How scoring works

Updated: May 2026

Every audit on etell carries three numeric scores out of 10. We score from observable signals in the captured artifact (the rendered email, the homepage screenshot) — not gut feel — so two persona-grounded reviews of the same artifact converge on the same number. This page is the rubric the persona applies; you can read each criterion, agree or disagree, and recalibrate.

The three scores

Business Impact answers: how well does this artifact target a person like the persona? Funnel A answers: will the persona move past the inbox / past the first screen? Funnel B answers: would the persona click / convert / add-to-cart?

The scoring method (every score)

Each score works the same way:

1. The score starts at 1.
2. The persona walks a 10-criterion checklist. For every criterion that's TRUE based on what's visible in the artifact, add +1 to the score.
3. The score caps at 10.
4. Each score is rendered with the per-criterion tally in the audit detail page so you can see exactly which signals counted.

That means a score of 10 is rare and means almost every signal is present. A score of 5 means about half of the signals are present. Most everyday emails / homepages land 4–7.

Email rubric

Business Impact (Email)

• Subject or hero copy explicitly references the persona's focus area.
• Sender is a brand the persona recognizes / is subscribed to.
• One concrete offer is visible (specific %/$ off, free-shipping, BOGO).
• Primary CTA is unambiguous (clear button copy + visible button).
• Visual hierarchy is clear — the eye lands on the offer / hero first.
• No render bugs (no overlapping text, broken images, lorem-ipsum, layout breakage).
• Demographic signals match the persona (age / gender / style of models).
• Email reflects current campaign / season (not stale promo language).
• Loyalty / member benefits visible if the persona is a member.
• Offer feels honest (no buried fine print, exclusionary language, bait-and-switch).

Open Likelihood (Email)

• Sender display name is recognizable.
• Subject is concrete (specific offer, product, urgency).
• Subject is relevant to the persona's focus area.
• Preview text complements the subject (does NOT repeat it).
• Preview text is real copy (not “view in browser” / unsubscribe junk).
• Subject is under ~50 chars (mobile-friendly).
• No spam signals (ALL CAPS, !!! exclamations, “FREE!!!”).
• Personalization or segmentation hints (the persona's name, category).
• Time-bounded urgency that feels credible.
• Cadence feels right (not the 4th near-identical promo this week).

Click Likelihood (Email)

• Hero offer is visible without scrolling on mobile.
• Primary CTA is in the persona's category.
• CTA copy is specific (“Shop Slip-ins”, not “Discover”).
• Offer reduces price OR has loyalty member pricing.
• Offer is time-bounded with a credible deadline.
• One specific product / hero linked (not just to homepage).
• Sizing / fit / availability info visible.
• Reviews or social proof visible.
• Brand voice consistent and trusted.
• No friction (no “view in browser” first, no broken-image gaps).

Web rubric

Business Impact (Web)

• Hero copy or imagery explicitly speaks to the persona (focus area, age, generation, lifestyle).
• A returning-shopper hook is visible (loyalty / member CTA, recently-viewed, rewards balance).
• One concrete offer is visible above the fold (specific %/$ off, free-shipping threshold).
• One unmistakable primary CTA is visible above the fold.
• Visual hierarchy is clear — the eye knows where to land first.
• No render bugs (overlapping text, broken images, placeholder lorem-ipsum, layout breakage).
• Demographic signals match the persona.
• Page reflects current campaign / season.
• Loyalty / membership benefits visible without scrolling.
• Offer feels honest (no buried fine print, exclusionary language, bait-and-switch).

Engagement Likelihood (Web)

The persona is already on the homepage — engagement scores whether they'd scroll past the first screen / tap a category / interact, vs. bounce.

• Hero relates to the persona's shopping focus area.
• Visible navigation makes it easy to reach the persona's category in 1 tap.
• A category or product image catches the eye for what the persona shops.
• A promo banner offers something the persona would actually use.
• The page renders cleanly (no obvious perf issues, no layout shift).
• Imagery includes someone like the persona (age / style / situation).
• Copy register matches the persona's reading style.
• Trust signals visible (review counts, badges, free returns).
• New-arrivals or “what's hot” rail visible in the persona's category.
• No dark patterns / forced modals making the persona want to bounce.

Conversion Likelihood (Web)

• Primary CTA is in the persona's category.
• CTA copy is unambiguous (“Shop Slip-ins”, not “Discover”).
• Active offer reduces price OR has loyalty member pricing.
• Offer is time-bounded with a credible deadline.
• Free-shipping threshold visible AND reachable for typical AOV.
• A specific product page is one tap away (not buried in nav).
• Sizing / fit info visible or one tap away.
• Returns / exchanges policy is mentioned.
• Reviews or ratings visible on featured products.
• Trust signals at checkout depth (secure-checkout, pay-later, etc.).

Why a checklist instead of a vibe

Persona-grounded scoring is inherently subjective — the persona is a 62-year-old comfort shopper, not a regression model. But the question we want answered isn't “what's the feeling?”; it's “what's actually on the page that a 62-year-old would respond to?” Anchoring scores to observable signals lets reviewers — including humans — disagree with the persona on a specific criterion (“the loyalty callout is there, you missed it”) instead of arguing about a vibe.

What scores are not

These are notprobability predictions. A “9/10 Open Likelihood” doesn't mean the email has a 90% open rate in production. It means the persona, with their particular shopping habits and their accumulated history with the brand, would likely open this one — based on the signals listed above. Treat the scores as relative: useful for comparing two emails or two homepages from the same persona's point of view, less useful as forecasts.

Questions or feedback?

Email alon@etell.app. The rubric is living — if a criterion is missing or wrong for your category, we'll iterate.