How Trustpilot's TrustScore is calculated (and 5 levers to improve yours)

Quick answer: how is TrustScore calculated?

Trustpilot's TrustScore is a Bayesian-weighted, time-decayed average of all reviews on a business profile, with extra weight for recent reviews. It's not the simple arithmetic mean of star ratings (which is what most other platforms display).

This article explains the actual formula (as documented in Trustpilot's own help center and reverse-engineered by independent researchers), why it matters for Shopify stores, and the 5 practical levers to improve yours.

The actual formula

Trustpilot's TrustScore for a business is calculated approximately as:

TrustScore = ( Σ(rating_i × weight_i) + (B × M) ) / ( Σ(weight_i) + B )

Where:

• rating_i = each review's 1-5 star rating
• weight_i = time-decay weight for each review (1.0 for last 12 months, decays exponentially)
• B = Bayesian prior weight (Trustpilot uses approximately B = 9 for new businesses)
• M = Bayesian prior mean (approximately 3.5/5)

In English: every business starts as if it had 9 imaginary reviews at 3.5 stars. As you collect real reviews, those imaginary reviews get washed out, and your real average dominates. This is why a brand-new profile with one 5-star review shows a TrustScore of around 3.65, not 5.0.

Time decay

Reviews older than 12 months are progressively down-weighted. Trustpilot's reverse-engineered decay function looks roughly like:

• 0-12 months old: 100% weight
• 12-24 months old: ~70% weight
• 24-36 months: ~45% weight
• 36+ months: ~30% weight

This is why active review collection matters: a profile that stopped collecting reviews 2 years ago will see its TrustScore drift toward 3.5 (the Bayesian mean) as old reviews lose weight.

Why your TrustScore feels "wrong"

The most common confusion: a merchant has 50 5-star reviews and 2 1-star reviews, calculates 4.85 mean, and sees a TrustScore of 4.4. Where does the 0.45 difference come from? Three reasons:

• Bayesian smoothing: with only 52 reviews, the imaginary 9 reviews at 3.5 still pull the score down ~0.3 points
• Time decay: if half your reviews are over 12 months old, their weight is reduced
• Hidden adjustments: Trustpilot occasionally applies penalties for suspicious patterns (sudden volume spikes, suspected fake clusters) that aren't disclosed publicly

Use our NPS calculator for the related (simpler) Net Promoter Score formula; TrustScore is more complex by design because Trustpilot wants to make manipulation harder.

The marginal impact of each new review

Most merchants want to know: "if I get one more 5-star review, how much does my score change?" The answer depends on your existing volume.

Two takeaways: (1) volume is your moat. A 5,000-review profile barely moves on any single review. (2) negative reviews hurt 2.5-3x more than positive reviews help at any volume, because they pull harder against the Bayesian mean.

How many 5-star reviews to "cancel out" a 1-star?

The classic question (and a top PAA on Trustpilot SERPs). The answer:

• From a 0-review baseline: 4 five-stars cancel 1 one-star to reach a 4.0 average.
• From a 4.5 average with 100 reviews: ~11 five-stars to recover from a single new 1-star.
• From a 4.5 average with 500 reviews: ~3 five-stars to recover.
• From a 4.5 average with 2,000 reviews: ~1 five-star is enough.

The pattern: the more reviews you have, the less any individual review (positive or negative) moves your score. Volume insulates.

Reviewz.ai for Shopify — automatically routes happy customers to leave reviews on Trustpilot, Google, and Judge.me, while privately catching unhappy ones in a feedback portal before they post a public 1-star. Re-engage every reviewer with upsell offers via WhatsApp, email, and SMS.

Install Reviewz on the Shopify App Store →

5 levers to improve your TrustScore

Lever 1: Volume (biggest impact)

Going from 100 to 500 reviews of the same average rating raises your TrustScore by ~0.15 points (because the Bayesian prior weight gets diluted). This is the single highest-leverage move. Active collection beats passive collection by 5-8x in volume.

For the playbook on collecting more, see our 9-tactic guide.

Lever 2: Recency

Reviews from the last 12 months get full weight. Reviews older than 24 months get 50% or less. If you collected most reviews 2 years ago and stopped, your TrustScore is silently fading. Resume collection to refresh the recency weighting.

Lever 3: Response rate

Trustpilot doesn't directly factor your response rate into TrustScore (officially), but Proserpio & Zervas (2017) documented a 12% lift in positive reviews for businesses that respond to negative ones. Higher response rate → more positive reviews → higher TrustScore over time.

For the framework, see our L.A.S.T. response guide.

Lever 4: Verified ratio

Reviews invited via Trustpilot's API (paid plan) carry a green "Verified" badge. While Trustpilot doesn't disclose if Verified reviews carry more weight in TrustScore, they're more resistant to flagging by other users (reducing the chance of false-positive removals that hurt your score).

To raise your verified ratio without paying for Trustpilot Standard, send invites from your own infrastructure pointing to your Trustpilot review URL; reviews collected this way don't get the badge but still count.

Lever 5: Removing flagged content

Genuinely fake reviews on your profile drag your TrustScore down. Flag them following our step-by-step removal guide. Trustpilot removes around 65% of valid flagged reviews; the rest stay but at least you've done your part.

The realistic timeline to improve a TrustScore

Trustpilot's algorithm rewards consistency, not bursts. From real Shopify merchants we've worked with:

• 4.0 to 4.5: typically 6-9 months of consistent collection (100+ new reviews/month)
• 4.5 to 4.7: 9-12 months
• 4.7 to 4.9: 18+ months and almost no negative reviews (which means actually fixing the underlying issues, not just collection volume)

The 4.9+ tier is rare and slow to reach. Customers reading reviews understand this; 4.5 with 1,000 reviews is more credible than 4.9 with 50 reviews. Volume + consistency over chasing perfection.

Reviewz.ai for Shopify — automatically routes happy customers to leave reviews on Trustpilot, Google, and Judge.me, while privately catching unhappy ones in a feedback portal before they post a public 1-star. Re-engage every reviewer with upsell offers via WhatsApp, email, and SMS.

Install Reviewz on the Shopify App Store →

Common myths about TrustScore

"Paying Trustpilot improves your score"

False, with a caveat. The TrustScore formula is the same on free and paid plans. However, paid plans give you the Invitation API (more reviews, faster) and the flagging tool (faster removal of clearly fake content), both of which indirectly help your score. The score itself isn't directly bought.

"You can ask Trustpilot to manually adjust your score"

False. Trustpilot's policy is clear: scores are algorithmic. They will remove specific reviews under valid grounds, but they won't manually boost or reset a score.

"Old reviews don't count anymore"

Partially true. Reviews never disappear, but their weight decays significantly after 12 months. After 36 months, they contribute about 30% of their original impact. So old reviews still matter, just less.

The bottom line

TrustScore is designed to be hard to manipulate, which is also why it can feel slow to move. The five levers that actually work are: collect more (volume), collect recently (recency), respond to all (engagement), use verified invites where possible (badge), and remove genuine fakes (cleaning). There are no shortcuts; Bayesian smoothing was specifically designed to prevent them.

If your TrustScore feels stuck below 4.5, the answer is almost always one of two things: not enough recent reviews (collection bottleneck) or unaddressed underlying customer experience issues (which no amount of new reviews will fully fix). Diagnose which one you have, then act.

References:

• Trustpilot TrustScore documentation. Link
• Proserpio & Zervas (2017). Online Reputation Management. Marketing Science. Link
• Bayesian average rating systems (general background). Wikipedia