How we rank CRO agencies

Every agency is scored on public evidence with the same seven weighted criteria. The calculation for every agency is published on its profile.

Four numbers we keep separate

Editorial score

A number out of 10 calculated from the weighted criteria below. It is our assessment of public evidence: not a customer rating, and never shown as stars.

Overall rank

An agency's position in the full index. Agencies are sorted by editorial score, ties broken alphabetically. The editors can fix a position with a disclosed editorial placement, labelled wherever it appears (see editorial placements). Demonstration records never receive a rank.

Position in a curated ranking

Pages such as best CRO agencies for SaaS order agencies by fit for one brief. That position can differ from the overall rank. The agency's editorial score does not change.

Third-party review rating

A rating from an external review platform, shown only with the platform, review count and date checked, and labelled with its verification status. We never invent or estimate review counts, and review ratings do not feed the editorial score.

Agency scoring criteria

The editorial score runs from 4.9 to 9.9. It is calculated in four steps:

  1. Each criterion below is scored 0 to 10 from public evidence. Criteria with no evidence are "Unknown" and score 0.
  2. Each criterion is graded against the strongest evidence in the index: vs best = score ÷ best score on that criterion × 10. The agency with the best evidence on a criterion sets the 10/10 anchor.
  3. Evidence index E = Σ(vs best × weight) ÷ 100, from 0 to 10.
  4. Editorial score = 4.9 + E × 0.5, rounded to one decimal and capped at 9.9. The 4.9 baseline is earned by passing our inclusion checks: a live website and a real CRO or experimentation service.

Profiles built from an automated evidence scan are labelled as pending editorial review, and each of their criteria is capped at 8.0 until an editor checks the evidence.

Agency criteria and weights
CriterionWeightWhat countsWhat does not count
Published proof of work
Case studies and named clients with specific, attributable outcomes on a public page.
25%
  • Case studies that name the client (or a clearly described anonymised client), the change tested and a measured outcome
  • Client names published by the agency or a bylined third party
  • "Fortune 500 brands" claims with no linked work
  • Aggregate revenue claims with no case-level detail
Experimentation rigor
How clearly the agency describes research, prioritisation, statistics, QA and decision-making.
20%
  • A published research and prioritisation framework
  • Stated approach to sample size, test duration, QA and result analysis
  • Free calculators, benchmarks or methodology content that show working
  • "Data-driven" claims with no described method
  • Win rates without a definition of a win
CRO specialization depth
How central conversion optimization is to the business and how complete the in-house CRO team is.
15%
  • CRO or experimentation as the primary published service
  • In-house strategy, research, design, development, analytics and QA roles
  • CRO listed as one item in a long service menu with no supporting detail
Commercial transparency
Whether buyers can see pricing, engagement models, ownership and leadership before a sales call.
15%
  • Published prices or price floors
  • Published engagement models, terms or guarantees with stated conditions
  • Stated ownership and named leadership
  • "Contact us for pricing" with no model described
AI readiness
Published, governed use of AI in research, ideation, test building or analysis, and readiness for AI search.
10%
  • A described AI workflow with human accountability for decisions
  • Stated data-governance terms for AI use
  • Answer engine / AI search optimization offered or practised (for example machine-readable FAQs)
  • "AI-powered" in a headline with no described use
Team and footprint
Published team structure, seniority, headcount, offices and years operating.
10%
  • Named roles or team structure
  • Published headcount or office locations
  • Years operating
  • A "global" claim backed by a single address
Third-party credentials
Platform partnerships, independent reviews and awards that can be checked.
5%
  • Profiles on independent review platforms with dated reviews
  • Named awards with a year
  • Tool partner listings
  • Unlinked badges

Worked example: Invesp

The table every profile shows. This is the calculation for Invesp:

Invesp editorial score calculation
CriterionWeightScore /10vs bestPointsEvidence note
Inclusion baseline4.90Earned by passing inclusion checks: live website and a real CRO or experimentation service.
Published proof of work 25% 7.5 8.3 1.04 Large published test and client counts and a case studies section. Named clients not checked in this cycle.
Experimentation rigor 20% 9.0 9.5 0.95 Published guides, a platform-backed ideation process and senior QA and analytics roles.
CRO specialization depth 15% 10.0 10.0 0.75 CRO only for 20 years.
Commercial transparency 15% 8.0 8.4 0.63 Publishes a no-lift-no-fee model and its guarantee. Fee levels not published.
AI readiness 10% 8.0 8.9 0.44 Proprietary pattern-mining platform used in ideation.
Team and footprint 10% 8.0 8.9 0.44 Since 2006 with named senior roles in each pod.
Third-party credentials 5% 6.0 8.0 0.20 Listed on several third-party CRO rankings.
Editorial score100%9.44.9 + 4.46 evidence points, rounded, max 9.9. "vs best" grades each criterion against the strongest evidence in the index.

Score labels

  • Exceptional evidence: 9.0 and above
  • Strong evidence: 8.0 and above
  • Good evidence: 7.0 and above
  • Limited public evidence: 0.0 and above

Specialization

We classify each agency so buyers can choose between pure specialists and broader partners. Classification does not affect the score directly; the specialization criterion does.

  • Dedicated CRO / experimentation: Conversion optimization and experimentation are the whole offer.
  • CRO-led with adjacent services: CRO is the core offer, supported by services such as design, development, SEO or analytics.
  • Multi-service agency with a CRO practice: CRO is one service among others such as paid media, demand generation or web builds.

AI readiness levels

  • AI built into delivery: Publishes how AI is used across research, ideation or test building, with human ownership of decisions and stated data governance.
  • AI used in parts of delivery: Publishes a proprietary AI or predictive engine, or a named AI service, used in parts of delivery.
  • AI mentioned, little detail: Mentions AI on its site but publishes little detail on how it is used or governed.
  • No public AI approach: We found no public description of AI use. That does not prove the agency does not use AI.

How evidence is verified

Every fact on a profile carries a status and, where possible, a claim-level source link:

  • Verified: Checked by our research desk against the linked public source on the research date.
  • Agency-reported: Published by the agency on its own website. We confirmed the claim is published, not that it is accurate.
  • Third-party reported: Reported by a third-party publication or directory. Not yet confirmed with a primary source.
  • Provided by agency: Supplied by the agency or its clients to our research desk. Awaiting a public source.
  • Not yet verified: Collected during research but not yet confirmed with a source.

We use third-party "best agency" lists only to discover agencies. Many are published by agencies that rank themselves; we flag those in source lists and never treat them as evidence of quality.

Tool scoring criteria

Software is scored separately, and tool scores are not comparable with agency scores.

  • Testing capability (30%): Client-side, server-side and feature-flag testing; targeting; personalization.
  • Statistics and analysis (20%): Published stats engine, variance reduction, guardrails and reporting.
  • AI features (15%): Documented AI features such as hypothesis generation, variant building, analysis or MCP access.
  • Privacy and governance (15%): Published compliance posture (GDPR, HIPAA, SOC 2), consent support and data residency.
  • Pricing transparency (10%): Published plans, free tiers or open-source options.
  • Ecosystem and integrations (10%): Analytics, CDP and commerce integrations; SDK coverage.

Editorial placements and independence

  • Listing is free. Positions are not sold. We do not earn affiliate commissions on agency retainers.
  • Curated ranking placement rules. Positions inside curated rankings are ordered by editorial score. Editorial rules can adjust one agency's position within a curated list (never its score or overall rank). The publisher of this site is a client of Wavesy. Wavesy did not pay for inclusion or placement. Its editorial score is calculated with the same published method as every other agency; its position inside some curated rankings is set by editorial rules described on our methodology page.
  • If paid placements are introduced in future, they will be labelled Sponsored, kept separate from editorial scores, and announced on this page.

Common questions

Can agencies pay to improve their score?

No. Scores come only from public evidence assessed with this method. Listing is free.

What happens when evidence is missing?

The criterion is marked Unknown and contributes 0 points. We say so on the profile. Missing public evidence does not prove a capability is absent, and agencies can send us sources.

How often are scores updated?

Every agency has a research date and a next review date. The whole index is reviewed each quarter, and individual profiles are updated when new evidence arrives.