How does ABA Rank decide the order of vendors on a category page?

Every category applies the same ranker to every listing. The order is produced by four disclosed signals, weighted so qualified reviews matter most: qualified review quality and volume (the largest signal, 50% of the Index), profile completeness (30%), recency of last verification (15%), and primary-category fit (5%). None of the four can be purchased. Sponsored entries sit above the ranked list, are visually distinct, and never reorder organic rank.

Does ABA Rank take money to change ranking position?

No. As of methodology v3.0 no paid product contributes any Index points. All 100 points are operational - qualified reviews (the largest share, 50 points), profile completeness, verification recency, and primary-category fit - and a vendor's subscription tier is not an input to the ranking formula at all. Sponsorship buys labeled slots that render above the ranked list, plus profile features; it does not change rank_score and it does not move a vendor within the ranked list. Earlier versions did carry a capped paid tier signal - see the methodology changelog at /how-we-rank for what changed and when. ABA Rank does not accept paid placement disguised as merit and does not run pay-to-rank ads.

Did ABA Rank ever sell ranking position?

Yes, and we stopped. Methodology versions before v3.0 (August 2026) included a tier signal worth up to 15 of the 100 Index points, which a Sponsor subscription received in full, plus an internal editorial multiplier that could raise a score by a further 10% and was not disclosed on this page. Stacked, they were worth up to roughly 24 points. Both were removed from the ranker in v3.0 rather than merely documented, and the 15 points they occupied were redistributed to qualified reviews and profile completeness. The changelog on this page records the change with its effective date.

What is the difference between sponsored placement and ranked listings?

Sponsored slots appear above the ranked list, are flagged as sponsored, and are paid placements. Ranked listings appear below and are produced by the disclosed signals. The two never mix. If a sponsoring vendor also appears in the ranked list, their organic rank reflects only the operational signals - the sponsored slot sits on top of that, not instead of it, and buying one does not move the other.

What does the Sponsor tier actually buy, if not rank?

A labeled slot above the ranked list on eligible category and location pages, profile features, and quarterly print-edition inclusion. It does not contribute Index points, it is not a tie-break, and it is not an input to the ranking formula in any form. A free Claimed listing and a Sponsor listing with identical reviews, completeness, verification recency, and category fit receive identical scores.

How often are rankings recomputed?

The Index is recomputed nightly. Reviewed corrections from the suggest-an-edit queue and freshly verified profiles flow into the next rerank pass. Top 10 leaderboards are republished quarterly and summarised in the print edition.

Which ranking signals matter most for vendors?

Qualified reviews are the single largest ranking signal - their quality, volume, freshness, and verified-interview presence carry the most weight (50% of the Index). Profile completeness is the largest lever a vendor controls directly: every present field (logo, description, website, HQ, founding year, employee band, integrations, pricing, certifications, payor coverage) pulls the score up; blanks pull it down. Verification recency and primary-category fit come next. Upgrading your subscription is not on the list, because it does not affect rank.

How does the clinic finder ranking on /find-aba differ from category rankings?

Find ABA results are sorted into three groups before the within-group sort: claimed first, then pending claims, then unclaimed. Claiming is free, so this is a verification signal rather than a commercial one - a claimed profile means an operator has confirmed the facts we publish. Within each group, results sort by waitlist availability (accepting first), profile completeness, and recency of last verification. A sponsor cap prevents one operator from crowding out a city or state page; it can only move a sponsor down, never up.

Can I see the exact ranking formula and weights?

Yes. ABA Rank publishes the weight of every signal group - machine-readable at /api/v1/methodology (index_weights) - though not the full per-field scoring internals. The signal groups and weights are qualified reviews (0.50), profile completeness (0.30), verification recency (0.15), and primary-category fit (0.05); they sum to 1.0, so there is no room for an undisclosed term. The clinic-finder ranker source is open at lib/rank/clinics-search.ts. Withholding the per-field internals prevents adversarial profile-stuffing while keeping the methodology fully auditable.

What can I do if I think a ranking looks wrong?

File a correction through /suggest-edit. Submissions land in the moderation queue, get reviewed by the editorial team, and reviewed corrections feed directly into the next nightly rerank pass. Corrections are about underlying facts (services, locations, credentials, pricing), not about negotiating position: there is no mechanism for an ABA Rank employee to adjust a vendor's score directly, and all 100 Index points derive from published facts. ABA Rank does not let vendors see neighbours' scores or submit counter-signals.

§ Methodology

How we rank.

Every category page on ABA Rank shows a ranked list of vendors. The order comes from four published signals applied uniformly to every listing in that category. Nothing in that formula is for sale: no tier, no subscription, no editorial thumb on the scale. If a provider is above another one, it is because of their reviews, their profile, and their verification date.

Updated August 5, 2026

§ 01 · Signals

The signals.

Four signals, 100 points, weights that sum to exactly 1.0 - so there is no room in the formula for a term we have not listed here. We publish every weight, machine-readable at /api/v1/methodology, and apply them the same way across every vendor in a category. Qualified reviews carry the most weight (50% of the Index). The full per-field scoring internals stay private to prevent profile-stuffing; the weights themselves do not.

Qualified reviews - 50 points: The largest ranking signal. Published review quality, volume, and freshness all count, and verified reviews carry additional trust because they are tied to a completed verification workflow, not just a public submission. A vendor with strong, genuine reviews outranks one that has merely filled in more profile fields - and outranks one that pays us, because paying us does nothing here.
Profile completeness - 30 points: How fully the profile is filled in - logo, description, website, HQ, founding year, employee band, integrations, pricing information. Every present field pulls the score up; blanks pull it down. This is the largest lever a vendor controls directly, and it is free: claiming a profile costs nothing.
Verification recency - 15 points: Profiles we’ve verified recently are treated as fresher than profiles that have gone a long time without a review. This prevents stale entries from squatting on top slots. The signal decays with time rather than latching, so it rewards keeping facts current rather than passing a check once. A paid Verified review is one way to refresh it; a free listing that keeps its own facts current earns the same points.
Primary-category fit - 5 points: A vendor can live in multiple categories, but every profile names exactly one as its primary. On a category page, vendors for whom the current category is primary are a better fit than vendors listed only as a secondary match. This keeps a practice-management suite from crowding out a session-note specialist on the session-notes page.

§ 02 · Money

What money buys.

ABA Rank sells subscriptions, so the honest question is what they change. The answer is reach and verification - never position in the ranked list. Two vendors with identical reviews, completeness, verification recency, and category fit receive an identical Index score whether one of them pays us nothing and the other pays us $1,490 a year.

What a paid tier buys

A labeled slot above the ranked list
An independent credential and ownership review
A verification badge with a public date
Profile features and a prospect inbox
Quarterly print-edition inclusion

What it does not buy

Index points - tier is not a term in the formula
A higher position in any ranked list
A tie-break against an equally-scored free listing
Placement in the clinic finder ahead of a free claimed clinic
Removal or suppression of a competitor

What we don’t do.

We don’t take money to reorder the ranked list. We don’t hand-tune the order for customers or prospects - there is no admin control that adjusts a vendor’s score, because we removed the one that existed (see the changelog below). We don’t let vendors see their neighbors’ scores or submit counter-signals. We don’t promise top-five placement as part of any commercial agreement. The methodology is applied consistently.

These are testable claims, not promises. The weights sum to 1.0 and are published, so an undisclosed term would have to displace a disclosed one. The scoring function takes four inputs and a subscription tier is not among them. Editorial reorders on cluster pages are logged with a required justification and any vendor can request the entries affecting their listing.

See a ranking that looks wrong? Tell us - the suggest-an-edit form goes to the moderation queue and reviewed corrections flow directly into the next rerank pass (nightly).

§ 04 · Clinic finder

Clinic finder ranking.

Results on Find ABA are sorted into three groups: claimed, pending claims, and unclaimed. A claimed clinic outranks an unclaimed one regardless of completeness or other signals - claiming is free, and a claimed profile means an operator has confirmed the facts we publish about them. Paying changes nothing here: a free claimed clinic and a Sponsor clinic sit in the same group and sort against each other on the same signals.

Within a group, we sort by waitlist availability (accepting first), profile completeness, and recency of last verification. Sponsor clinics carry a visible Sponsored chip, and we cap how many can appear near the top so a single multi-site operator can’t crowd out a city page. That cap can only move a sponsor down the list - it is a diversity guard, not a ranking signal.

State pages, plan-only pages, and state-plus-plan pages use the same ranker as city pages - applied to the broader result set. The three ranking groups, the within-group sort signals (waitlist availability, completeness, last verified), and the sponsor cap all behave identically across every /find-aba surface.

Source · lib/rank/clinics-search.ts

§ 05 · FAQ

Methodology questions, answered.

The questions vendors, families, and reporters ask most often about how the directory is ranked - answered against the same disclosed methodology described above.

How does ABA Rank decide the order of vendors on a category page?: Every category applies the same ranker to every listing. The order is produced by four disclosed signals, weighted so qualified reviews matter most: qualified review quality and volume (the largest signal, 50% of the Index), profile completeness (30%), recency of last verification (15%), and primary-category fit (5%). None of the four can be purchased. Sponsored entries sit above the ranked list, are visually distinct, and never reorder organic rank.
Does ABA Rank take money to change ranking position?: No. As of methodology v3.0 no paid product contributes any Index points. All 100 points are operational - qualified reviews (the largest share, 50 points), profile completeness, verification recency, and primary-category fit - and a vendor's subscription tier is not an input to the ranking formula at all. Sponsorship buys labeled slots that render above the ranked list, plus profile features; it does not change rank_score and it does not move a vendor within the ranked list. Earlier versions did carry a capped paid tier signal - see the methodology changelog at /how-we-rank for what changed and when. ABA Rank does not accept paid placement disguised as merit and does not run pay-to-rank ads.
Did ABA Rank ever sell ranking position?: Yes, and we stopped. Methodology versions before v3.0 (August 2026) included a tier signal worth up to 15 of the 100 Index points, which a Sponsor subscription received in full, plus an internal editorial multiplier that could raise a score by a further 10% and was not disclosed on this page. Stacked, they were worth up to roughly 24 points. Both were removed from the ranker in v3.0 rather than merely documented, and the 15 points they occupied were redistributed to qualified reviews and profile completeness. The changelog on this page records the change with its effective date.
What is the difference between sponsored placement and ranked listings?: Sponsored slots appear above the ranked list, are flagged as sponsored, and are paid placements. Ranked listings appear below and are produced by the disclosed signals. The two never mix. If a sponsoring vendor also appears in the ranked list, their organic rank reflects only the operational signals - the sponsored slot sits on top of that, not instead of it, and buying one does not move the other.
What does the Sponsor tier actually buy, if not rank?: A labeled slot above the ranked list on eligible category and location pages, profile features, and quarterly print-edition inclusion. It does not contribute Index points, it is not a tie-break, and it is not an input to the ranking formula in any form. A free Claimed listing and a Sponsor listing with identical reviews, completeness, verification recency, and category fit receive identical scores.
How often are rankings recomputed?: The Index is recomputed nightly. Reviewed corrections from the suggest-an-edit queue and freshly verified profiles flow into the next rerank pass. Top 10 leaderboards are republished quarterly and summarised in the print edition.
Which ranking signals matter most for vendors?: Qualified reviews are the single largest ranking signal - their quality, volume, freshness, and verified-interview presence carry the most weight (50% of the Index). Profile completeness is the largest lever a vendor controls directly: every present field (logo, description, website, HQ, founding year, employee band, integrations, pricing, certifications, payor coverage) pulls the score up; blanks pull it down. Verification recency and primary-category fit come next. Upgrading your subscription is not on the list, because it does not affect rank.
How does the clinic finder ranking on /find-aba differ from category rankings?: Find ABA results are sorted into three groups before the within-group sort: claimed first, then pending claims, then unclaimed. Claiming is free, so this is a verification signal rather than a commercial one - a claimed profile means an operator has confirmed the facts we publish. Within each group, results sort by waitlist availability (accepting first), profile completeness, and recency of last verification. A sponsor cap prevents one operator from crowding out a city or state page; it can only move a sponsor down, never up.
Can I see the exact ranking formula and weights?: Yes. ABA Rank publishes the weight of every signal group - machine-readable at /api/v1/methodology (index_weights) - though not the full per-field scoring internals. The signal groups and weights are qualified reviews (0.50), profile completeness (0.30), verification recency (0.15), and primary-category fit (0.05); they sum to 1.0, so there is no room for an undisclosed term. The clinic-finder ranker source is open at lib/rank/clinics-search.ts. Withholding the per-field internals prevents adversarial profile-stuffing while keeping the methodology fully auditable.
What can I do if I think a ranking looks wrong?: File a correction through /suggest-edit. Submissions land in the moderation queue, get reviewed by the editorial team, and reviewed corrections feed directly into the next nightly rerank pass. Corrections are about underlying facts (services, locations, credentials, pricing), not about negotiating position: there is no mechanism for an ABA Rank employee to adjust a vendor's score directly, and all 100 Index points derive from published facts. ABA Rank does not let vendors see neighbours' scores or submit counter-signals.

§ 06 · Editorial policy

Editorial policy.

Cluster pages (our “best of” software guides) layer editorial judgment on top of the ranked Index. To keep that layer accountable, every editorial action is governed by the four rules below - and the rules themselves are public.

Cluster seed rules

Every cluster roster starts from the same place: the ABA Rank Index for the cluster’s target category (the rank order recorded in vendor_category.rankPosition). For cross-category clusters like Best ABA Software, we seed from the top-N vendors by global rank score. Editors may reorder vendors after seeding, but any insertion outside the seeded top-N requires an editorial blurb explaining the move - and the override is logged (see below).

Override logging

Every editorial reorder writes an auditLog row with action cluster_roster_override, recording the previous and new sort positions and the editor’s required justification text. The audit trail is permanent and any vendor can request access to entries that affect their listing.

Takedown process

Vendors who believe their inclusion on a cluster page is inaccurate, or who prefer not to be listed, can request removal by emailing editor@abarank.com. We respond within five business days. Approved removals are reflected on the next publish cycle.

Review cadence

Each cluster’s last_reviewed_at timestamp is touched whenever the cluster is (re)published. After 90 days without a re-review, the cluster surfaces a staleness warning in our admin console; editors revisit pricing, integrations, and editorial copy before re-publishing. The “Reviewed {month year}” chip on each cluster page reflects the most recent review.

§ 07 · Changelog

Methodology changelog.

We disclose substantive methodology changes here so vendors and reporters can see exactly how rank inputs evolve over time. Effective dates are when the change began affecting rank order, not when the document was published.

Methodology v3.0 - effective August 2026

Methodology v3.0 removes every paid input from the ranked list. Until this version, a Sponsor subscription bought real Index points, and an internal flag could raise a score further without appearing in any public disclosure. Both are deleted from the scoring code rather than merely documented, because a disclosure cannot fix a formula that families are relying on to compare care providers.

The tier signal is gone. v2.0 weighted listing tier at 0.15 and gave a Sponsor subscription the full value of that term - 15 of 100 Index points, bought. Subscription tier is no longer an input to the ranking formula in any form: not a weighted term, not a tie-break, not a filter.
The undisclosed editorial multiplier is gone. A per-vendor bonus_ranking flag, settable by ABA Rank staff, multiplied a vendor’s finished score by 1.1. It was not described on this page or in /api/v1/methodology. It has been removed from both rankers and the admin control that set it has been deleted.
Stacked, those two mechanics were worth up to 24.1 of 100 Index points to a vendor positioned to benefit from both. That is the number this change removes.
Redistribution: the freed 0.15 went to the two signals that reflect provider quality - qualified reviews 0.40 → 0.50, profile completeness 0.25 → 0.30. Verification recency (0.15) and primary-category fit (0.05) are unchanged. Weights still sum to 1.0.
Clinic finder: the four result groups collapse to three. “Paid & claimed” and “claimed free” merge into a single claimed group, so a paying clinic no longer outranks an equally-complete free one. The hidden 10% completeness bonus that applied within groups is removed.
What sponsorship still buys: labeled slots above the ranked list, profile features, and print-edition inclusion. Those were always rendered separately from organic rank and still are.
Expect rank movement. Vendors who were benefiting from the tier signal or the editorial flag will fall; well-reviewed vendors on the free tier will rise. This is the intended effect, and the first nightly rerank after the effective date applies it.
Effective date: August 2026.

Methodology v2.0 - effective July 2026 (superseded)

Methodology v2.0 makes qualified reviews the center of gravity. Earlier versions weighted profile completeness most heavily; v2.0 rebalances so a vendor’s customers, not the length of its profile, drive rank. The published weights below now exactly match the scoring engine.

Index weights (each signal’s share of 100 points): qualified reviews 40, profile completeness 25, verification recency 15, capped tier signal 15, primary-category fit 5. Machine-readable at /api/v1/methodology.
Qualified reviews (quality, volume, freshness, and verified-interview presence) are now the single largest signal, up from an effective ~13% in v1.x.
The Sponsor tier signal is capped at 15 of 100 Index points; per-page sponsored placements render in labeled slots above the ranked list and never change rank_score. The two mechanics are disclosed separately.
Effective date: July 2026.

Methodology v1.6 - effective May 2026 (superseded)

Methodology v1.6 published the coach (provider) ranking inputs and disclosed a paid-boost ceiling for Featured coaches. In retrospect that disclosure was incomplete: the “at most ~1.5% of the final score” figure below was accurate for the Featured tier but not for Sponsor, which received the full 15-point tier term. v3.0 removes the tier term entirely, which makes the distinction moot.

Coach ranking score = verified credentials + profile completeness + responsiveness + recency.
Paid boost is capped and disclosed: a Featured coach receives at most a +0.10 tier boost (FEATURED_BOOST_MAX), Pro receives +0.05 - the tier component is weighted at 0.15, so tier lifts a coach by at most ~1.5% of the final score and can never override the unbuyable base score.
Featured search results are labeled “Featured - paid” so families can see paid placements.

Methodology v1.5 - effective May 2026

Methodology v1.5 adds five new vendor profile fields, normalizes the integrations table, and rewards vendors who publish more transparent product data.

Five new vendor profile fields: pricing_starting_usd, pricing_notes, deployment_model, free_tier, and last_price_verified_at. Each is rendered on the public profile where set.
Normalized vendor_integration table - vendors list their integrations once and they surface across every cluster the vendor appears on, instead of being re-typed per cluster.
+5 points added to the 50-point profile-completeness portion of the Index, capped at one point each across the five fields above (free tier explicitly set, deployment model disclosed, at least one pricing field present, at least one integration recorded, and a price-verification date set).
New cluster-page editorial process - every cluster roster is seeded from the Index and every reorder is logged. See editorial policy above for the full rules.
Effective date: May 2026.