Trust Grade Methodology

How every Trust Grade is computed

Every letter grade you see on Howsociable is derived deterministically from cited signals. No editor tweaks the score by hand. This page documents the exact formula — what feeds in, how each source is weighted, and where the letter-grade thresholds sit.

See every Trust Grade Trust Center

1. Signals are collected with a cited source URL

A “trust signal” is a single externally-verifiable data point — a Trustpilot aggregate, a Reddit thread, a BBB complaint count, a news article, or a result from our own 30-day test. Every signal an editor records has:

A source (e.g. Trustpilot, Reddit, HowSociable test)
A polarity — positive, negative, or neutral context
A category (authenticity, retention, delivery, support, …)
A sourceUrl linking directly to the primary source so any reader can verify it
An optional admin weight override(1-10) when a default doesn't fit the context

Nothing is paraphrased or generated. If we can't link to the origin, we don't cite it.

2. Each source has a default reliability weight

Not all sources are created equal. A moderated review on a regulated platform counts for more than an anonymous tweet. The rule of thumb: the harder it is to fake and the stricter the moderation, the higher the weight. Our own 30-day tests top the scale because we can guarantee they happened.

Source	Default weight
HowSociable 30-day test	10/10
Better Business Bureau	8/10
Trustpilot	8/10
G2	8/10
Trusted Reviews	7/10
Sitejabber	6/10
Reviews.io	6/10
Google Reviews	6/10
News / press	6/10
Reddit	5/10
Published refund policy	5/10
Independent blog	4/10
Forum	4/10
Domain WHOIS	4/10
Social media	3/10
Search engine	3/10
SSL certificate	3/10
Other source	2/10

Admins may override any single signal's weight (capped at 10) when context warrants — for example, reducing a Trustpilot row if the profile has obvious fake-review patterns.

3. Negative signals hit 30% harder than positives

A positive signal contributes weight × 1.0 to the score. A negative signal contributes weight × −1.3. Neutral signals add context but don't move the score.

The ~30% penalty mirrors what Trustpilot, Sitejabber, and the BBB all do: critical reviews are costlier to fabricate, and satisfied buyers rarely write one. Weighting negatives harder is how these platforms prevent a product from drowning out legitimate red flags in a flood of five-stars.

4. Signals roll up into nine categories

Every signal is tagged with one category. We aggregate signals within each category (weighted sum, then normalized to 0-100) to produce the spider-chart you see on every Trust Report. This makes it obvious whether a service is strong on retention but weak on support, for example.

Authenticity
Are followers / engagement real humans, or bots that will be swept away?
Quality
How closely does what's delivered match what's advertised?
Value
Is the price fair for what you actually receive?
Support
Can you reach a human, and do they help?
Retention
Do the followers / likes stick, or drop off within days?
Delivery
Does it arrive on time and at the promised volume?
Privacy
How are your account credentials and personal data handled?
Refund handling
If something goes wrong, do they honour the refund policy?
Transparency
How transparent is the company about who they are?

5. Overall score maps to a letter grade

The overall score is the weighted average across every signal — each signal contributes effectiveWeight × polarityMultiplier. The result is normalized to a 0-100 scale where 50 = neutral, 100 = fully positive, and 0 = fully negative. That number maps to a letter:

Letter	Score range
	95–100
	90–94
	85–89
	80–84
	75–79
	70–74
	65–69
	60–64
	55–59
	50–54
	40–49
	0

Thresholds mirror the spread used by Trustpilot / Sitejabber / BBB: most “good” companies cluster in the B to A- range. A+ is reserved for standout performers.

We refuse to grade thin data

Below a minimum combined weight (roughly the equivalent of one high-reliability signal plus one mid-tier signal), we publish “Gathering data” instead of assigning a letter. Inventing a grade from one or two data points would be misleading — a statistical accident, not a judgment. Products in this state show up as Pending on the Trust Reports index.

What the Trust Grade is not

Not our editorial verdict
That lives in the review itself. The Trust Grade is a deterministic signal aggregate, not an opinion piece.
Not pay-to-play
No vendor can buy a grade. No affiliate relationship moves the score. Ad disclosures live on every affected page.
Not a Trustpilot clone
Trustpilot, Sitejabber and BBB are inputs into our grade — weighted alongside independent sources, news coverage, and our own tests. We never treat any single platform as ground truth.
Not frozen in time
Every grade recomputes whenever a signal is added, retracted, or re-weighted. The observedAt date on every source tells you how fresh each input is.

See the methodology applied

Every reviewed service, ranked by its cited Trust Grade.