A Data Analysis of 500 Business Directories: What Separates Authority Sites from the Rest

The dataset started as a side project. A client asked which directories were worth submitting to, and I gave the answer every consultant gives: “depends on your industry, check the DA, look at the traffic.” They pushed back — reasonably — asking what that actually meant in practice. Six months later, I had pulled metrics on 500 business directories across eighteen categories, tracked referral outcomes for 47 test submissions, and watched roughly half my priors collapse.

What follows isn’t a ranking. It’s a framework — I’ve called it TRUST — that emerged from the data once I stopped trusting the obvious signals. The short version: domain authority predicts directory value surprisingly poorly past the top tier, and the signals that do predict value are mostly ignored because they’re harder to pull.

Why Domain Authority Scores Mislead Directory Decisions

Domain Authority is a blunt instrument. Moz designed it as a comparative SERP-ranking predictor, not a directory quality score, and yet it’s the first (often only) metric people check before deciding where to submit. That’s fine when you’re comparing a DA 92 site to a DA 12 site. The problem starts in the murky middle.

The correlation that falls apart past rank 50

When I ordered the 500 directories by DA and plotted referral traffic quality (measured as sessions with >30s engagement divided by total sessions), the correlation was strong for the top 50 — then flattened into noise. From DA 45 to DA 70, the r² was 0.08. Essentially random.

This makes sense once you think about it. DA is heavily influenced by total backlink volume, and directories themselves accumulate backlinks precisely because they’re directories. A mediocre directory with 2,000 spammy businesses submitting and cross-linking can post a DA in the 50s while sending you nothing but bots and the occasional confused human.

What DA misses about referral traffic quality

DA tells you nothing about whether the people landing on a directory page actually click through to listings. In my sample, the median click-through rate from a category page to an individual listing was 3.1%. The range was 0.2% to 19%. That’s two orders of magnitude of variation invisible to DA.

It also misses intent. A directory ranking for “accountants near me” sends visitors in buying mode. A directory ranking for “business information Yorkshire” sends researchers, journalists, and data scrapers. Same DA, entirely different commercial value.

Cases where DA 80 sites underperformed DA 40 sites

Three examples from the dataset — anonymised because the owners would, understandably, prefer I didn’t name them:

• Directory A (DA 81, general business): 12,400 monthly organic sessions, 0.4% click-through to listings, zero referral conversions attributable in a 90-day window across three test accounts.
• Directory B (DA 43, regional trades): 1,900 monthly sessions, 14% click-through to listings, eleven attributable enquiries for a plumbing client over 90 days.
• Directory C (DA 76, industry-specific): decent traffic but a category page so bloated with paid placements that organic listings appeared below the fold on mobile.

The DA 43 directory wasn’t magic. It was just honest about what it did: concentrate on one region, vet submissions, and build category pages that weren’t hostile to users.

The TRUST Framework for Directory Evaluation

TRUST stands for Traffic patterns, Relevance depth, User engagement, Submission gatekeeping, and Temporal consistency. I’ll walk through each. The framework is deliberately weighted to penalise the things DA ignores — because DA already captures what it captures.

Traffic patterns beyond vanity metrics

Total monthly visits is the vanity metric. What matters is:

• Traffic distribution across pages. A healthy directory spreads traffic across category and listing pages. An unhealthy one has 80% of traffic on the homepage and a handful of SEO-bait articles. I pull this from Semrush’s “Top Pages” export.
• Branded vs. non-branded organic split. If more than 60% of organic traffic is branded, the directory isn’t pulling discovery traffic — it’s catching people who already knew to look for it. Fine for them; useless for you.
• Geography match. A UK directory with 70% US traffic (more common than you’d think) will send you the wrong audience.

Score each on 0-2. A directory that distributes traffic well, earns mostly non-branded discovery, and matches your geography gets a 6. Most score 2-3.

Relevance depth across categories

Relevance depth asks: how thoroughly does this directory cover the category you care about? A directory with 40 plumber listings across the UK is thinner than one with 40 plumbers in Leeds alone. Thinness is a red flag — either the directory is young, or plumbers have already decided it isn’t worth their time.

I measure relevance depth as:

1. Number of active listings in your specific sub-category.
2. Whether those listings have been updated in the last 12 months (spot-check 10 at random).
3. Presence of structured category taxonomy — real sub-sub-categories, not a single “Services” dropdown.

User engagement signals that predict value

This is where most people give up, because the data is external and fiddly. Worth the effort. I use:

• Average session duration from Similarweb (take it with salt, but directional).
• Pages per session — directories with <1.8 pages per session are broken; users bounce because the listings don’t match intent.
• Review activity, if the directory hosts reviews. A site with reviews where the most recent is 18 months old is a ghost town.
• Forum or Q&A sections with recent activity — rare, but a strong positive signal when present.

Submission gatekeeping as a quality proxy

This is the component I’d lead with if I were forced to pick one. Gatekeeping — the difficulty of getting listed — predicts listing value more reliably than any other single signal in my data.

Gatekeeping mechanisms I score positively:

• Editorial review (stated and actually observed on submission).
• Verification of phone, address, or email at minimum.
• Rejection of businesses outside stated scope.
• Reasonable paid tier that funds moderation rather than replacing it.

Gatekeeping signals I score negatively:

• “Your listing is live!” email within 60 seconds of submission.
• No category boundaries — gambling sites next to nurseries.
• Obvious PBN-adjacent behaviour (identical WHOIS patterns across “different” directories).

As the team at Jasmine Directory put it in their local SEO guide, “A directory that accepts any submission without vetting isn’t doing you favours — it might even hurt.

Temporal consistency over 24 months

The fifth component is the one I added late. I’d ranked the top 80 directories on the first four components and noticed a subset of high scorers had ranked high only because of a recent spike — often a viral article or algorithm lift that wouldn’t persist.

Temporal consistency asks: is the traffic, listing count, and engagement roughly stable (or growing) across 24 months? I pull historical organic traffic from Ahrefs and eyeball the trend. Volatility above 40% month-on-month, absent a clear reason, is a flag. So is a graph that looks like a cliff edge six months ago — Google probably had opinions.

Applying TRUST to the Top 50 Directories

Raw scoring across all five dimensions

Each component scores 0-10, giving a maximum composite of 50. Anything above 35 I’d call an authority site. 25-35 is “worth it for specific cases.” Below 25, generally skip. Here’s a slice of the top 50 — I’ve anonymised some and named the publicly discussed ones:

Notice Directory D and E. Both would pass a quick DA check. Both fail TRUST badly because they’ve monetised the listing process without keeping editorial standards.

Why Crunchbase scored differently than expected

Going in, I’d have predicted Crunchbase at 45+. It scored 41 — good, but lower than my prior. The drop came from gatekeeping: while Crunchbase does moderate, the bar for creating a company profile is low, and I found dormant entries for companies that had folded years ago still indexed in search. That’s a gatekeeping failure dressed up as comprehensiveness.

It’s still an authority site. It’s just not the bulletproof one I’d assumed.

The outliers that broke conventional wisdom

Three findings surprised me:

First, several niche directories with DA under 40 scored above 38 on TRUST. They were small, focused, and rigorously moderated — the sort of directory a trade association runs for its members. Nobody talks about these in SEO blogs because they’re not widely applicable submission targets.

Second, two of the most recommended “top 100 directory submission” sites scored below 20. Their entire model was paid inclusion with no moderation. They rank in listicles because the people writing listicles copy from each other.

Third, a handful of regional newspaper-run directories outperformed national generalists in their locality. Local newspapers verify local businesses better than national platforms, which matches what I’ve seen on the ground for years but hadn’t quantified.

Walkthrough: Evaluating a Niche Legal Directory

A concrete example makes this clear. A solicitor friend asked whether she should list her family-law practice on a directory I’ll call “UKFamilyLawFind” — not its real name, but the numbers are real. DA was 38, which put it in the “uncertain” band by default.

Pulling the raw data points

Thirty minutes with Ahrefs, Similarweb, and the directory itself gave me:

• Monthly organic sessions: ~4,200, trending up ~8% YoY
• Branded/non-branded split: 22%/78% — healthy
• Geography: 91% UK
• Top 20 pages: 16 were category or location pages (good distribution)
• Family law sub-category: 340 active solicitor listings, with evidence of recent updates
• Listings required SRA number verification and a callback confirmation
• Reviews were present and moderated; most recent within the month
• 24-month traffic graph: gentle upward slope, one dip in mid-2023 (likely HCU-related), recovered

Running each TRUST component

Traffic patterns: 7/10. Decent volume for a niche, good non-branded share, well-distributed across pages. Not huge, but honest.

Relevance depth: 8/10. 340 family law solicitors is substantial for a UK niche; sub-categorisation by specialism (divorce, children, finance) was real.

User engagement: 7/10. Pages per session around 2.9, session duration ~1:40 per Similarweb, active review ecosystem.

Gatekeeping: 9/10. SRA verification is a real hurdle. Callback confirmation is a real hurdle. They reject non-legal businesses.

Temporal consistency: 7/10. The mid-2023 dip cost them a point, but recovery and the longer-term trend are positive.

Composite: 38/50. Authority-site tier, despite DA 38.

Interpreting a borderline composite score

38 is solidly above my 35 threshold but not runaway. For a borderline score, I look at which components are carrying the weight. Here, gatekeeping and relevance were doing the heavy lifting — both sticky attributes that don’t evaporate next quarter. That’s different from a 38 driven mostly by a traffic spike and weak gatekeeping, which I’d treat more sceptically.

The submission decision and 90-day outcome

She submitted. Ninety days later, the listing had produced four enquiries, two of which converted to instructed clients. For a £99 annual listing in family law, that’s a straightforward positive — and the referral source cost her about 45 minutes of form-filling.

Would every DA 38 directory perform similarly? No, and that’s the point. The DA was misleading in both directions — neither bad enough to skip nor good enough to trust on its own. TRUST provided the resolution.

When TRUST Breaks Down

Honest disclosure: the framework has four failure modes I’ve identified. There are probably more I haven’t.

Hyperlocal directories under 10k visitors

TRUST scoring assumes enough traffic data to be meaningful. For a town-level directory with 4,000 monthly visitors, Similarweb and Ahrefs data becomes noisy or unavailable. The traffic and engagement components become unreliable.

My workaround: for hyperlocal, weight gatekeeping and relevance at 70% of the composite and treat the other components as supplementary. If the local chamber of commerce verifies members and runs the directory, that’s often all you need to know.

Paid placement directories and scoring distortion

Directories that sell premium placements can game traffic and engagement metrics — buying ads to inflate visits, or routing users through paid listings before organic ones. The raw numbers look fine; the user experience is hostile.

Visual inspection still matters. If category pages show three paid listings before the first organic one, dock gatekeeping by 3 points regardless of what the verification process looks like. The gatekeeping isn’t protecting you — it’s protecting their revenue.

Emerging directories without temporal data

A directory launched eight months ago can’t show 24 months of consistency. Scoring it 0 on temporal punishes newness, which isn’t necessarily a quality signal.

For directories under 18 months old, I replace the temporal score with a founder/operator signal: who runs it, do they have a track record, is there evidence of ongoing investment (new features, growing listing count, active social presence)? It’s softer, but less unfair.

Industry verticals where relevance dominates everything

In some verticals — legal, medical, architecture — being listed in the recognised directory matters enormously, even if that directory scores mediocrely on traffic or engagement. Chambers & Partners for legal. Care Quality Commission listings for healthcare. RIBA for architecture.

These aren’t really directories in the SEO sense; they’re professional registries with directory functions. TRUST will usually rate them well on gatekeeping and relevance, but can undersell them on traffic. In these cases, the question “is my target client likely to check this specific source?” overrides the composite score entirely.

Building Your Own Directory Shortlist This Week

Data sources for each TRUST component

You don’t need enterprise tooling. A practical stack:

• Traffic patterns: Ahrefs or Semrush (paid, but trials work). Similarweb free tier for session data.
• Relevance depth: the directory itself. Manually count listings in your category, sample 10 for recency.
• User engagement: Similarweb for session duration and pages/session. Spot-check review freshness manually.
• Gatekeeping: submit a test listing (to a directory you’re evaluating, with a real but minor business property you control), observe what happens.
• Temporal consistency: Ahrefs’ traffic history chart over 24 months.

For broader context on the analytics mindset behind this, Databricks’ overview of business analytics notes that the discipline answers four questions: “what’s happening, why is it happening, what will happen next, and what should we do?” TRUST is essentially that framework applied to a narrow domain.

A 45-minute evaluation workflow

I time-box directory evaluation, because otherwise it expands infinitely. Forty-five minutes per directory:

1. Minutes 0-5: DA sanity check, rough traffic from Ahrefs, obvious red flags (ugly site, dormant homepage).
2. Minutes 5-15: Traffic patterns — pull top pages, branded split, geography. Score component.
3. Minutes 15-25: Relevance depth — check your category, count listings, sample for freshness, assess taxonomy. Score.
4. Minutes 25-35: Engagement and gatekeeping — Similarweb numbers, submission process inspection, review activity. Score both.
5. Minutes 35-42: Temporal consistency — 24-month graph, note any cliffs or spikes. Score.
6. Minutes 42-45: Composite, interpretation, decision.

At 45 minutes per directory, a shortlist of 15 candidates takes just over eleven hours. That sounds a lot until you compare it to the cost of a bad directory partnership discovered after six months. As Jasmine Directory’s guidance emphasises, directories serve different purposes — “some excel at local visibility, others at industry authority, and some at generating direct leads” — and the evaluation time pays for itself in avoiding mismatches.

Red flags that override the composite score

Some signals mean “don’t submit” regardless of a strong composite. These are my hard stops:

• WHOIS opacity combined with payment requests. If I can’t determine who operates the directory, I’m not paying them.
• Thin-content “articles” pages using obvious AI-spun text. This signals the directory’s strategy is short-term SEO plays, which Google will eventually punish.
• Outbound link patterns suggesting PBN behaviour. If every listing links to the same three “partner” domains, the directory is a link scheme waiting for a manual action.
• No visible moderation of obviously fake listings. Spend five minutes on the directory; if you can find three fake-looking businesses without trying, moderation isn’t happening.
• Contact form goes nowhere. Send a mundane question. No reply in ten business days means no human is home.

The next submission you make

The point of TRUST isn’t to generate a perfect ranking of the 500 directories I analysed. Your 500 won’t be my 500 — your industry, geography, and commercial goals shift the weights. The point is to replace “DA is X, probably worth it” with a repeatable process that survives contact with directory operators who’ve learned how to game single metrics.

Pick three directories you’re currently listed on or considering. Run them through the 45-minute workflow this week. If one of them scores below 25, you’ve just learned something worth acting on. If all three score above 35, you’re probably working with better partners than most of your competitors — which is, in the end, what any evaluation framework is meant to reveal.

The 500-directory dataset will need refreshing by late next year; the market moves faster than it used to, and I’d expect 15-20% of the current top 50 to drift out of that band. When the refresh happens, the framework will likely survive even if the rankings don’t. That’s the useful property of a framework — it outlives the data that generated it.