Low confidence
Useful for early context, but too thin for a strong ranking claim.
Research methodology
The score is the start. The sample tells you how much to trust it.
HarborMeds turns public review data into category scores, then keeps review volume, freshness, and known limitations beside the result. This page explains the process without hiding the awkward parts.
Methodology reviewed July 21, 2026
SourceRecent reviews
Ratings, dates, titles, and review text
SortTopic matches
Support, shipping, value, billing, and other categories
ScoreCategory ratings
Average star rating converted to a 100-point score
WeightProvider score
Category scores weighted by the number of mentions
Trustpilot review scoring
How a provider score is calculated
HarborMeds normally analyzes reviews dated within the last 180 days. If fewer than 20 dated reviews fall inside that window, the report uses all available dated reviews and labels the result as historical rather than recent.
Reviews are matched to topic categories using plain keyword rules. The current categories cover support, price and value, shipping, getting started, medication experience, side effects, and billing and refunds. One review can match more than one category.
The formula
Two calculations, both shown on a 100-point scale.
Category score(Average star rating / 5) multiplied by 100
Provider scoreCategory scores weighted by their number of mentions
The provider score is not the same as the Trustpilot star rating. Categories with more mentions carry more weight. Every matched category with a rating contributes to the provider score, including categories that have too few mentions to appear as a strength or risk. If no category scores are available, the report falls back to the average rating for the selected review window.
Confidence labels
A high score and a high-confidence score are different claims.
Medium confidence
Enough recent reviews to compare, with sample size kept visible.
High confidence
A larger sample that gives repeated category patterns more weight.
Category labels
Strength, mixed, or risk
A category needs at least eight mentions before HarborMeds displays it as a strength, mixed area, or risk. A strength usually needs a score of at least 75 and a negative share below 30%.
Risk thresholds are stricter for billing and refunds. That category becomes a risk when at least half of its mentions are negative or its score falls below 60. Other categories become risks when negative mentions reach 60% or the score falls below 55. Side-effect praise is treated more cautiously and needs at least 15 mentions, a score of 85, and a negative share below 20%.
Reddit discussion
Context, not verified customer data
Reddit reports collect public posts and comments that match provider names or configured topic searches. HarborMeds uses them to look for recurring questions about pharmacy handoffs, shipping, refunds, support, and pricing.
Reddit identity and purchase history are not verified. Those discussions do not affect the Trustpilot-based provider score, and the site labels them as community discussion rather than customer reviews.
Automation
What software does, and where it can be wrong
Trustpilot category matching uses fixed keyword rules. Reddit sentiment and topic tags use an OpenAI model in batches. If that analysis fails, the system records a neutral result instead of inventing a positive or negative score.
Automated labels can miss sarcasm, context, or who a complaint is actually about. Site summaries and excerpts are reviewed before publication, but that review does not turn public comments into verified facts.
Rankings
How scores become a shortlist
Rankings begin with the current provider score. Review count, freshness, and confidence decide how much weight that number deserves. A high score from a small or old sample can sit below a slightly lower score backed by many recent reviews.
Affiliate and referral relationships do not set ranking order. Pages can include monetized links, but those relationships are disclosed separately and do not remove cautions from the research.
Limits
What these numbers cannot tell you
- Public reviewers are self-selected and may not represent every customer.
- HarborMeds cannot independently verify every review or Reddit account.
- Saved source pages may not contain every review on a provider profile.
- Keyword matching can place one review in several categories.
- Overall confidence does not make every category equally well supported.
- Prices, pharmacies, medication availability, and program terms can change.
- The scores measure reported service patterns. They do not measure medical quality.
Corrections and updates
See something that needs another look?
Send the page URL, the specific statement, and any supporting source to support@harbormeds.com. Provider corrections are reviewed against current public information. A correction request does not guarantee a score or ranking change.