How the IPTV Help Score works

The IPTV Help Score is a number from 0 to 100 that summarizes what members report about an IPTV provider. Our ranking system calculates it automatically, with the same formula for every provider.

Nobody at IPTV Help can set or adjust a provider’s score, and no provider can pay to change its score or its place on the list. Scores come only from the automatic ranking system, using the formula on this page. Editors can place a provider at a chosen position on the public list; those positions are always marked “Editor’s pick” (see below). Current formula version: ihs-1.0.0-launch.

At a glance

• Customer rating (the stars): the plain average of all published customer reviews, shown with the number of reviews next to it, from the first review onwards. Staff and affiliated accounts are never included.
• Score: 0 to 100, shown to one decimal, with a ± margin and a confidence label. Unlike the stars, the score uses the weighted formula below.
• Make-up: 90 % member ratings + 10 % measured reliability.
• Waiting period: every review counts 1 hour after it is published, positive or negative alike.
• Ranked when: at least 1 counted reviews, enough review weight (evidence of at least 0.3), at least 1 review from the last 365 days, and at least 1 day of website checks.
• Updates: the ranking recalculates every 15 minutes, and again after each review is published.
• Never counted: payment of any kind, provider requests, and the site owner’s commercial interest. The owner’s own service is not in the rankings at all.

What goes into the score

90 % comes from member reviews. Each review has an overall star rating (1 to 5, whole stars only) and six sub-ratings. The overall stars and the six sub-ratings each count for half of the member rating: the overall stars capture everything we didn’t ask about, and the sub-ratings make sure the things that matter most, such as stream stability, carry the most weight.

Stream stability carries the most weight because it is the most common reason members leave a provider, and you can’t check it before you pay. Ease of setup carries the least because it is a one-time cost.

10 % comes from reliability, described next. Reliability gets a small share on purpose: a reachable website is objective, but it is only a weak sign of service quality.

Reliability (10 %)

Our automated website monitor checks each provider’s public website every 5 minutes. Reliability looks at the last 30 days and has two parts:

• Website reachability (60 % of reliability). The share of checks where the website responded. At or below 90 % of checks this part scores zero; at 99 % or more it scores full marks, with a straight line in between.
• Member outage reports (40 % of reliability). A day counts as an outage day when at least 3 different members report a problem with the provider. It counts in full if Watchtower also saw at least 2 failed checks that day, and at 50 % otherwise, because streams can fail while the website stays up. 10 outage days in the window bring this part to zero. If no outage days are recorded, this part is at its maximum.

Website checks are not stream checks. The monitor only asks whether the public website responds. It never logs in, never uses credentials and never opens a stream, so it says nothing about picture quality, buffering or channel availability. Those come only from member reviews.

If a provider has fewer than 1 day of website checks, its reliability is set to a neutral middle value and it is not ranked yet.

When a review starts to count

A published review is visible straight away, but it starts counting toward the score 1 hour after it is published. The wait is the same for positive and negative reviews. If positive reviews counted instantly while critical ones waited for a moderator, scores would drift upward and fake praise would always be faster than honest criticism. The wait also gives our checks time to run before anything moves a score.

Some reviews are checked by a moderator before they are published; for those, the 1-hour wait starts when the moderator publishes them:

• Reviews from people without an account are always checked by a moderator before publishing. Their author starts at the lowest trust weight (0.60), like any brand-new account.
• Reviews from new members are checked the same way.
• Reviews with an overall rating of 2 stars or less are checked so that serious claims meet our review guidelines. They are held for the same reason, not because they are negative.

Invited reviews. Our team sometimes sends a customer a one-time link to review a provider they use. The customer writes the review themselves. It is published straight away at any star rating, is marked “Invited review”, waits the same 1 hour before it counts, and carries the same trust weight as any new reviewer. Staff can’t write reviews, and each link works once and expires after at most 30 days.

If a reviewer edits a review, it drops out of the score and waits another 1 hour from the edit.

Not every review counts the same

Each review gets a weight. Weights are measured in “baseline reviews”: a fresh review from an established member, with no helpfulness votes, no integrity concerns and a 1 to 6 month subscription, weighs exactly 1.0. The factors below are multiplied together, and no single review can count for more than 1.5 baseline reviews.

Reviews from accounts that work for, own or resell any IPTV service weigh zero, as do reviews from accounts banned for manipulation.

Few reviews are pulled toward the typical rating

A provider with three glowing reviews should not outrank one with fifty good ones. So each provider is scored as if it already had 5 baseline reviews at a typical rating. As real reviews build up, they outweigh that starting point. Statisticians call this a Bayesian average.

The typical rating starts at 3.0 out of 5. Once at least 10 providers each have at least 3 baseline reviews of weight, it becomes the average rating across those providers, kept between 2.75 and 3.75 so that site-wide enthusiasm or negativity can’t give unknown providers a head start or sink everyone. The overall stars and each sub-rating are averaged this way separately.

The formula

For readers who want the exact maths. “Σ” means “add up over all counted reviews”, and prior_k is the typical rating described above.

w      = trust × account age × recency × integrity × helpfulness × subscription
         (one review never counts for more than 1.5)

S_k    = (5 × prior_k + Σ w × stars_k) / (5 + Σ w)
         (for the overall stars and for each sub-rating separately)

Sub    = 0.30 × S_stability + 0.15 × S_quality + 0.15 × S_content + 0.10 × S_setup + 0.15 × S_support + 0.15 × S_value
MR     = 0.5 × S_overall + 0.5 × Sub          (1 to 5 stars)
M      = (MR − 1) / 4                                  (0 to 1)

Rel    = 0.6 × uptime part + 0.4 × outage-report part   (0 to 1)

Score  = 100 × (0.9 × M + 0.1 × Rel)

A provider rated 1 star on everything can score at most 10.0, and only with perfect reliability. A provider rated 5 stars on everything with perfect reliability scores 100.0.

When a provider is ranked

A provider appears in the rankings only when all of these are true:

• it is listed in our directory and is not the site owner’s own service;
• it has at least 1 counted reviews (one per person);
• it has evidence of at least 0.3. Evidence is the total review weight, reduced when a few reviews dominate (technically, the smaller of the total weight and the effective number of reviews);
• it has at least 1 review from the last 365 days;
• it has at least 1 day of website checks.

Until then it has no score and no rank from customers. Its customer rating (the plain average of its published reviews) is shown from the first review, always next to the number of reviews, so you can see how many opinions it rests on: “5.0 (1)” is one person. On the list it appears after scored providers: those with customer reviews first (more reviews, then higher average), then the rest by live website speed. Staff reviews never affect this order. We don’t show a “needs N more reviews” countdown, because it would invite solicited reviews.

These are launch thresholds, set low so the leaderboard can form within days. We expect to raise them as more members review; any change gets a new formula version and an entry in the changelog.

Margin and confidence

Each score shows a margin, for example 72.4 ± 6.1. It is an approximate 90 % range for the member part of the score. It is wide when there are few reviews or when reviewers disagree, and it narrows as evidence builds up. Providers whose ranges overlap may not really differ.

The confidence label comes from the evidence figure: Low below 10, Medium from 10 up to 30, and High at 30 or more.

Ties, rank numbers and movement

Competition ranks. Ranks are decided on the score rounded to 1 decimal. Equal scores share a rank and the next rank is skipped (1, 2, =2, 4). Within a tie, the list order is: higher lower bound (score minus margin), then more evidence, then more counted reviews in the last 90 days, then an internal ID. The ID is not the name, so calling yourself “AAA IPTV” gains nothing.

Weekly movement arrows. Each ranked provider’s rank is compared with its rank 7 days earlier: an up or down arrow with the number of places moved, a dash for no change, or New if it was not ranked then.

When scores update

The ranking system recomputes every provider’s score and rank every 15 minutes, and again after each review is published. A newly published review changes the score at the first recompute after its 1-hour wait ends. The rankings page and the home page show when the last recalculation ran.

What never affects a score

• Payments of any kind: advertising, sponsorship, commissions or “featured” slots. None of these exist in the ranking system.
• Requests from providers, including requests to remove reviews.
• Page views, clicks, listed prices or a provider’s own description of its service.
• Provider replies to reviews. They are shown but never scored.
• Staff opinions, forum posts that are not reviews, and AI-written summaries. (Editors can pin a list position, always labelled “Editor’s pick”, but can’t change a score.)
• The site owner’s commercial interest. The owner’s own service has no listing, no score, no rank and no reviews here, and owner and staff accounts can’t submit provider reviews.

Editor’s picks

Our editors can pin a provider to a chosen place on the public list, for example to feature a newly listed provider. Every pinned place shows an “Editor’s pick” label, so you can always tell an editorial position from a customer-score position. Pinning never changes a provider’s score, stars or reviews, and no provider can pay for a pinned place. IPTV Help is run by the owner of other IPTV services (see our disclosure), so keep this in mind when you see an editor’s pick.

How we limit manipulation

• One review per person per provider, plus posting limits and an anti-spam check on the review form. Invite links work once, and every review that came from one is labelled.
• The 1-hour wait and the account-age and trust factors make bursts from new accounts slow and weak.
• Accounts that work for, own or resell any IPTV service don’t count toward any score.
• Moderators remove reviews only for clear rule violations, such as a second account, an undisclosed connection to a provider, or a review written in exchange for payment or a free month. We never remove a review because it is negative, because a provider asked, or because it suits the site owner.
• Reviewers and providers can appeal decisions. See the community guidelines.

Limitations

• Reviews are self-reported experiences. We read them and run checks, but we can’t confirm what happened to each reviewer.
• No purchase verification. We can’t confirm that a reviewer actually paid for the service, so we show no “verified” badges.
• Website checks are not stream-quality checks. A provider’s website can be online while its streams are failing, and the reverse.
• Early scores are rough. With the low launch thresholds, a score may rest on only a few reviews. Look at the margin and the confidence label.
• Determined manipulation is possible. A patient attacker with many aged accounts can still move a score. Our checks make that slow, costly and visible, but not impossible.
• Your experience may differ. Results vary with your internet provider, device, region and time of day. A high score doesn’t guarantee a service will work well for you.
• The score says nothing about legality. We don’t assess whether a service is licensed or legal to use in Canada. Check the rules that apply to you before you subscribe.

Changelog

The formula is versioned. Every change applies to all providers at once and is listed here with its date.

Questions

If something here is unclear or looks wrong, email help@supportiptv.ca or see our contact page.