Customer Testimonials That Sell Without Making Shit Up

Direct answer: Customer testimonials sell when they support one specific business claim with a real customer’s relevant experience. The customer should be accurately identified or honestly anonymized, the quotation should preserve what they actually meant, any unusual result or material connection should be disclosed, and the evidence should appear beside the buyer’s doubt. A floating quote that says “Great team!” proves only that somebody found three words and a pair of quotation marks.

TL;DR: Stop collecting compliments like decorative throw pillows. Build an evidence system. Match each testimonial to a claim, choose customers whose circumstances resemble the intended buyer, document permission and source material, edit only without changing meaning, disclose incentives or atypical results, and place the testimonial at the decision it helps someone make. Scope Design calls this the CLAIM Test: Claim, Likeness, Attribution, Integrity, and Moment.

What this guide covers

• What customer testimonials actually are
• Why testimonials can help—and why the usual statistics are bullshit
• The CLAIM Test
• What a persuasive testimonial includes
• How to choose and request testimonials
• How to edit, approve, and disclose them honestly
• Where testimonials belong on a website
• Written, video, audio, and social formats
• Testimonials in AI-assisted marketing
• How to measure whether they help
• Common testimonial mistakes
• A practical publishing workflow
• Frequently asked questions

Customer testimonials are selected evidence, not independent applause

A customer testimonial is an advertising message presented by a business that buyers are likely to understand as reflecting a customer’s actual opinion, belief, or experience. It may be a written quotation, a video, a short audio clip, an attributed social post, or part of a larger customer story.

The business chooses what to publish and where to publish it. That is the important bit everybody politely steps around. A testimonial may come from a real customer and still be curated marketing. It is not automatically false, and it is not automatically independent.

The Federal Trade Commission’s Consumer Reviews and Testimonials Rule Q&A draws a useful U.S. distinction: a review is submitted to and published through a site or platform that receives and displays customer evaluations, while a testimonial is an advertising message used by the business. The context can change the classification. Reusing customer feedback in company-controlled advertising is not the same act as leaving it in its original review environment.

That distinction helps businesses stop treating every nice sentence as interchangeable proof.

Our guide to how customer reviews influence buying decisions owns the public-review side of this cluster. This guide owns what happens when a business selects customer evidence for its own marketing.

Why testimonials can help—and why the usual conversion statistics are bullshit

People evaluate claims partly through their source. A customer describing an actual experience may be more believable than a company describing its own magnificence. That is not a shocking discovery. It is also not permission to staple “testimonials increase conversions by 80%” onto every marketing article until the internet collapses from recycled statistics.

A 2013 Public Relations Review experiment compared a company spokesperson with a customer testimonial across business communication channels. The 2-by-7 experiment included 514 participants. It found partial and contextual support for customer quotations improving message credibility and attitudes toward the product. It did not find universal evidence that a customer spokesperson is always more credible, and it did not produce a magic website conversion rate. Read the study abstract and design.

The broader source-credibility literature is not a confetti cannon either. A 2026 systematic review and meta-analysis covered 40 effects from 34 papers with 7,995 participants. It found a small positive average effect of credible sources on performed behavior, with meaningful differences by context. The authors explicitly cautioned that the cost of creating credible-source interventions should be balanced against their effectiveness. See the open research summary and article.

The practical conclusion is more useful than a suspiciously precise percentage:

• credible customer evidence can strengthen a relevant claim;
• the source, audience, channel, context, and claim all matter;
• a testimonial cannot rescue an irrelevant offer, confusing message, broken form, or nonexistent demand;
• evidence should reduce a specific uncertainty, not merely decorate a layout.

That fits Scope Design’s larger position on brand authority: authority comes from specific claims, shown work, named constraints, and a willingness to state what the evidence does not establish. Looking established may buy the first meeting. Useful proof helps a buyer survive the cheaper comparison.

The CLAIM Test for customer testimonials

Before publishing a testimonial, run it through five questions.

C — Claim: What exact statement does this evidence support?

Start with the business claim, not the compliment.

“They were wonderful” has no clear evidentiary job. “The team explained the migration plan before touching the live system” supports a claim about process clarity and risk management. “The new site is easier for our staff to update” supports maintainability. “They answered the phone” may be valuable if response anxiety is the buyer’s actual objection.

Write the claim in plain language. Then ask whether the customer’s words genuinely support it. If the quote and the claim do not match, do not force them into an arranged marriage.

L — Likeness: Can the intended buyer recognize relevant circumstances?

Likeness does not mean demographic cosplay. It means the customer story contains the context a buyer needs to judge relevance:

• business or use-case type;
• role in the decision or implementation;
• problem being solved;
• meaningful constraint;
• stage, scale, or technical environment when it affects the result.

A testimonial from a nationally recognized company may look impressive and still answer nothing for a local owner wondering whether the process will overwhelm a three-person team. Relevance beats logo worship.

A — Attribution: Is the source real, accurate, and authorized?

Record who said it, when, where the original statement lives, and what identifying details may be used. If the customer permits a full name, role, company, and photo, those details make the evidence easier to evaluate. If anonymity is necessary, use a truthful description rather than a theatrical one.

“Operations director at a regional manufacturer” can be legitimate. “Fortune 100 executive” is not acceptable if the person was an intern at a vendor that once sold the company paper clips.

Attribution also means disclosing a relationship a reasonable buyer would not expect: employee, relative, affiliate, reseller, investor, paid creator, free-product recipient, or another material connection.

I — Integrity: Did the published version preserve the experience and its limits?

Editing for spelling, length, or clarity can be reasonable. Editing “the kickoff was organized” into “Scope Design transformed our revenue engine” is fiction with administrative overhead.

Integrity requires:

• preserving the customer’s intended meaning and tone;
• retaining qualifications that affect interpretation;
• verifying any number before publication;
• explaining unusual results or limited applicability;
• disclosing relevant incentives or relationships;
• keeping the approved version and original source connected.

M — Moment: Is the evidence where the doubt occurs?

Put proof beside the claim it supports.

A testimonial about smooth implementation belongs near the implementation process. Evidence about staff editing belongs near maintainability. A relevant customer quote about pricing clarity belongs near cost context. A testimonial page hidden in the navigation is a museum: technically full of artifacts and practically absent from the decision.

What a good customer testimonial actually contains

A useful testimonial does not need to be long. It needs enough information to be judged.

The strongest structure is usually:

1. Situation: What was happening before the work or purchase?
2. Decision: Why did the customer choose this approach or provider?
3. Experience: What specifically happened during the work?
4. Outcome: What changed, was completed, or became possible?
5. Constraint: What conditions or limits affect how the result should be interpreted?

Not every quote needs all five. A short testimonial may support one narrow claim. A full customer story can carry the larger sequence.

Illustrative testimonial rewrites

The examples below are writing demonstrations, not Scope Design customer claims.

Vague praise: “Amazing company. Highly recommended.”

Useful version: “We had three teams editing the same product information in different places. The project gave us one approved workflow and made ownership clear before new software was added.”

The useful version supports process diagnosis. It gives a buyer a recognizable problem and a concrete change without inventing a percentage.

Vague praise: “They built us a beautiful website.”

Useful version: “Our staff needed to update ordinary pages without calling a developer. The new page system gave us reusable sections and a documented publishing process.”

This supports maintainability, not aesthetics, sales, or search rankings.

Vague praise: “Best investment ever.”

Useful version: “Discovery showed that the requested dashboard would duplicate a report our existing platform already produced. We used the native report and spent the project budget on the broken intake process instead.”

This supports diagnostic honesty and solution restraint. It does not claim a quantified return unless one was measured.

Notice what the stronger examples do not contain: imaginary names, celebrity headshots, suspicious revenue figures, or the phrase “game changer.” Civilization continues.

How to choose testimonials for a website

Do not begin by sorting quotes from nicest to nicest. Begin with the claims and objections on the page.

Create a simple evidence inventory with these fields:

This prevents a common error: trying to solve every trust problem with a testimonial. Some claims need a demonstration. Some need a policy. Some need a case study with measured data. Some need the business to stop making the claim.

Select evidence using four priorities:

• relevance before fame;
• specificity before enthusiasm;
• traceability before polish;
• coverage before volume.

Coverage means your evidence addresses the important doubts across the buying journey. Twenty quotes about friendliness do not compensate for silence about implementation, cost, ownership, accessibility, maintenance, or results.

How to ask for a testimonial without writing it for the customer

Ask after a credible moment of value, while the customer can still describe the situation clearly. That moment might be a successful launch, an avoided error, a completed migration, the first independent use of a new system, a resolved service problem, or a documented business outcome.

Do not ask, “Can you say something nice about us?” That produces nice-sounding mush. Ask questions that recover evidence:

• What was happening before this work?
• What were you worried about?
• What did we recommend or do that was most useful?
• What changed after the work?
• What surprised you?
• Who would this approach fit, and who might need something different?
• May we follow up to verify details and request publication permission?

The customer can answer in writing, on a recorded call, or through a short interview. The format matters less than preserving the original response and approving the published version.

Public review requests have different platform and policy considerations. Our ethical Google review generation guide covers timing, scripts, review gating, platform rules, and the Scope Design FAIR Review Loop. Do not turn a testimonial request into a disguised demand for a five-star public review.

Can a business edit a customer testimonial?

Yes, a business can edit a testimonial for length, grammar, clarity, or format if the result still accurately reflects the customer’s actual opinion and experience. The FTC’s Endorsement Guides state that an endorsement need not use the endorser’s exact words unless the advertisement says it does, but the message cannot be presented out of context or reworded to distort the opinion or experience.

Use an editorial chain of custody:

1. Preserve the original email, recording, form response, review, or interview transcript.
2. Mark deletions and changes in a working copy.
3. Verify names, roles, companies, dates, products, and any factual result.
4. Send the exact edited version and intended attribution to the customer.
5. Record approval, permitted channels, and any time limit.
6. Store the final approved asset with the original source and disclosure notes.
7. Recheck it when the offer, product, relationship, or customer’s role materially changes.

Safer edits

• correcting obvious spelling and punctuation;
• removing filler or repetition;
• shortening without removing a material condition;
• replacing a confusing pronoun with the accurate company or product name;
• dividing a longer statement into clearly faithful excerpts;
• adding brackets or ellipses sparingly when they help readers understand an edit.

Edits that cross the line

• adding an outcome the customer did not report;
• changing “helped” to “transformed”;
• removing “for our small team” when scale affects relevance;
• combining statements from different customers into one synthetic person;
• converting an employee’s praise into an ordinary-customer endorsement;
• turning an unmeasured impression into a percentage;
• attaching a stock portrait or AI-generated person to a real quotation;
• keeping a testimonial live after learning it no longer reflects the customer’s view.

If your edit needs a defense memo, the edit probably sucks.

Permission is part of the evidence

FTC guidance focuses on truthfulness and deception, not on giving businesses a universal release-form template. Other rights, contracts, privacy laws, industry rules, and platform terms can matter. Get legal advice for your jurisdiction and risk profile.

Operationally, written permission gives a business something essential: proof that the source authorized the represented use. A permission record should cover, as applicable:

• the approved words or recording;
• editing for length and clarity;
• the customer’s name and preferred attribution;
• job title and company name;
• photograph, voice, likeness, and video;
• website, email, social media, proposals, presentations, paid advertising, and other channels;
• whether the business may create shorter excerpts;
• incentives or material relationships;
• an expiration or review date, if appropriate;
• a practical contact and removal process.

Public praise is not a bottomless asset library. A LinkedIn comment, Google review, private email, project chat, and recorded interview arrive in different contexts. Before moving content into company-controlled advertising, verify the original platform terms, accuracy, attribution, and permission.

Scope Design applies the same standard to its own work: no named-client performance claim goes public without documented evidence and appropriate permission. When a story can be used only in anonymized form, the description must remain truthful and omit identifying details rather than adding theatrical ones.

Incentives, insiders, and atypical results

The United States FTC requires endorsements to reflect honest experience and prohibits claims that would be deceptive or unsubstantiated if the advertiser made them directly. It also requires clear disclosure of material connections that a reasonable audience would not expect and that could affect the weight or credibility of the endorsement.

Material connections can include:

• payment;
• free products or services;
• discounts, credits, upgrades, or sweepstakes entries;
• affiliate commissions;
• employment or family relationships;
• reseller, investor, consultant, or partner relationships;
• another benefit tied to providing the endorsement.

Put the disclosure near the testimonial in language a normal person can understand. Do not bury “paid relationship” in a hover state, legal page, pale-gray footnote, or sixteenth accordion called “transparency.”

The FTC’s small-business advertising FAQ also explains that an endorsement reflecting an exceptional result may mislead if the advertising does not clearly state what consumers generally can expect or the limited applicability of the endorser’s experience. “Results may vary” and “not typical” are not magic erasers.

Ask four questions before publishing a result:

1. Did the customer actually experience and report it?
2. Can the underlying number or event be verified?
3. What would a reasonable buyer infer about expected results?
4. What context, typical-result information, or limitation must appear with it?

Regulated health, financial, legal, employment, and other sensitive claims can carry additional rules. This article is a marketing operations framework, not individualized legal advice.

Where testimonials should go on a website

Put each testimonial immediately before, beside, or after the claim or objection it helps resolve. The correct location depends on the evidence job.

Homepage testimonials

Use a small number of broad, high-relevance testimonials that support the homepage’s main claims. Do not install an unreadable auto-rotating carousel and call the trust problem solved. If evidence changes before a visitor can finish reading it, the interface is fighting the content.

Service and product pages

Use evidence specific to that service, product, problem, and buyer. A general compliment about the company is weaker than a quotation about the exact uncertainty on the page.

Pricing and comparison pages

Use customer evidence to clarify value, fit, switching, implementation, or the cost of a problem—not to pressure someone with an outlier result. Pair every quantified claim with its context and substantiation.

Forms and calls to action

Evidence near a form should address the last remaining risk: what happens next, whether the conversation is useful, how information is handled, or whether the process respects the buyer’s time. “Join 50,000 happy customers” is not reassuring if the visitor is worried about a four-day response time.

Dedicated testimonial pages

A testimonial page can be a useful evidence library for sales teams and committed researchers. It should not be the only place testimonials exist. Organize it by buyer question, service, outcome, use case, or industry—not as a random wall of praise. Link each item to deeper context when available.

Written, video, audio, and screenshot testimonials

Choose the format that preserves the useful evidence and fits how buyers evaluate it.

Written testimonials are fast to scan, easy to place near a claim, and easy to keep accessible. They become weak when over-edited, unattributed, or presented as identical cards with no context.

Video testimonials can preserve voice, emotion, hesitation, and a fuller explanation. They are also expensive, slower to consume, harder to update, and not inherently more truthful. A polished production can make an honest customer look scripted. Keep the customer’s language and conditions intact.

Audio testimonials can work when voice is meaningful and video adds little. Provide a transcript for access and scanning.

Screenshots of messages or social posts can preserve original context, but they may expose personal information, become unreadable on phones, and create permission or platform-term questions. Transcribe the relevant text accessibly, retain the source, and obtain appropriate authorization.

For prerecorded synchronized video, WCAG 2.2 guidance on captions requires synchronized text for the audio information at Level A. The W3C’s transcript guidance explains when basic or descriptive transcripts support additional access needs. Captions and transcripts also make the evidence searchable and reusable without forcing everyone to watch a video.

There is no universal ideal length. Use the shortest version that preserves the claim, context, and limits. That may be one sentence beside a feature, a 45-second clip near a process explanation, or a full case study for a complex purchase.

Can AI write or edit customer testimonials?

AI can help organize source material. It cannot manufacture the customer’s experience.

Reasonable uses include:

• transcribing a recorded interview;
• finding repeated themes across approved responses;
• suggesting shorter versions that a human verifies;
• tagging evidence by claim, buyer doubt, service, or format;
• flagging numbers, names, and material connections for review;
• generating captions, subject to human correction;
• comparing the approved quotation with the published version.

Unacceptable uses include:

• inventing a customer;
• generating praise and asking a customer to rubber-stamp it;
• attaching an AI face to a quotation;
• merging several customers into a fake composite without clear disclosure;
• fabricating emotion, outcomes, or precise metrics;
• changing the meaning because the original did not sound exciting enough.

Use AI for assembly and checking. Keep human approval for publication, attribution, legal review, and any judgment about whether the result remains truthful. Fake proof is not automation. It is lying faster.

How to measure testimonial performance

Do not judge testimonial strategy by quote count. A swollen asset folder is not a business outcome.

Measure the last useful behavior the evidence can reasonably influence:

• clicks from an evidence block to a relevant case study or process explanation;
• qualified form starts and completions on pages where evidence was added;
• progression from service or comparison content to a high-intent next step;
• sales use of testimonial assets and the objections they help answer;
• lead quality, held conversations, proposal progression, or sales-cycle movement where attribution is credible;
• evidence coverage across priority claims and objections;
• percentage of published assets with documented source, permission, disclosure, and review status.

A/B testing can help on high-traffic pages, but low-volume service businesses should not spend eighteen months testing whether Sarah’s quote belongs two inches higher. Start with claim coverage, sales feedback, usability observation, and qualified behavior. Our guide to web design versus content explains why diagnosing the actual constraint matters more than polishing the most visible thing.

Testimonial evidence should also feed operations. If several customers independently praise a process step, the business may have a differentiator worth documenting. If the strongest available evidence is about friendliness while sales promises technical expertise, the evidence gap deserves attention. If customers repeatedly mention a problem, the answer is not to hide those comments and buy a happier widget. The online reputation management guide owns the correction loop.

Common testimonial mistakes

• Publishing vague praise. “Great service” consumes space without reducing uncertainty.
• Hiding every quote on one page. Proof separated from the claim arrives too late.
• Using fame instead of relevance. A recognizable logo does not answer a buyer’s actual question.
• Inventing specificity. A fake percentage is not stronger than an honest qualitative result.
• Losing the source. If nobody can recover the original statement or approval, the asset is weak and risky.
• Over-editing. Removing qualifications can change the testimonial’s net impression.
• Ignoring permission. Publicly visible does not mean reusable everywhere forever.
• Forgetting disclosures. An employee, affiliate, free product, or unusual result can change how the evidence should be understood.
• Using dark-pattern proof. Fake purchase popups, fabricated counters, and stock-person testimonials are manipulation, not evidence. The ethical marketing psychology guide covers the Scope Design CLEAN Test.
• Treating video as automatically superior. Format cannot compensate for weak substance.
• Measuring raw conversions only. More low-quality submissions can create a reporting win and an operational loss.
• Never retiring old evidence. Roles, products, opinions, permissions, and expected outcomes can change.

The Scope Design testimonial workflow

1. Map claims and doubts. List what each priority page asks a buyer to believe and why they may hesitate.
2. Choose the right evidence type. Do not demand a testimonial when a demonstration, policy, case study, public review pattern, or measured result would be stronger.
3. Identify relevant customers. Choose people with actual experience that matches the claim and audience.
4. Ask evidence questions. Recover situation, decision, experience, outcome, and constraint in the customer’s language.
5. Preserve the source. Store the original response, recording, or message with date and context.
6. Verify factual details. Check names, roles, company, scope, timeframe, and every number.
7. Edit faithfully. Shorten or clarify without strengthening the claim or removing material conditions.
8. Approve and authorize. Confirm the exact wording, attribution, media, channels, and disclosures in writing.
9. Apply the CLAIM Test. Reject or revise any asset that fails Claim, Likeness, Attribution, Integrity, or Moment.
10. Publish near the doubt. Put the evidence where a buyer encounters the relevant uncertainty.
11. Measure qualified behavior. Watch coverage, sales usefulness, and business outcomes rather than counting quotes.
12. Review and retire. Recheck high-use evidence when facts, relationships, products, or permissions change.

This is slower than asking an AI tool for twelve ecstatic testimonials from fictional CEOs. It is also how a business builds a body of proof instead of a content costume.

Frequently asked questions about customer testimonials

What is a customer testimonial?

A customer testimonial is an advertising message presented by a business that buyers are likely to understand as reflecting a customer’s real opinion, belief, or experience. It can be written, recorded, filmed, or republished from another source with appropriate accuracy, attribution, permission, and disclosure.

What is an example of a good testimonial?

A good testimonial identifies a recognizable situation, describes what specifically happened, and supports a limited claim. For example: “Our staff needed to update ordinary pages without calling a developer. The reusable page system and publishing guide let the communications team handle those updates directly.” This is an illustrative example, not a claimed customer quotation.

What makes a testimonial credible?

Credibility comes from a real and relevant source, specific context, accurate attribution, preserved meaning, verified facts, honest limits, and disclosure of incentives or relationships that could affect how buyers evaluate the endorsement. Polish without traceability is just shinier uncertainty.

What is the best format for a testimonial?

The best format is the one that preserves useful evidence and fits the buyer’s decision. A short written quote works well beside a narrow claim. Video can preserve emotion and explanation. A case study is stronger for complex context and measured outcomes. Format should follow the evidence job.

How long should a testimonial be?

Use the shortest version that retains the relevant situation, experience, outcome, and limits. There is no universal ideal length. One or two sentences may support a page claim; a complex purchase may need a longer video, transcript, or case study.

How should a testimonial start?

Start with the customer’s situation or concern, not with praise for the company. “We had three teams updating the same information in different places” gives buyers a recognizable problem. “Scope Design is amazing” gives them an adjective and no reason to believe it.

Where should testimonials go on a website?

Place each testimonial near the claim or doubt it addresses: fit evidence near service context, implementation evidence near the process, maintainability evidence near editing or ownership, and verified outcome evidence near cost or decision-stage content. A dedicated library can supplement this placement but should not replace it.

How do I put testimonials on my website?

Map the page’s claims, select a relevant approved testimonial for each important doubt, include truthful attribution and necessary disclosure, format it accessibly, and connect it to deeper evidence when available. Avoid auto-rotating carousels that disappear before people can read them.

Is a testimonial the same as a customer review?

No. A public review is generally a customer’s evaluation submitted to a platform or system that displays customer evaluations. A testimonial is an endorsement selected and presented in business-controlled advertising. Reusing a public review on a company website can change its context and obligations.

What is the difference between a testimonial and a case study?

A testimonial is an attributed customer endorsement, often short and focused on one experience or claim. A case study is a structured account of the situation, work, constraints, and outcome, usually assembled by the business and verified with the customer. Case studies can carry more context but still require evidence and permission.

Can a business edit a customer testimonial?

Yes, editing for grammar, length, clarity, or format can be acceptable when it does not distort the customer’s opinion or experience. Preserve the original, verify facts, send the exact edited version for approval, and never add outcomes or remove conditions that change the meaning.

Do I need permission to publish a testimonial?

Written permission is the safest operational standard, especially when using a person’s name, image, voice, employer, or statement in new advertising contexts. FTC guidance addresses truth and disclosure, while contracts, privacy, publicity rights, platform terms, and industry rules may add requirements. Get legal advice for the applicable jurisdiction.

How should testimonial incentives be disclosed?

Disclose a material connection clearly, conspicuously, and near the testimonial in plain language—for example, that the customer received a free product, discount, payment, affiliate commission, or another benefit. Do not condition incentives on positive sentiment or hide the disclosure in distant fine print.

Are video testimonials better than written testimonials?

Not automatically. Video can preserve voice, emotion, and a fuller explanation, while written testimonials are faster to scan, place, update, and make accessible. Substance, relevance, permission, and placement matter more than format. Prerecorded synchronized video should include accurate captions.

Can AI write customer testimonials?

AI can transcribe, organize, tag, and suggest faithful edits to real customer source material. It should not invent customers, praise, outcomes, faces, or composite experiences presented as real. A human should verify the source, facts, permission, disclosure, and final wording before publication.

How many testimonials should a website have?

Enough to cover the important claims and objections without repetition. A page may need two highly relevant testimonials rather than twelve generic ones. Measure evidence coverage and usefulness, not the size of the testimonial pile.

How do you measure whether testimonials work?

Track qualified behavior near the evidence: relevant case-study clicks, form starts and completions, progression to high-intent pages, sales use, objection resolution, lead quality, and sales-cycle movement where attribution is credible. Also audit whether every published asset has a traceable source, approval, disclosure, and review status.

Build proof buyers can inspect

The goal is not to make every customer sound ecstatic. The goal is to help the next buyer judge a claim with less uncertainty.

Map the doubt. Find the relevant experience. Preserve the source. Verify the facts. Obtain permission. Edit without strengthening. Disclose what changes interpretation. Put the evidence beside the decision. Then measure whether better-informed buyers take a useful next step.

That is customer testimonial strategy. Everything else is a quote carousel auditioning for a role in a website nobody trusts.

Sources

• Federal Trade Commission: Consumer Reviews and Testimonials Rule Questions and Answers
• Federal Trade Commission: Endorsement Guides — What People Are Asking
• Federal Trade Commission: Advertising FAQs for Small Business
• Howes and Sallot: Company spokesperson vs. customer testimonial
• Hamer et al.: How Effective Are Credible Sources in Changing Behavior?
• W3C WAI: Understanding Captions for Prerecorded Media
• W3C WAI: Transcripts