What a healthy click quality score looks like (and why it's not a gate)
Here is the verdict first: the click quality score is a signal, not a verdict on a person. It reads the shape of the traffic an athlete's tracking link sends, gives brands a sortable number to weigh alongside everything else, and runs in shadow mode. A low score does not block an athlete, does not remove anyone from a program, and never reverses a commission on its own. Money only moves through the ledger; the score never touches it.
The rest of this article covers what the score measures, why it stays a signal rather than a gate, what healthy and unhealthy traffic look like at a high level, and how confirmed fraud is handled on a separate, human-reviewed track.
What the click quality score measures
The score looks at the pattern of clicks an athlete drives over time, not at any single click. It is built from three plain inputs.
- IP and fingerprint diversity. Real audiences come from many devices and networks. The score looks at how spread out the clicks are across distinct IP addresses and browser fingerprints. A healthy spread reads as a real, varied audience.
- Bot health. Each click carries a bot score from the tracking edge. The quality score rolls those up to gauge how much of the traffic looks automated versus human.
- Conversion sanity. The score checks whether the relationship between clicks and sales is sane. A real audience clicks, browses, and some share buys. Traffic that converts in a way no real audience would is what this input is built to notice.
These run nightly across an athlete's recent clicks. The result is a single number, the click quality score, that summarizes traffic quality, so a brand can sort partners by it the way they would sort by total sales or conversion rate.
The score reads traffic, not the person
The click quality score is computed from click and conversion data, not from an athlete's audience size, follower count, or content. It is a read on the shape of the traffic, and it can move up or down as that traffic changes.
It runs in shadow mode, as a signal for brands
The score runs in shadow mode. That means it is computed and shown, but it does not act on anything by itself. It changes no balances, sends no payouts backward, and removes no one from a program.
For a brand, it is one sortable column among several. When a brand reviews partners in athlete search, the score sits next to sales, conversion rate, and the rest of the picture. It points a brand toward where to spend attention, and it informs a conversation. It never makes a decision for them.
That design is deliberate. A single number, computed automatically, is a useful prompt and a poor judge. Traffic patterns have honest explanations: a viral post, a feature in a newsletter, a slow month, a new audience that has not warmed up yet. A signal invites a human to look. A gate would punish the honest cases along with the rest, so we built a signal.
It is not a gate
This is the part worth saying plainly, because it is where most platforms quietly do the opposite.
- A low score does not block an athlete. Browsing and applying stay open. Approved partners keep their tracking link and keep their place in the program.
- A low score does not remove a commission. Earned commission stays earned. The score has no path to the ledger and cannot reverse, hold, or reduce a payout.
- A low score is not a strike. It is a snapshot of recent traffic that can recover as the traffic does. It carries no permanent mark.
Payouts run on their own track and on their own rules: you earn the commission the brand sets, in full, and the 20% platform fee is billed to the brand on top, never deducted from your payout. The score sits entirely outside that path. If you are an athlete watching your score move and wondering whether it affects your money, it does not. See athlete troubleshooting if a number on your dashboard looks off.
What healthy vs unhealthy traffic looks like
At a high level, the contrast comes down to whether the traffic looks like a real, varied audience or like a machine. Here is the shape of each, without the exact thresholds.
| Signal | Healthy traffic | Unhealthy traffic |
|---|---|---|
| IP and fingerprint diversity | Many distinct devices and networks, spread out over time | The same handful of addresses or fingerprints repeating |
| Bot health | Clicks read as human, with normal timing | A high share of automated clicks, or clicks in machine-like bursts |
| Conversion sanity | Clicks and sales hold a normal relationship for the audience | A click-to-sale pattern no real audience would produce |
| Over time | Rises and falls with real posts and real audience behavior | Spikes that do not match any real activity |
A healthy score follows real life. It dips in a quiet stretch and climbs when a post lands, and none of that movement touches your payouts. For brands, this is exactly the kind of context that makes running a curated program work: a sortable read on traffic quality, paired with two-way approval and identity-verified partners.
How confirmed fraud is handled, separately
The quality score and actual fraud handling are two different things, on purpose.
The score is a soft signal that never moves money. Confirmed fraud is a separate track that a person drives. When a real fraud signal surfaces, it goes to a review queue for an admin to look at, not to an automatic action. From there:
- An admin reviews the case with the underlying signals in front of them.
- If it is dismissed, nothing happens. The athlete and their balances are untouched.
- If fraud is confirmed, it routes into the existing dispute and writeoff path, where any reversal is a deliberate, recorded decision.
There is no auto-clawback anywhere in this flow. Nothing reverses a commission because a number crossed a line. A human reviews, a human decides, and the decision is written to the ledger like any other. That keeps the honest cases safe and keeps every reversal accountable to a person.
The short version
The score is a signal for sorting. Fraud handling is a separate, human-reviewed track that routes confirmed cases to dispute or writeoff. Neither one reverses your commission automatically.
FAQ
Does a low click quality score get an athlete kicked off?
No. The score runs in shadow mode and does not block, suspend, or remove anyone. Browsing and applying stay open, and approved partners keep their tracking link and their place in the program. A low score is a snapshot of recent traffic that a brand can weigh; it can recover as the traffic does, and it carries no permanent mark. Removing a partner is a brand decision made by a person, never an automatic action triggered by the score.
What makes click traffic look healthy?
Healthy traffic looks like a real, varied audience. That means clicks spread across many distinct IP addresses and device fingerprints, a bot health read that looks human rather than automated, and a sane relationship between clicks and sales for that audience. Healthy traffic also rises and falls with real activity: it climbs when a post lands and dips in a quiet stretch. The score reads the shape of the traffic over time, not any single click.
Can a quality score automatically reverse my commission?
No. The click quality score has no path to the ledger and cannot reverse, hold, or reduce a payout. You earn the commission the brand sets, in full, and the 20% platform fee is billed to the brand on top, never deducted from your payout. Commissions only reverse through two recorded paths: a refund inside the program's refund window, which is pro-rated to the amount returned, or confirmed fraud that an admin reviews and routes to dispute or writeoff. Both are deliberate and written to the ledger. There is no auto-clawback from a score.
The Harmonia team — Notes from the team building the US Health & Wellness partner platform.