Skip to main content
GrowthLayer buyer tool · no email gate

Compare CRO agency proposals on evidence—not sales polish.

Score three providers against the same 15 weighted questions. Missing evidence stays visible, hard stops override points, and your work remains in this browser.

Written by Atticus Li · Evidence reviewed under the GrowthLayer editorial policy · Updated July 22, 2026

01

Before you score

Make every provider answer the same decision.

Send the questions with your RFP or use them during proposal review. Score only what the provider documents; do not fill evidence gaps with assumptions from the sales call.

  1. 01

    Define the outcome

    Name the journey, business decision, primary metric, guardrails, deadline, and scope exclusions.

  2. 02

    Request comparable evidence

    Ask every provider for the same source links, assumptions, owners, costs, and commercial terms.

  3. 03

    Score before negotiating

    Leave unknowns blank. Resolve hard stops first, then turn remaining gaps into contract conditions.

02

Interactive worksheet

Score three CRO agency proposals side by side.

The normalized score reflects answered criteria. Evidence coverage shows whether enough of the weighted scorecard has been answered to make that number trustworthy.

Loading saved worksheet… No email or account is required.

Blank CSV

Proposal status

Scores are directional buying support, not a provider certification.

Evidence coverage

0%

Normalized score

—

Not comparable yet

Rate enough source-backed criteria to reach 80% evidence coverage.

Evidence coverage

0%

Normalized score

—

Not comparable yet

Rate enough source-backed criteria to reach 80% evidence coverage.

Evidence coverage

0%

Normalized score

—

Not comparable yet

Rate enough source-backed criteria to reach 80% evidence coverage.

Rating scale: what 0–4 means
  1. 0Absent, contradicted, or unacceptable
  2. 1Generic promise with no evidence or owner
  3. 2Partially specified; important dependencies remain
  4. 3Specific, evidenced, and mostly complete
  5. 4Source-linked, measurable, owned, and contract-ready

Evidence-weighted comparison

A criterion counts toward coverage only when it has both a rating and a nonblank evidence note. Leave the rating blank when the proposal does not let you judge it. A zero means you reviewed the criterion and found it absent, contradicted, or unacceptable.

Criterion and evidence requestProvider AProvider BProvider C
Fit7% weight

Decision and scope fit

What business decision and customer journey will this engagement make easier?

Ask for: Named decision, journey boundary, primary outcome, exclusions, and acceptance criteria.

Fit6% weight

Business-model and journey fit

What directly comparable experience do you have with this model, audience, and buying motion?

Ask for: Named comparable clients, relevant journey, constraints, and important differences.

Measurement8% weight

Measurement integrity

How will you validate events, the baseline, primary metric, guardrails, and attribution?

Ask for: Tracking audit, metric definitions, QA owners, SRM monitoring, and incident process.

Research6% weight

Research plan

Which methods, samples, and evidence sources will be used, and why?

Ask for: Named methods, sample expectations, synthesis, limitations, and decision supported.

Feasibility10% weight

Test and traffic feasibility

Show the baseline, MDE, allocation, runtime, and low-volume alternative.

Ask for: Traffic and conversion inputs, MDE, duration, stopping rule, and fallback validation method.

Evidence6% weight

Comparable public proof

Provide a named comparable case with its outcome, baseline, timeframe, and source.

Ask for: Named client, comparable context, baseline, effect, timeframe, and source URL.

Evidence8% weight

Proof methods and limitations

Explain the design, attribution, confounders, and what the case cannot establish.

Ask for: Method, attribution boundary, material concurrent changes, and explicit limitations.

People5% weight

Named team and seniority

Name the roles, seniority, availability, and subcontractors assigned to the work.

Ask for: Named team, relevant experience, allocation, substitution rules, and subcontractor disclosure.

Operations9% weight

Delivery ownership

Who owns research, copy, design, development, QA, launch, analysis, and implementation?

Ask for: RACI or equivalent ownership table for every delivery stage.

Operations4% weight

Client-side load

State the access, people, hours, approvals, and turnaround times required from us.

Ask for: Weekly client hours, named roles, access requirements, approval SLA, and dependencies.

Operations4% weight

Insight timing and cadence

When is the first decision-ready insight, and what supports the proposed cadence?

Ask for: Research milestone, first decision date, test cadence assumptions, and delay consequences.

Commercials10% weight

All-in cost and terms

Include fees, tools, incentives, change orders, term, cancellation, and performance-fee rules.

Ask for: Itemized quote, assumptions, pass-through costs, change control, renewal, and exit terms.

Security & privacy8% weight

Security, privacy, and deployment controls

How will you protect our data, restrict access, manage credentials and subprocessors, control deployments, delete retained data, and notify us of incidents?

Ask for: Contract-ready DPA and subprocessor list; least-privilege access and credential controls; retention and deletion schedule; buyer deployment approval and rollback plan; incident-notification timing and owner.

Durability5% weight

Knowledge transfer and ownership

What data, research, code, designs, records, and decision logs are returned to us?

Ask for: Ownership terms, repository format, continuous access, handoff schedule, and retention policy.

Risk4% weight

No-lift and inconclusive handling

What happens when evidence is inconclusive, implementation slips, or no lift appears?

Ask for: Decision rule, learning deliverable, rework boundary, schedule impact, and commercial consequence.

Hard-stop review

A checked hard stop overrides the point score. Resolve it in writing before legal, procurement, or a pilot.

Provider A
Provider B
Provider C

How to use the result

Shortlist only after a proposal reaches at least 80% weighted evidence coverage. Put every unresolved dependency, owner, metric, and exit term into the final statement of work. Then verify named references directly; GrowthLayer does not certify the provider or guarantee an outcome.

03

Decision rule

Coverage comes before ranking.

An agency should not win because unanswered, difficult criteria disappeared from the denominator without being noticed.

Under 80% coverage
Not comparable. Ask for the missing evidence before using the normalized score.
70–84 score
Advance only with written conditions, provided evidence coverage is at least 80%.
85+ score
A strong shortlist signal only when coverage is at least 90% and no hard stop is present.
Any hard stop
Do not advance on points. Resolve the measurement, ownership, feasibility, security, privacy, or commercial risk first.

04

Continue the decision

Set the operating model and budget before choosing a logo.

Pricing and proof transparency indexCheck which public pricing and proof claims have a current source trail.CRO agency pricingCompare the all-in scope, dependencies, and commercial model.Agency vs alternativesDecide whether an agency is the right operating model first.CRO audit guideDefine a bounded diagnostic and objective acceptance criteria.CRO agency directoryBrowse evidence-reviewed providers after your requirements are explicit.