How the Storefront Evidence Score works

Seven areas, led by the brand itself

Value proposition, point of view and the offer carry 47% of the score, because a clear promise and a strong offer move revenue more than any app. Tech stack is a supporting score.

What each category asks

• Value Prop & Positioning (20%). Can a first-time visitor tell what you sell, who it is for and what it does for them, within a few seconds?
• Point of View (11%). How strongly the brand stands for something only it can say: explicit differentiation versus the alternatives and a range of distinct hooks, each backed by a quote from its own site.
• Third-Party Validation (14%). What other people say, shown where shoppers decide: review volume and rating, a star rating on the product page, press, awards and expert authority, certifications or clinical backing, and creator, affiliate or customer content. Only what is visible on the site or detected in its public code counts; we never estimate review counts or follower numbers. Ratings and counts that load after the page source (Bazaarvoice, Yotpo, Okendo and similar widgets) are checked on a rendered capture of the product page, the way a shopper sees it. A star rating earns full credit from 25 reviews up; below that, the review items earn a share in proportion to the count.
• Offer & Angles (16%). The offer, risk reversal and a reason to buy now.
• Messaging Consistency (9%). Do the homepage and the product page tell the same story with the same numbers?
• Site Experience (19%). Scored with the Brand Tested CRO Checklist by Cam Gawley: every sitewide, homepage and PDP item we can observe in public page code or on a dated capture, weighted 1 to 3. Cart and checkout items need a purchase, so they are listed but never scored. Items that a custom or headless build hides from public code are marked not scored instead of failed. Non-Shopify stores skip the Shopify-specific items (such as Shop Pay express checkout).
• Tech Stack (11%). Coverage of the seven app categories strong DTC brands run. Left out and reweighted for limited scans.

The rubric

Each category is a checklist. Points awarded divided by points possible gives a 0 to 100 score. Every awarded point on a brand page is backed by a short quote or observation.

Value Prop & Positioning

Point of View

Third-Party Validation

Offer & Angles

Messaging Consistency (site portion; ads and social added in full scans)

Site Experience

Tech Stack

We detect apps from public storefront code (scripts, app blocks and pixels). Seven core categories:

• Subscriptions
• Popups / list growth
• Post-purchase / upsells
• Reviews / UGC
• Referrals / affiliates
• Retention email / SMS
• Attribution

Limited scans. Our scanner reads public storefront code and sees the most on standard Shopify themes. When a storefront is not on Shopify, or is custom or headless and we detect fewer than 2 apps, we label it “Custom build, partial scan”. We show whatever we did detect, but we do not publish a Stack Score for that brand and Tech Stack is left out of its overall score, which is reweighted across the categories we could measure. A thin scan should never read as a weak stack.

Apps that run only in checkout or the back office can hide from a scan, so “not spotted” doesn’t mean missing. Brands can tell us through the corrections process.

Grade bands

The number is always the points a brand earns out of 100 on the checks above; we never curve it. The letter is set against the DTC field we score: our checks are strict enough that a perfect 100 means every checklist item is met, so even the best storefronts we have seen land in the high 80s. On our scale 82 or higher is in the A range, the middle of the field is a C+, and an F means a storefront is missing most of the basics (no visible proof, no clear offer, no clear headline).

Bands locked Oct 10, 2026 for methodology v1.3. They move only with a new methodology version, never brand by brand. Today 12 As, 65 Bs, 58 Cs, 35 Ds, 10 Fs across the 180 brands we have fully scored.

Who counts as the top 20

Whenever we compare stacks (“55% of the top 20 run an upsell app” and lines like it, on brand pages, The Stack and our data stories), the top 20 are the 20 brands with the highest score with the Tech Stack sub-score left out. Otherwise brands would rank high partly because of the apps they run, and the comparison would prove itself. The leaderboard and every brand’s published score still use the full weights above.

One public score

The Storefront Evidence Score is the only public score we publish. The Stack Score is its Tech Stack area, shown on its own. The Quick Score is a free preview of the same checks from one homepage and one product page. Customer sentiment is shown as the store’s own average star rating, never as a score.

Retail-first brands

The score grades a brand’s own online store, so it only ranks brands for which that store is a primary sales channel. A brand is proposed as retail-first when public evidence shows at least two of: a store locator or “find in stores” as a top-nav item or hero call to action; no add to cart on the main product page; the site sending shoppers to Amazon or a retailer for the main purchase; an editor flag (for example majority grocery or mass distribution, or ownership by a large CPG company). Cam Gawley approves the final list. Approved brands keep their page and storefront score with the label “Retail-first: we scored the online store, which isn’t this brand’s main channel. Not ranked on The Board.”, and leave The Board, category ranks, top 20 comparisons and the score mix. The first list is under review, so no brand is excluded yet.

Source of truth

Every count on this site comes from one of these definitions, as of our Oct 9, 2026 scan.

• Tracked brands: 1021. Every DTC and CPG brand in our index, scored or not.
• Successful scans: 982. Tracked brands whose storefront answered our scan.
• Stack denominator: 936. Successful scans where we could read the storefront code for apps. Every “X% of storefronts” stat uses this number.
• Fully scored: 180. Brands with a Storefront Evidence Score on every area of the checklist we can observe. These make up The Board.
• Coverage state. Each brand page says which it is: fully scored, scanned for its stack only, or custom build (partial scan, Tech Stack left out).

Top 20 comparisons use the 20 highest scores with Tech Stack left out (why).

Refresh

Graded brands are re-scanned over time. Each re-grade is dated; the previous grade stays visible in the page history.

Independence

Money never changes a grade. Payment, affiliate or partner status, claimed profiles and advertising have no effect on any grade, leaderboard position, inclusion or recommendation. Scores are produced with this methodology before commercial terms are considered. Read our independence statement.

Disclosure Standard. Every relationship between our editor and a brand we score is disclosed on that brand’s page and on any page that features it, and those brands never appear in homepage features. Scores are never changed for money or relationships. The full standard.

Customer sentiment

Each brand page carries a Customer Sentiment panel built only from ratings and reviews shown on the brand’s own product pages. Hero products are the top reviewed items in the store’s own best-selling sort. The average rating and review count are as displayed. Themes come from the most recent review text the page lets a visitor read; quotes are verbatim and trimmed to one sentence, with no names. When the page shows highest-rated reviews first, we say so and use the displayed rating breakdown instead of inventing complaints. We show the average star rating as the store displays it. It is not a score and does not feed the Storefront Evidence Score. The panel needs at least 25 public reviews, the same line where a star rating earns full credit in the score. Under 25, it says “Not enough public reviews yet.”

Corrections and appeals

Factual errors. If we got a fact wrong (a price, a feature, a detected app, a quote), email hello@brandtested.com (subject: Correction) or use the rescan form on the brand’s page, with the page URL, the statement you believe is wrong and your evidence. We acknowledge within 2 business days and resolve within 30 days. Corrections are noted on the page with the date.

Grade appeals. A brand may ask for a re-review if we made a factual error that affected the score, or if it has changed its site since our scan date. Disagreement with our judgment alone is not grounds for a change. A re-review produces a new dated grade.

Brand responses. A verified brand may submit a response of up to 150 words, published on its page and labeled as a response from the brand.

What we will not do. We do not remove accurate grades because a brand dislikes them, and we do not accept payment to change, hide or delay a grade.

Changelog

• Oct 10, 2026: v1.3. Social Signals removed from scoring. It never had measured data, so it was always left out and every score carried a “provisional” label; that label is gone. Its 10 points move to the other areas in proportion to their old weights, rounded to whole numbers: Value Prop 20, Point of View 11, Validation 14, Offer 16, Consistency 9, Site 19, Tech Stack 11. Rounding moved 15 of 180 scores by one point (11 up, 4 down); every other score is unchanged. The score is renamed the Storefront Evidence Score: how clearly a live storefront earns the next click. Customer sentiment now shows the average star rating instead of a 100-point number, so there is one public score.
• Oct 10, 2026: Stack comparisons (top 20 vs all storefronts) now rank the top 20 without the Tech Stack sub-score, so stack data never grades itself. Published scores and the leaderboard are unchanged.
• Oct 10, 2026: Customer Sentiment now needs at least 25 public reviews, matching the 25-review line for full star-rating credit in the score (it was 20). Labels: “Limited scan: custom platform” is now “Custom build, partial scan”. No score changed.
• Oct 10, 2026: v1.3. New grade scale, set against the DTC field: A range from 82 (A+ from 88), B range from 73, C range from 64, D range from 55, F below 55 (full bands above). The number is unchanged, still points earned out of 100. Under the old school-style scale no brand reached an A and the median brand sat on the C-/D+ line. Same release: header cart, search, bundle and upsell widgets, product-page reviews and homepage review carousels are now checked on a rendered capture of the page, the way a shopper sees it. Where an item shows only after the page loads, it earns 0.75 of its points (visible, quality unverified). 91 checks across 72 brands moved; no score went down.
• Oct 9, 2026: Review ratings and counts that load after the page source are checked on a rendered capture of the product page. A star rating earns full credit from 25 reviews up; below that, the review items earn a share in proportion to the count.
• Oct 9, 2026: Customer Sentiment added to brand pages: hero products, what people love and don’t, and a sentiment score from public reviews on the brand’s own product pages.
• Oct 9, 2026: v1.1 adds two sub-scores, Point of View (10%) and Third-Party Validation (13%). Their checks move out of Value Prop and Angles so nothing counts twice; other weights were trimmed to fit.
• Oct 9, 2026: Site Experience now runs on the Brand Tested CRO Checklist v1 (70 items, adapted from Cam Gawley’s Ultimate CRO Checklist for Shopify). Old-school advice such as security seal badges is replaced with modern equivalents like Shop Pay and express checkout. Scores moved for most brands.
• Oct 9, 2026: v1 published. Brand-first weights (value prop and angles lead; tech stack secondary).
• Oct 9, 2026: Expanded to 180 full audits and 1021 scanned brands across 12 categories. Same rubric. Limited-scan rule added: custom or headless storefronts with fewer than 2 detected apps get no Stack Score and Tech Stack is excluded from their overall score. The “Subscription or cadence option” check is scored only for products people reorder (consumables, supplements, personal care and similar); for durable goods such as apparel, jewelry and equipment it is left out and the Site Experience score uses the remaining checks.