Marketing Evidence Library Setup

TL;DR

Set up a marketing evidence library that connects product claims to primary sources, approval status, expiry dates, and reusable citation-ready language.

A systematic evidence library turns unsubstantiated claims into a defensible, scalable asset that accelerates approvals, reduces legal risk, and increases conversion rates by up to 34% (Gartner research).

The Problem

Most founders and marketing leaders operate with a fragmented, reactive approach to claim substantiation. A claim is written, the legal team reviews it once, and then the evidence (a PDF of a study, a customer testimonial, a third-party report) is stored in a shared drive, email thread, or Slack channel. When the same claim is reused six months later, no one knows where the original supporting data lives, whether it has expired, or whether a newer, stronger study exists. The result: claims are repeated without fresh evidence, audits fail, and regulators or competitors force costly retractions.

The underlying failure is structural. Marketing claims are not treated as governed assets with a lifecycle. Evidence is not paired to claims in a machine-readable, version-controlled system. Teams lack a common taxonomy for evidence strength (RCT vs. survey vs. anecdote), so a single testimonial is given the same weight as a double-blind study. This creates a dangerous false sense of substantiation. According to the Federal Trade Commission’s substantiation doctrine, any claim that is explicitly or implicitly made must be supported by “competent and reliable scientific evidence” at the time it is disseminated. Without a living library, companies cannot prove that they met that standard.

A second pain point is the time cost. When a new campaign is being built, marketers spend an average of 11 hours per week hunting for evidence, according to a 2023 Gartner survey. That time is wasted on non-value-adding search. Meanwhile, legal teams are bottlenecked because they must manually verify every claim’s source. The lack of a single source of truth leads to inconsistent messaging across channels, which erodes brand trust and complicates global compliance.

Core Framework

Key Principle 1: Claim-Evidence Pairing with a Strength Hierarchy

Every marketing claim — whether “clinically proven to reduce wrinkles,” “#1 rated by customers,” or “FDA-cleared” — must be explicitly paired with one or more pieces of evidence. The pairing is not a loose link; it is a formal relationship in a database. Each evidence artifact is assigned a tier based on its methodological rigor. This hierarchy creates a clear decision tree: a claim can only be made if an evidence tier of a certain minimum exists.

Example hierarchy: - Tier 1: Double-blind, randomized controlled trial (RCT) with peer-reviewed publication. - Tier 2: Non-randomized controlled study, large-scale survey (n>500), or meta-analysis. - Tier 3: Case study, cohort study, or small survey (n<500). - Tier 4: Customer testimonial, expert opinion, or internal white paper. - Tier 5: Anecdotal or unsubstantiated (cannot be used for any claim).

A claim like “reduces fine lines in 4 weeks” requires Tier 1 or Tier 2 evidence. A claim like “customers love our product” can be backed by Tier 4 (representative testimonials) but must also include a supporting Tier 3 survey. This prevents over-leveraging weak evidence.

Key Principle 2: Lifecycle Governance with Versioning and Expiry

Evidence is not static. A study’s results can be superseded, a regulatory standard can change, or a customer testimonial can become outdated (e.g., the customer’s experience may have changed). The library must treat each evidence item as a versioned record with a clear creation date, review date, and expiry date. Governance policies define who can add, edit, or retire evidence, and a timestamped audit trail is maintained.

Example: A clinical study published in 2021 might be valid for 3 years. When the library’s automated system flags it as “expiring in 60 days,” the evidence owner is notified to either renew (by confirming no newer contradictory research exists) or replace it with a more recent study. If neither action is taken, the claim that depends on that evidence is automatically suspended from use in marketing materials.

Step-by-Step Execution

1. Audit Your Current Claims and Evidence

Start by scraping every piece of marketing material: website copy, email campaigns, product pages, paid ads, social media posts, PR releases, and sales decks. List every claim that could be interpreted as a factual statement about your product. Use a tool like Screaming Frog or a simple manual extraction. For each claim, note the channel, date first used, and the person who wrote it.

Create a spreadsheet with columns: Claim ID, Claim Text, Channel, First Used Date, Last Used Date, Current Evidence Reference (if any), Evidence Location (URL, file path, email). This inventory is the baseline. Expect to find 50–200 claims for a mid-sized B2B SaaS company and 200–500 for a consumer health or supplement brand.

Output: A living inventory of all active claims. Flag any claim that has no evidence reference — these are the highest-risk items.

2. Define Evidence Tiers and Scoring Criteria

Formalize the hierarchy from the Core Framework into a documented rubric. Use a table with clear definitions, examples, and acceptable use cases.

Tier	Name	Methodology	Acceptable Claims	Sample Size Requirement	Example
1	RCT	Double-blind, randomized, placebo-controlled	Efficacy, safety, causality	n≥100 per arm	“Clinically proven to lower blood pressure”
2	Controlled Study	Single-blind, non-randomized, or large-scale survey	Correlation, user satisfaction, preference	Surveys: n≥500; Studies: n≥50	“9 out of 10 users reported fewer breakouts”
3	Observational	Cohort, case-control, small survey	Trend, user experience, behavioral	n≥30	“Our users sleep 20 minutes longer on average”
4	Testimonial	Individual customer story, expert quote	Anecdotal support, satisfaction	n=1	“I lost 15 pounds using this plan”
5	Unsubstantiated	No data, internal speculation	Never use	N/A	“The best product on the market”

Score each claim by the highest tier of evidence available. Publish this table internally and make it the single source of truth for claim strength.

3. Collect and Tag All Existing Evidence

Gather every piece of evidence you have: past clinical studies, third-party reports, customer surveys, case studies, testimonials, certification documents, regulatory filings, and even competitor comparisons. Assign each artifact a unique ID. Tag it with metadata:

Source: (e.g., “Journal of Dermatology, 2022, Vol 34”)
Sponsor: (e.g., “Internal R&D” or “Third-party lab”)
Date Conducted: (YYYY-MM-DD)
Date Published: (YYYY-MM-DD)
Methodology: (RCT, survey, etc.)
Sample Size: (number)
Population: (e.g., “adults 25–40 with mild acne”)
Key Findings: (list of supported claims)
File Location: (URL or cloud storage path)
Owner: (person responsible for this evidence)
Expiry Date: (set based on domain: e.g., 3 years for clinical, 1 year for surveys)

Use a tool like Airtable or Notion with a database view. For each claim in your inventory, link the relevant evidence IDs. This step reveals gaps: claims that have no linked evidence, or evidence that is too weak for the claim’s tier.

4. Build the Library Database

Choose a platform that supports relational data, versioning, and access control. Options:

Notion – good for small teams (<10), limited automation.
Airtable – moderate scalability, powerful linked records and automation.
SharePoint – enterprise-grade, integrates with Microsoft 365.
Dedicated platform – speciality tools like Consignor or ClaimMaster.

Design the schema:

Table: Claims - Claim ID (text) - Claim Text (long text) - Channel (multi-select) - Status (active, suspended, retired) - Minimum Required Tier (number) - Current Evidence Tier (lookup from linked Evidence) - Last Reviewed Date (date) - Owner (user)

Table: Evidence - Evidence ID (text) - Title (text) - Tier (number) - Source (text) - Methodology (text) - Sample Size (number) - Conducted Date (date) - Expiry Date (date) - File (attachment) - Owner (user) - Status (active, expired, replaced)

Link table: Claim-Evidence Mapping (many-to-many).

Example schema in YAML:

tables:
  claims:
    fields:
      claim_id: string
      claim_text: string
      channels: [string]
      status: string (active|suspended|retired)
      min_required_tier: integer
      last_reviewed: date
      owner: string
    linked_to:
      evidence: many-to-many via claim_evidence
  evidence:
    fields:
      evidence_id: string
      title: string
      tier: integer
      source: string
      methodology: string
      sample_size: integer
      conducted_date: date
      expiry_date: date
      file_url: string
      owner: string
      status: string (active|expired|replaced)
    linked_to:
      claims: many-to-many via claim_evidence
  claim_evidence:
    fields:
      claim_id: string
      evidence_id: string
      relationship: string (primary|supporting)
      created_at: timestamp

5. Implement a Review Workflow

Define roles: Evidence Owner (usually a data scientist or product manager), Reviewer (legal or compliance), Approver (head of marketing or CMO). Set up a periodic review cadence — quarterly for high-risk claims (health, safety, financial), semi-annually for lower-risk claims.

Use automation to trigger reviews:

Airtable Automations or Zapier: When a claim’s linked evidence expires, flag the claim and send an email to the owner.
Google Forms + Apps Script: Create a simple submission form for adding new evidence. On submission, a new record is created in the library and an approval request is sent to the reviewer.
Slack bot: Notify the team when a claim is about to expire or when a new evidence is added.

Maintain a change log. Every addition, edit, or deletion must timestamp the user and the action. This provides an audit trail for regulators.

6. Establish Governance Policies

Document the rules in a single-page policy document. Key policies:

Minimum Evidence Requirement: All claims must have at least one piece of Tier 2 or better evidence, unless the claim is purely subjective (e.g., “we think you’ll love it”). Subjective claims must still be flagged.
Evidence Expiry: Default expiry is 3 years from publication for clinical studies, 1 year for surveys, and 6 months for testimonials. Exceptions require written approval from legal.
Evidence Addition: Only evidence owners and above can add new records. Each addition must include a source, date, and methodology.
Claim Retirement: When a claim is no longer used, it must be marked “retired” in the library. The evidence can remain active for other claims.
Dispute Resolution: If a team member believes a claim is unsupported, they can open a dispute ticket. The evidence owner and legal must resolve within 7 days.

7. Train and Enforce

Conduct a one-hour workshop for the marketing, product, and legal teams. Show them how to:

Look up a claim and see its evidence tier.
Request the addition of new evidence.
Understand the expiry alert system.
Follow the policy for creating new claims.

Appoint a Evidence Librarian (a part-time role, often a marketing operations manager) who is responsible for the library’s health. Run a weekly 15-minute standup to review new claims added that week, expiring evidence, and blocked claims. Use a dashboard (e.g., Airtable’s interface, or a Google Data Studio report) that shows:

% of claims with Tier1 evidence vs. Tier2 vs. Tier3+.
Number of claims expiring in next 30 days.
Number of claims with no evidence linked.

Common Mistakes

❌ No evidence expiry dates. Claims become stale, and when a competitor or regulator challenges an old claim, you cannot defend it. Set expiry dates and automate alerts.
❌ Treating all evidence as equal. A single testimonial is not comparable to a peer-reviewed study. Without a tier system, weak evidence can be used to make strong claims, creating legal exposure.
❌ Storing evidence in silos. Emails, shared drives, and Slack channels fragment the library. No single person can find all evidence quickly. Centralize immediately.
❌ No version control. When you update a study, you lose the old version. If a regulator requests the original data, you cannot produce it. Use a database with version history or a VCS like Git LFS for large files.
❌ Over-engineering before the audit. Building a complex database before you know what claims you have is a waste. Start with a simple spreadsheet, then migrate to a tool once the inventory is clean.

Metrics to Track

Metric	Definition	Target	How to Measure
Claim Coverage Rate	Number of claims with Tier1 or Tier2 evidence / total active claims	>80%	Library query
Evidence Freshness	Average age (in months) of all active evidence	<12 months	Automated calculation
Audit Pass Rate	% of claims that pass a random legal review	100%	Quarterly audit
Time to Find Evidence	Average time (seconds) from claim lookup to opening the evidence file	<30 seconds	Usability test (5 users)
Claims per Evidence	Average number of claims supported by a single evidence item	<3	Library query
Expired Evidence Count	Number of evidence items past their expiry date	0	Weekly automated check

Checklist

[ ] Complete inventory of all current marketing claims across all channels.
[ ] Define evidence tier definitions and scoring rubric (documented).
[ ] Gather all existing evidence into a central repository (cloud folder or database).
[ ] Tag each evidence with metadata (source, date, sample size, methodology, tier).
[ ] Build the library database (choose tool, design schema, link claims to evidence).
[ ] Set up a workflow for adding new claims and evidence (with approval step).
[ ] Establish governance policies (minimum evidence, expiry, roles, dispute resolution).
[ ] Train marketing, product, and legal teams on library usage.
[ ] Run first audit cycle: review all claims against evidence; suspend unsupported claims.
[ ] Set up automated expiry alerts and a weekly health dashboard.

How to Implement the Evidence Library in 7 Days

Day 1: Inventory claims. Use a web scraper or manual audit to list all claims. Aim for 100% capture. Output: a spreadsheet of 50–200 claims with columns: Claim ID, Channel, First Used Date, Current Evidence (if any).

Day 2: Define tiers and gather evidence. Write the tier rubric (see Step 2). Search your files, emails, and CRM for all evidence. Add each to a folder with a standard naming convention: YYYY-MM-DD_EvidenceType_Summary.pdf. Create a second spreadsheet for evidence.

Day 3: Build the database in Airtable. Create the two tables: Claims and Evidence. Set up a linked record field between them. Import your spreadsheets. Use the Airtable interface to add missing metadata.

Day 4: Map claims to evidence. For each claim, link at least one evidence record. If no evidence exists, flag the claim as “unsupported” and set status to “suspended.” This will likely be 20–40% of your claims. Note: you must be honest about gaps.

Day 5: Set up automation and alerts. Create an Airtable automation that triggers when a claim’s linked evidence has an expiry date within 30 days. Send an email to the evidence owner. Create a simple form (Typeform or Google Forms) for submitting new claims or evidence.

Day 6: Write governance policy. Produce a one-page document (see Step 6). Share with the team via a Google Doc. Get sign-off from legal and the CMO.

Day 7: Train and launch. Hold a 60-minute session. Walk through the library interface, the policy, and the alert system. Assign a weekly standup for the first month. Run a live audit against three random claims to demonstrate the process.

Frequently Asked Questions

What is the minimum evidence needed for a marketing claim?

The FTC requires “competent and reliable scientific evidence” that is appropriate for the claim. For health or performance claims, that usually means at least a well-designed controlled study with statistically significant results (Tier 2 or above). For subjective claims like “most loved,” a robust survey (Tier 2) with proper sampling is acceptable. Always check local regulations (e.g., FDA, ASA, CAP).

How often should evidence be reviewed?

At least annually. For claims in fast-moving fields (tech, supplements, fashion), review quarterly. Use automated expiry alerts so you don’t rely on memory. The library should flag any evidence that is older than the allowed window (e.g., 3 years) and suspend the linked claim until renewal.

Can we use customer testimonials as evidence?

Yes, but only for claims of satisfaction, personal experience, or subjective opinion. They are Tier 4 evidence and cannot support efficacy or safety claims. The testimonial must be genuine, representative, and not misleadingly edited. The FTC’s Endorsement Guides require that testimonials reflect typical results, not exceptional ones.

Who should own the evidence library?

A cross-functional team: marketing (owns the claim inventory), legal (owns the policy and audit), product/R&D (owns the evidence generation), and data science (owns the tier scoring). The day-to-day manager is often a marketing operations or product manager with a “Evidence Librarian” title. Budget for 2–4 hours per week for maintenance.

What if we don’t have internal studies?

You can use third-party research, industry benchmarks, white papers, or partner studies. However, you must ensure that the evidence is directly applicable to your product. If the evidence is for a different product or population, you cannot use it without appropriate disclaimers. Create a plan to fund internal studies for the most critical claims. The library can also include “evidence gaps” as a status to prioritize future research.

How do we handle claims based on competitor comparisons?

They are high-risk. You need rigorous evidence from a head-to-head trial or independent benchmarking using the same methodology. The FTC and ASA require that any comparative claim be substantiated by testing that is fair and representative. Avoid vague claims like “better than X.” Instead, use specific, data-backed comparisons: “Users of our product reported 20% faster time-to-value than users of Product X in a controlled study (n=200).”