determines the allocation rather than the other way round, and any agency that proposes a fixed citation-to-directory ratio without auditing the client’s specific footprint is selling a template, not a strategy.

Looking forward to 2027 and the conditions under which the patterns described here would persist or break: the projection on current trajectories is that the citation channel will continue to commoditise toward an annual-hygiene baseline, with the marginal return on incremental citation work compressing further as aggregator pipelines absorb more of the long-tail propagation work. The directory channel, conversely, will continue to bifurcate, with a small number of editorially curated properties capturing a growing share of both human referral traffic and AI-system citation share, and a long tail of low-quality submission-driven properties becoming progressively less useful. The combined effect is that the budget allocation observed in the engagement — roughly one-third citations, one-half directories, one-sixth measurement — will probably move further in the directory direction over the next twelve to eighteen months, perhaps toward a 25/60/15 split for similarly positioned clients.

This projection holds under three conditions. The first is that AI answer systems continue to favour editorially vetted sources over aggregator data, which has been the consistent direction of travel through 2025 and 2026 but is not guaranteed to continue if the underlying ranking systems are reweighted. The second is that the curated directory landscape continues to maintain its editorial standards, which depends on the commercial sustainability of the properties involved; if a wave of consolidation produces lower-quality merged properties, the channel’s value will compress. The third is that the regulatory environment around AI-generated answers does not introduce sourcing requirements that displace third-party directories in favour of, for instance, government-maintained registers as the canonical entity sources for healthcare services. Any of these three conditions failing would falsify the projection in a recognisable way, and practitioners working in the space should be watching all three indicators rather than committing to the allocation as a fixed strategy.

If the projection holds, the practical implication for agencies and in-house teams planning their 2027 work is that the directory shortlist deserves more attention now than it received in the 2024-2025 cycle, and that the editorial-relationship muscle — the ability to negotiate a placement, supply documentation, manage a category negotiation, and follow up through an editorial queue — is the capability that will discriminate between effective and ineffective local search programmes. Citation hygiene will remain necessary, but it will increasingly be the floor of the work rather than the substance of it.