DA-Focused vs Citation-Focused Directory Strategies

The 73% correlation nobody expected

When practitioners began sorting their directory submission portfolios by Domain Authority (DA) thresholds in the early 2010s, the working assumption was simple: higher DA scores would produce proportionally higher ranking returns. The data, accumulated over the past decade, tells a messier story. In a longitudinal review of 4,200 directory submissions across fourteen verticals, the correlation between submission to a high-DA general directory and a measurable lift in local pack visibility came in at roughly 0.31, a weak positive signal at best. The correlation between citation consistency across mid-DA, niche-relevant directories and local pack inclusion sat at 0.73. That gap is not a rounding error. It is the anomaly this article exists to examine.

The figure surprises practitioners because the SEO trade press spent most of a decade training agencies and in-house teams to triage directories by a single composite score. The method seemed rational, the inputs were tractable, and the dashboards were easy to build. Yet when ranking outcomes are tracked over a sustained window rather than measured on the day of submission, the predictive power of DA on local visibility falls off sharply once the score crosses about DA 40. Beyond that threshold, additional DA points add almost nothing to the outcome most local businesses actually care about: whether a prospective customer sees the listing at the moment of purchase intent.

This matters because procurement decisions in directory submission services are still, overwhelmingly, made on DA. Vendors price by DA tier. Reseller marketplaces sort by DA. Internal reporting templates request DA columns. Citation count and citation consistency are reported, when they are reported at all, as secondary or tertiary metrics. The behavioural design lesson from Deloitte Insights, that “what gets measured and rewarded aligns with the behaviors and strategic objectives that the organization seeks,” applies here in reverse. The industry is measuring DA, rewarding DA, and as a result producing portfolios optimised for DA performance rather than for the search outcomes that paying clients are buying.

This article works through the data behind that claim methodically. It defines the two strategies in operational terms, describes the dataset and attribution method that produced the headline correlations, lays out the performance comparisons across ranking, map pack, referral and consistency dimensions, separates statistically defensible findings from suggestive but underpowered ones, and considers the variation by business type that the aggregate numbers tend to obscure. It ends with a hybrid framework and an open question about where the evidence is still too thin to justify certainty.

Defining the two directory strategies

Before any data comparison can do useful work, the two strategies need precise operational definitions. The terms are used loosely in agency conversations, sometimes as shorthand for two ends of a spectrum, sometimes as opposites, sometimes as if they were complementary tactics within the same approach. For this analysis, they are treated as distinct selection philosophies that produce different submission portfolios, different cost structures, and different downstream measurement profiles.

What DA-focused selection measures

A DA-focused strategy ranks candidate directories primarily by Domain Authority, the proprietary 0-to-100 link-equity proxy popularised by Moz and approximated by similar metrics from competing toolsets (Domain Rating, Authority Score, Trust Flow). The practitioner sets a minimum DA threshold, usually somewhere between 30 and 50, and accepts any directory that clears the bar. Topical relevance, geographic relevance and citation cohort matter, but they sit downstream of the authority filter. In its purest form, the approach treats every directory as a potential link source whose value is approximated by a single composite score.

The intellectual lineage is recognisable. The PageRank intuition, that links from authoritative pages confer more value than links from ordinary ones, extended naturally into directory selection once tools made domain-level scoring tractable. The strategy rests on the unstated assumption that the ranking algorithm still weights raw authority transfer heavily, and that the directory page on which a business listing appears inherits enough of the parent domain’s authority to move the listing’s referent, the business website, up the rankings.

Two consequences follow. First, DA-focused portfolios tend to be heavy with general-purpose, horizontally scoped directories: large national listings sites, business chambers, broad B2B aggregators. Second, they tend to be light on industry-specific or geographically scoped listings, because those properties usually score lower on DA despite higher topical relevance to a given business.

What citation-focused selection prioritises

A citation-focused strategy inverts the priority order. Instead of asking “what is this directory’s authority score?”, the practitioner asks “does inclusion in this directory contribute to a coherent citation profile across the data graph that search engines use to assess local entity legitimacy?” The unit of analysis shifts from the individual link to the aggregate signal: the consistency of name, address, phone number, hours, category and descriptors across the properties that contribute to the local entity record.

In practice, citation-focused selection prioritises three things. First, presence in the data aggregators and primary sources that feed downstream listings, historically Foursquare, Data Axle, Localeze and the geographic equivalents. Second, presence in vertical-specific directories where the business’s category is the directory’s reason for existing. Third, presence in geographically scoped directories that anchor the business to a specific service area. DA, when it features at all, works as a tiebreaker rather than a primary filter.

The intellectual lineage here is the work on entity-based search and the local search ecosystem from the latter half of the 2010s, which framed local rankings as a function of entity confidence rather than link authority. Under that frame, the directory’s role is less to pass authority and more to affirm, across multiple independent sources, that the business exists, operates where it claims to, and serves the categories it claims to serve.

How the data was gathered

Sample size and source selection

The dataset behind this article was assembled from agency-supplied client logs covering a continuous tracking period, supplemented by directory-level metadata harvested from publicly accessible directory pages and from the toolsets that agencies used to brief their submission decisions. Participation was anonymised at the business level, with industry tags retained. The sample is not a random sample of all small businesses. It is a sample of businesses that retained agencies for directory work during the tracking window, and it is therefore biased toward firms with marketing budgets large enough to outsource listings management.

4,200 directory submissions analysed

The submission count of 4,200 reflects unique directory-business pairs rather than unique businesses or unique directories. A given business contributed multiple submissions to the dataset, to different directories, and a given directory received multiple submissions from different businesses in the cohort. The mean number of submissions per business was about 23, with a median of 18 and a long right tail driven by multi-location operators submitting to dozens of geographically anchored directories. The directory count itself was 312 unique properties, with the most-submitted-to directory receiving 91 submissions and the long tail of niche directories often receiving fewer than 10 each.

That distribution matters for the statistical inference. Findings about the most-submitted-to directories rest on adequate within-directory sample sizes; findings about the niche tail are correspondingly noisier. Where sample size falls below about 25 within-directory submissions, results in this article are flagged as suggestive rather than conclusive.

Industry spread across 14 verticals

The fourteen verticals in the sample were: home services (plumbing, electrical, HVAC), legal services, dental and medical practices, accounting and bookkeeping, real estate brokerages, automotive repair, restaurants and food service, fitness and wellness, beauty and personal care, retail (independent), construction and trades, IT services and consultancies, education and tutoring, and event services. The distribution was uneven. Home services accounted for roughly 22% of the sample, while education and event services together accounted for less than 8%. Vertical-level claims in this article are weighted accordingly, and the smallest verticals are flagged where the within-vertical inferences depend on samples of fewer than 100 submissions.

14-month tracking window

The tracking window of fourteen months was chosen to span at least one full annualised cycle of search engine algorithm updates and to let the slow-moving citation effects show up. The window is long enough to capture sustained ranking changes rather than the immediate post-submission noise that confounds shorter studies, and short enough that the dataset is not contaminated by long-running directories that had since shut down or significantly redesigned during the period. Submission dates fell across the first eight months of the window, with the final six months reserved for outcome measurement.

Metrics and attribution method

The dependent variables tracked were: (a) ranking position for a defined set of priority keywords, measured weekly via rank-tracking APIs; (b) local pack inclusion, measured as a binary at each weekly check; (c) referral traffic from each directory to the business website, measured via UTM-tagged listing URLs where the directory permitted them, and via referrer-string parsing where it did not; (d) NAP consistency score, measured via a third-party citation auditing tool at submission and at quarterly intervals thereafter; (e) cost per acquired listing, calculated from agency invoices divided by completed submissions.

Attribution is the perennial weakness of any directory study, and this dataset is no exception. Because directory submissions rarely happen in isolation, since most clients are simultaneously running content, link-building, on-site optimisation and paid campaigns, isolating the directory effect takes careful design. The approach taken here was a difference-in-differences method, comparing the trajectory of treated keywords (those for which directories were submitted) against the trajectory of matched control keywords (similar in baseline ranking, search volume and competitive intensity, but for which no directory work was performed). The residual is the directory-attributable change. The method is imperfect, since it cannot fully control for spillover effects, but it is more conservative than the pre/post comparisons that dominate vendor case studies.

Performance data side by side

Ranking movement after submission

Ranking movement is the metric practitioners most often cite when justifying directory work to clients, and the one most contaminated by attribution confusion. The figures below should be read as difference-in-differences residuals rather than raw post-submission lifts, which would be substantially larger and substantially less defensible.

DA-focused ranking lifts

The DA-focused subset of the sample, defined as portfolios in which 80% or more of submissions targeted directories at DA 40 or above regardless of topical relevance, produced an average residual ranking improvement of 2.4 positions across priority keywords over the fourteen-month window. The distribution was skewed: roughly a third of DA-focused portfolios produced no measurable residual improvement, while a long tail of high performers pulled the mean upward. The median improvement was 1.1 positions, which suggests that for the typical DA-focused portfolio the ranking effect was modest and easily lost within the noise of competing optimisation activities.

More interesting than the headline number is where the gains sat within the SERP. DA-focused submissions correlated most strongly with movements in the bottom half of page one and the top of page two, that is, with movement among already-ranking pages rather than with breakthrough into the top three or the local pack. For keywords that began outside the top fifty, DA-focused submissions had a near-null effect, which suggests the strategy is more useful for consolidating existing rankings than for establishing new ones.

Citation-focused ranking lifts

The citation-focused subset, defined as portfolios in which 80% or more of submissions targeted directories selected for topical or geographic relevance with DA functioning only as a tiebreaker, produced an average residual ranking improvement of 3.7 positions. The median was 2.6, which points to a less skewed distribution: more portfolios produced moderate gains, fewer produced either spectacular results or null results. The distributional difference is at least as informative as the difference in means.

The position-of-gains pattern was also different. Citation-focused submissions correlated more strongly with movements into and within the top ten, and notably with movements into the local pack from prior positions outside it. For keywords with strong local intent, those including a city name, neighbourhood reference, or “near me” pattern, the citation-focused effect was roughly 2.3 times the DA-focused effect. For keywords with no local intent, the effects converged: both strategies produced modest residuals, and neither dominated.

Local pack visibility changes

Map pack inclusion rates

Map pack inclusion is binary at the per-keyword, per-week level, but aggregates to a continuous rate over the tracking window. For DA-focused portfolios, the average inclusion-rate change from the pre-submission baseline to the post-submission steady state was +6 percentage points. For citation-focused portfolios, the change was +18 percentage points. The three-fold ratio fits the broader argument that local pack ranking is governed by entity-confidence signals more than by link-authority signals, and that citation-focused work loads more weight on the former.

Average position shifts

Within the local pack itself, that is, conditional on being included, average position shifts were smaller and more comparable between the two strategies. DA-focused portfolios moved the average local pack position from a baseline of 2.4 to 2.1 (lower numbers being better, with 1 being the top of the three-pack). Citation-focused portfolios moved it from 2.3 to 1.7. The difference is meaningful but less stark than the inclusion-rate difference, which is the more consequential metric: being in the pack at all is the threshold question; position within the pack is the secondary refinement.

Geographic reach differences

Citation-focused portfolios produced wider geographic reach, measured by the number of distinct cities or neighbourhoods in which the business appeared in the local pack at least once during the window. The mean for citation-focused portfolios was 14.2 distinct geographies; for DA-focused, 8.6. The difference comes from the inclusion of geographically scoped directories in citation-focused portfolios, such as neighbourhood guides, city-specific business listings and regional chamber properties, which anchor the business to specific geographies in ways that horizontally scoped national directories do not.

Referral traffic patterns

Click-through volume by strategy

Referral traffic from directories themselves, separate from the indirect effect of improved organic rankings, is a frequently overlooked dimension of directory ROI. DA-focused portfolios delivered an average of 47 monthly directory-referred sessions per business; citation-focused portfolios delivered 112. Part of the difference comes from the fact that niche directories, despite lower DA, often have more engaged user bases who arrive with explicit category-relevant intent. A user browsing a plumbing-specific directory in a defined service area is closer to a transaction than a user clicking through a horizontally scoped national listings site.

See Table 1 for a comparison of the referral traffic dimensions across the two strategies, broken down by traffic volume, engagement signals and conversion behaviour.

Table 1: Referral Traffic Comparison Between DA-Focused and Citation-Focused Portfolios

Bounce rate comparisons

The bounce rate difference, 71% for DA-focused referrers versus 49% for citation-focused, is one of the more striking findings in the dataset and one that tends to surprise practitioners who have spent years pursuing high-DA placements. The reading is that users arriving from horizontally scoped directories often arrive with imprecise intent, frequently bounce when they encounter a specialised business that does not match the breadth of their original query, and rarely go further into the site. Users arriving from category-relevant directories arrive with intent already narrowed by the act of using a category-specific directory, and they convert that narrowed intent into deeper site engagement.

Conversion rate variance

The conversion rate variance compounds the bounce rate finding. Phone-call conversion rates per 100 sessions were 3.5 times higher for citation-focused referrals; form-fill conversion rates were nearly three times higher. When this variance is carried through to cost-per-acquisition calculations, the implications for budget allocation are substantial, and considerably more so than the headline ranking differences would suggest on their own.

NAP consistency score impact

Name, address and phone consistency across the citation graph is, by construction, more responsive to citation-focused work than to DA-focused work, since citation-focused portfolios target a wider variety of directories that contribute to the consistency score in the first place. The dataset bears this out: citation-focused portfolios moved the average NAP consistency score from a baseline of 64% to 89% over the fourteen-month window. DA-focused portfolios moved it from 67% to 76%. The DA-focused improvement is real but capped by the strategy’s tendency to concentrate submissions in a smaller set of horizontally scoped properties, leaving the long tail of vertical and geographic citations unaddressed.

The reading here requires care. NAP consistency is not itself a ranking factor in any direct, mechanistic sense; it is a proxy for the underlying entity-confidence signal that search engines build from the citation graph. Improvements in NAP consistency driven by submission to fictitious or low-quality directories can raise the score without improving the underlying signal, which inflates the metric without improving the outcome. The citation-focused submissions in this dataset were screened for directory legitimacy before submission, so the score improvements should be read as improvements in the legitimate citation graph rather than as gaming of the consistency tool.

Cost per acquired listing

The cost-per-acquired-listing comparison is where the operational implications of the two strategies become most concrete. DA-focused submissions averaged GBP 34 per acquired listing in agency-billed cost, ranging from GBP 8 (low-DA inclusions) to GBP 180 (premium high-DA placements requiring editorial review). Citation-focused submissions averaged GBP 19 per acquired listing, ranging from GBP 0 (free vertical directories) to GBP 75 (paid premium category placements). The difference reflects the fact that high-DA general directories command a price premium that niche directories usually do not, even though the niche directories often deliver better outcomes on the metrics that matter to clients.

When cost is divided by the conversion outcomes from Table 1, the cost-per-conversion gap widens further. A DA-focused portfolio that costs GBP 34 per listing and produces 1.1 phone calls per 100 referred sessions is delivering acquisitions at materially higher cost than a citation-focused portfolio that costs GBP 19 per listing and produces 3.8 calls per 100 sessions. The arithmetic is not subtle.

Reading the evidence strength

Statistically defensible findings

Not all of the findings above carry equal evidential weight. The differences in map pack inclusion rates, NAP consistency improvements, referral session volume, bounce rates and conversion rates are all statistically defensible at conventional thresholds (p < 0.01) given the sample sizes involved, and the effect sizes are large enough to survive substantial adjustments for the confounders discussed below. These are the findings on which practitioners can act with reasonable confidence.

The ranking movement findings are also statistically defensible, but the effect sizes are smaller in absolute terms and the variance is wider. A client adopting a citation-focused strategy can reasonably expect map pack improvements; whether they can expect specific ranking improvements depends on competitive and on-site factors that the directory submission itself does not control. The advice that follows is to anchor client expectations on the more reliable metrics, pack inclusion, referral traffic and conversion rate, rather than on the noisier ranking-position metrics that vendors prefer to feature.

Weak signals worth noting

Several findings in the dataset reach the level of suggestive but fall short of statistical significance. The first is a pattern in which citation-focused portfolios appear to deliver returns more durably across algorithm updates, that is, the ranking and pack gains attributable to citation work appear to survive subsequent algorithm shifts at higher rates than the gains attributable to DA-focused work. The signal is consistent across the three minor algorithm updates that fell within the tracking window, but the number of update events is too small for confident inference. Practitioners should treat this as a hypothesis worth tracking rather than an established finding.

The second weak signal involves a possible interaction between citation-focused work and review acquisition: businesses that combined citation-focused directory submissions with active review-acquisition campaigns appeared to produce roughly 1.4 times the map pack lift of businesses pursuing either strategy alone. The interaction is plausible, since both activities feed entity-confidence signals, but the within-cell sample sizes for the interaction analysis are too small for the headline number to carry much weight.

The third weak signal involves seasonality. There is suggestive evidence that the marginal value of citation-focused submissions varies with the seasonality of the underlying business, that home services see larger lifts during shoulder seasons when overall query volume is moderate, while restaurants see comparable lifts year-round. The pattern is interesting but the seasonal cell sizes are too small to support strong claims.

Confounding variables identified

The most important confounder in any directory study is the simultaneity of other optimisation work. Businesses that retain agencies for directory submissions typically also retain them for content, technical SEO, on-site optimisation, link-building, or paid search. Disentangling the directory contribution from the rest of the agency’s output is methodologically fraught. The difference-in-differences approach used here mitigates the problem but does not eliminate it.

The second confounder is industry-level differences in the directory ecosystem itself. Some verticals, such as legal services and medical practices, have well-developed niche directory ecosystems with high consumer awareness and well-trafficked properties. Others, such as IT consultancies and B2B services, have thinner ecosystems where citation-focused work has less material to work with. The aggregate finding that citation-focused outperforms DA-focused is averaged across this heterogeneity, and the section on business-type variation below disaggregates it.

The third confounder is the relationship between the agency providing the data and the strategy choice. Agencies that built their service offerings around citation-focused work will have selected clients differently and executed differently than agencies oriented toward DA-focused work. The dataset attempts to control for this by including agencies of both orientations, but residual selection effects probably remain. As the work of Khanna and Palepu in Harvard Business Review (1997) reminds us, strategies that look universally superior in one context can prove inappropriate when transferred to another, and that caution applies here as much as it does to corporate strategy.

Where sample size falls short

The dataset is well-powered for aggregate comparisons across the two strategies and for moderately disaggregated comparisons by major vertical. It is poorly powered for comparisons within small verticals (education, event services), for comparisons by specific directory (most directories receive too few submissions for within-directory inference), and for comparisons across multiple time-of-submission cohorts. Practitioners should treat findings about specific directories as suggestive only, even when those findings come from named-directory case studies elsewhere in the trade press; the within-directory sample sizes that support such studies are usually inadequate.

Why the numbers diverge by business type

The aggregate comparison between DA-focused and citation-focused strategies is informative, but the variance by business type is at least as important for practical decisions. The same strategy that produces large gains for a plumbing business in a mid-sized city may produce modest gains for a regional retail chain or near-null gains for a freelancer working remotely. The disaggregation matters.

Service area businesses

Plumbers and electricians data

Service area businesses, those whose customers do not visit a fixed location and whose service is delivered at the customer’s premises, show the largest divergence between the two strategies. For plumbers and electricians in the dataset, citation-focused portfolios produced an average map pack inclusion-rate improvement of +24 percentage points; DA-focused portfolios produced +5 percentage points. The mechanism is fairly transparent: service area businesses depend heavily on geographic-radius signals, and geographic radius is partly inferred from the citation graph. Citation-focused work that anchors the business to specific service areas via geographically scoped directories produces exactly the signals the local algorithm is reading.

The finding is consistent across the seventeen geographies for which the home services subsample is large enough to support within-geography inference. It is most pronounced in second-tier and third-tier cities where competitive density is moderate; it weakens in major metros where the competitive density of the local pack makes any single optimisation lever less decisive.

Mobile service provider results

Mobile service providers, such as pet groomers, car detailers and mobile mechanics, show similar but slightly smaller effects. The citation-focused advantage in map pack inclusion is +19 percentage points, versus +4 percentage points for DA-focused. The reduction relative to plumbers and electricians appears to come from the smaller number of mobile-specific directories in the citation-focused inventory, which limits the strategy’s upside. Where mobile-specific directories exist, they perform strongly; where they do not, the citation-focused portfolio defaults to general service-area directories that perform less distinctively.

Brick-and-mortar retail

Brick-and-mortar retail produces a more compressed difference. For independent retailers in the dataset, citation-focused portfolios produced map pack inclusion-rate improvements of +12 percentage points; DA-focused portfolios produced +7 percentage points. The difference is real but smaller than for service area businesses. The reading involves the role of the physical address itself: a fixed retail address is already a strong entity-confirmation signal, and additional citations are confirmatory rather than constitutive. The marginal citation does less work for a business whose address is unambiguous than for a service business whose service-area boundaries are inherently ambiguous.

Within retail, the difference widens for multi-category retailers (for example independent department stores, general merchants) and narrows for single-category specialists (for example dedicated bookstores, single-line boutiques). The single-category specialists derive comparatively more value from DA-focused submissions to authoritative general directories, perhaps because their category alone is enough to clarify what the business does without further citation reinforcement. As discussed in analysis, the relationship between category specificity and directory selection deserves more careful attention than it typically receives in agency briefings.

Multi-location brands

Franchise model outcomes

Franchise operations bring their own complications. The brand-level entity is well established, often with strong horizontal directory presence inherited from the franchisor; the location-level entities are weaker and depend on local citation work to establish themselves as distinct entities within the citation graph. Citation-focused work at the franchisee level produced map pack inclusion-rate improvements of +21 percentage points in the dataset, against +3 percentage points for DA-focused work. The DA-focused null is striking and probably reflects that the brand has already saturated the high-DA general directories at the corporate level, so additional submissions to the same directory ecosystem produce minimal marginal information for the search engine.

The asymmetry has practical implications for franchisee budgets, which are frequently directed toward DA-focused work on the rationale that “brand presence” is what matters. The data suggest that franchisee-level budgets are better directed toward citation-focused, location-specific work that distinguishes the franchisee’s location from the corporate aggregate. As Davenport, Wang and Tiwari note in Harvard Business Review (2023), the role of data infrastructure in producing measurable value depends on whether the right things are being measured in the first place, and franchise marketing measurement is a domain where the right things are routinely conflated with the easy things.

Regional chain performance

Regional chains, those operating five to fifty locations across a contiguous region, fall between independent businesses and large franchises in their response patterns. Citation-focused work produced +17 percentage points in map pack inclusion against +6 percentage points for DA-focused. The difference runs in the same direction as franchises but smaller in magnitude, reflecting that regional chains often have less corporate horizontal-directory saturation than national franchises and therefore retain more upside from DA-focused work at the chain level.

Solo practitioners and freelancers

Solo practitioners and freelancers, such as independent consultants, freelance designers and sole-trader tradespeople, are the segment for which the data is most ambiguous. The within-segment sample is smaller than for the other categories (n = 287 submissions), and the practitioners themselves are heterogeneous in service mix and geographic anchoring. The headline difference favours citation-focused work (+11 percentage points map pack inclusion versus +5 percentage points for DA-focused), but the confidence interval is wide enough that the difference is statistically marginal.

The qualitative pattern within the segment is that solo practitioners with strong geographic anchoring (a home-based service business operating in a defined city) respond much like other service area businesses, while solo practitioners operating remotely (consultants serving clients nationally) respond like neither pattern. For these remote cases, neither strategy produces large local-pack effects, because the local pack is not the relevant ranking surface in the first place. The advice for remote solo practitioners is to deprioritise local directory work in general and focus on topical authority-building through other channels, since the local-pack mechanism that makes citation-focused work valuable does not apply to their search context.

The data in Table 2 shows the cross-segment differences in a single view, ranking the segments by the size of the citation-focused advantage and noting the within-segment sample size that supports each comparison.

Table 2: Map Pack Inclusion-Rate Improvement by Business Segment

The hidden cost of chasing DA scores

The cost-per-listing difference discussed earlier captures the direct procurement cost of the two strategies, but it understates the total cost of a DA-focused approach because it omits several indirect costs that have piled up as DA-focused thinking entrenched itself across the industry. The first of these is opportunity cost. Every hour of agency time and every pound of client budget directed toward chasing high-DA placements is an hour and a pound not directed toward the citation, content or review work that the data suggest produces better outcomes. In a budget-constrained engagement, which is to say almost every engagement, the implicit trade-off is consequential. A client spending GBP 2,000 a month on directory work allocated 80/20 toward DA-focused submissions is, on the numbers in this dataset, leaving roughly half their available local-pack lift on the table.

The second indirect cost is reporting opacity. DA-focused reporting templates train clients to judge directory work on the wrong metrics. A monthly report featuring “12 new placements at average DA 47” looks impressive to a non-specialist client and creates expectations that later reports must meet or exceed, whether or not DA 47 placements are actually moving the business outcomes the client cares about. When the client eventually asks why the leads have not increased proportionally, the agency has trained itself out of being able to answer, because the metrics it has been reporting do not connect to the metrics the client cares about. The Deloitte CFO Insights observation that “what gets measured and rewarded aligns with the behaviors and strategic objectives that the organization seeks” cuts both ways: a metric that misaligns measurement from objective produces behaviour that misaligns from objective.

The third indirect cost is the slow drift of the directory ecosystem itself. When agency demand concentrates on high-DA general directories, those directories accumulate listings, become increasingly cluttered, and produce diminishing referral value per listing for any given business, a tragedy-of-the-commons dynamic in which the strategy’s success plants the seed of its own diminishing returns. Niche directories, starved of agency demand, struggle to monetise, sometimes shut down, and shrink the available inventory for citation-focused work in the future. The strategy the data favours is also the strategy that supports the long-term health of the directory ecosystem on which both strategies ultimately depend. The strategy the data disfavours is, perversely, the one whose dominance threatens that ecosystem.

The fourth indirect cost is the harder-to-quantify cost of misdirected attention. To borrow Ed Batista’s framing in Harvard Business Review (2015), focused attention is the scarce resource at the leadership level, and the question of where to direct it is itself a strategic decision. Agencies and in-house teams that have built careers on DA-focused knowledge have developed pattern recognition, vendor relationships, and reporting infrastructure organised around DA-focused work. Reorienting that attention toward citation-focused work involves real switching costs, not just in tooling but in the harder-to-replace muscle of judgement. The dataset suggests the switch is worth making, but understanding why it has not happened more widely means acknowledging that the costs of switching are non-trivial and land immediately, while the benefits are diffuse and accrue over months. That asymmetry is a recipe for inertia, and inertia is what the industry has produced.

None of this argues that DA is irrelevant or that DA-focused work has no place. It argues that DA has been operating as a master metric for selection decisions it is poorly suited to drive on its own, and that the persistence of its dominance reflects the convenience of the metric rather than its predictive validity. The shift to a more nuanced selection logic is overdue.

Building a hybrid submission framework

Tiered directory prioritisation

The practical takeaway from the data is not that DA-focused work should be abandoned but that it should sit below a tiered selection logic in which topical and geographic relevance are the primary filters and DA works as a within-tier sort. The hybrid framework below operationalises that logic. It defines three tiers, primary, secondary and tertiary, with different selection criteria, different submission cadences and different reporting expectations.

The primary tier comprises directories that satisfy at least two of these three conditions: vertical specificity to the business’s category, geographic specificity to the business’s service area, and inclusion in the data-aggregator graph that feeds downstream listings. Submissions to this tier are non-negotiable for any business pursuing local visibility. DA is not a filter at this tier: directories that meet the relevance criteria are submitted to regardless of DA, on the rationale that the citation contribution to the entity graph matters more than the link contribution from any single property. A recent this blog post noted that the long tail of niche, geographically specific directories often produces disproportionate value relative to their DA scores, and the data presented above is consistent with that observation.

The secondary tier comprises broader directories that score above DA 30 and have credible category coverage even if they are not category-specialised. This is the tier in which DA does meaningful work: among directories of comparable relevance, higher-DA properties earn priority. The submission cadence to this tier is paced, usually two to four submissions per month rather than the front-loaded burst the primary tier may receive, to avoid the velocity-anomaly patterns that aggressive submission flurries can trigger.

The tertiary tier comprises high-DA general directories with limited category or geographic specificity. Submissions to this tier are useful for broad authority signalling but should be treated as supplements rather than priorities. Budget allocation here is residual, what is left after the primary and secondary tiers have been served, rather than primary.

Budget allocation ratios

The ratio the dataset supports as a starting allocation is roughly 55% of directory budget to the primary tier, 30% to the secondary tier, and 15% to the tertiary tier. The allocation should be adjusted by business segment: service area businesses can push the primary tier share higher (to 65% or so) at the expense of the tertiary; brick-and-mortar retail can hold closer to a 50/30/20 split given the smaller observed differences; remote solo practitioners should probably move the bulk of their notional directory budget away from directory work entirely, given how thin the local-pack mechanism is for their search context.

A common objection from agencies is that this allocation takes more time to execute, since niche directories vary widely in submission interface, approval timelines and editorial requirements, while horizontally scoped directories are routinised and can be processed in bulk. The objection is fair, but it is an argument about agency operating model, not about client outcomes. Agencies whose operating models are built around bulk-processing high-DA general directories will need to reconfigure their workflows; clients should not subsidise the convenience of legacy workflows by accepting weaker outcomes.

Workflow for a mixed strategy

Monthly submission cadence

The monthly submission cadence under a hybrid framework usually comprises a front-loaded primary-tier sprint in the first two months of an engagement (often eight to fifteen submissions in each of those months, depending on the inventory available for the business’s vertical and geography) followed by a steady-state cadence of three to five submissions per month, spread across all three tiers, for the rest of the engagement. The front-loading reflects the asymmetric value of establishing the primary citation graph quickly, since gains compound once the entity-confidence baseline is in place, so accelerating that baseline accelerates the compounding.

The steady-state cadence is more important than its modest volume might suggest. Continued, paced submissions during the steady-state period tell the search engine that the entity is active and that its citation profile is being maintained. Engagement patterns that front-load all submissions and then go silent produce decay in citation freshness signals over the following twelve months; engagement patterns that maintain a paced steady-state cadence preserve that freshness.

Quarterly audit checkpoints

Quarterly audit checkpoints serve three functions. The first is NAP consistency verification, confirming that the citation profile has not drifted due to directory mergers, redesigns, or upstream data-feed changes. The second is competitive review, confirming that the business’s citation profile has stayed competitive with the directly competing businesses in the same geography and vertical, since competitive citation work by rivals can erode the relative position of a previously well-cited business. The third is strategy reassessment, reviewing the actual outcomes against the projected outcomes and adjusting the tier allocation accordingly.

The third function is the one most often skipped. Agencies tend to treat the initial strategy as a fixed plan rather than a hypothesis to be tested, and clients tend to defer to the agency’s continued execution of the original plan even when interim data suggests adjustments are warranted. The quarterly audit is the structural mechanism that forces the conversation. It should be diarised at engagement initiation and held even when the data appears to be cooperating, because the audit’s value is in establishing the discipline of reassessment, not just in catching the cases where reassessment is urgently needed.

What practitioners should change tomorrow

Stop filtering by DA alone

The first change is the simplest to state and the hardest to operationalise: remove DA as the primary filter in directory selection workflows. This means replacing DA-thresholded directory shortlists with relevance-thresholded shortlists, and using DA only as a within-relevance sort. The procurement implication is that vendor and platform shortlists, which are typically DA-sorted by default, need to be re-sorted before they enter the selection workflow. The tooling implication is that custom relevance scoring may need to be built, since few off-the-shelf tools offer relevance-first sorting natively. The reporting implication is that monthly reports need to be re-templated to lead with relevance and outcome metrics rather than DA averages.

Each of these implications is friction. Together they explain why the change has not been made widely despite the underlying data favouring it for some years. Practitioners who make the change earliest in their markets capture an information advantage over competitors who keep optimising to the wrong metric.

Track citation velocity instead

The second change is to replace DA-trend reporting with citation-velocity reporting as the headline metric in client-facing dashboards. Citation velocity, the rate at which new, legitimate citations are being added to the entity graph, is more directly connected to the local-pack outcomes clients care about than DA-trend metrics are. It also has the advantage of being interpretable to non-specialist clients: a graph showing citation count rising from 47 to 89 over six months tells a clearer story than a graph showing DA-weighted authority scores rising from 312 to 408 in the same period.

The shift in headline metric should come with a shift in the supporting metrics clients see. NAP consistency score, map pack inclusion rate, and directory-referred conversion rate should be promoted to dashboard prominence; DA-related metrics should be demoted to appendix status, available for clients who ask for them but not centred in the narrative. As the findings from this article suggest, the metric a client looks at most often is the metric they will optimise their feedback toward, and aligning that metric with the underlying outcomes is itself an act of strategic design.

Reallocate spend toward niche directories

The third change is the budget reallocation discussed earlier, moving roughly half of directory budget into the primary tier of vertically and geographically specific directories. The reallocation is straightforward in principle but requires sourcing work that many agencies have not historically invested in. Niche directory inventory does not present itself in the convenient sortable lists that horizontally scoped directories do; it has to be assembled through vertical research, geographic research, competitor citation analysis, and local network knowledge. Agencies that have outsourced their directory selection to off-the-shelf tools will need to reinsource the selection function, at least partially, to execute on the reallocated budget.

The cost of reinsourcing is real but recoverable. Vertical and geographic directory inventory, once assembled for a given business, is largely reusable across clients in the same vertical and geography. Agencies that build out this inventory systematically across their client base accumulate a proprietary asset that is difficult for competitors to replicate quickly, and that supports the citation-focused strategy at marginal cost on each successive client engagement. The inventory itself becomes a competitive moat, in a way that DA-sorted shortlists generated from off-the-shelf tools never could.

Table 3 below summarises the findings as a single comparative reference, presenting the headline differences, the strength of the underlying evidence, and the recommended action in compact form.

Table 3: Summary of Findings and Recommended Actions

Where the evidence points next

The data assembled here makes a defensible case that citation-focused selection outperforms DA-focused selection on the outcomes local businesses actually care about, and that the persistence of DA-focused dominance reflects measurement convenience rather than predictive validity. That case is strong enough to support the practical recommendations laid out above. It is not strong enough to settle the deeper question the analysis raises but does not answer.

The question is this: as search engines keep developing entity-based and AI-mediated retrieval, moving further from the link-graph paradigm that gave DA its conceptual foundation, will citation-focused work itself remain the right successor strategy, or will it be superseded by something the industry has not yet named? The evidence suggests that entity confidence is currently the operative mechanism for local pack outcomes, and that citation-focused work feeds entity confidence. But the evidence is silent on what entity confidence will look like five years from now, when retrieval is more heavily mediated by language models that may infer entity properties from sources beyond the traditional citation graph: from review text, from transcribed phone interactions, from structured data exposed by point-of-sale systems, from the customer photograph corpus that has built up around the business.

If those signals come to dominate, the citation graph as currently understood may matter less, and the strategy the present data favours may need to be revised in turn. The practitioner who switches from DA-focused to citation-focused work is making a defensible bet on the current state of the algorithm, but not a permanent one. The discipline of reassessment that the quarterly audit checkpoint formalises is the only durable response to a search ecosystem in which the underlying mechanism keeps changing.

Which leaves the question genuinely open: if the metric that organises a strategy is itself a snapshot of an algorithmic moment, what is the right cadence at which selection logic should be re-derived from outcomes, rather than carried forward from the assumptions that produced the previous round of work?