Report card

hostingstep.com

Payout–Rank not computable — in every article, every pick we can price carries the same rate (1 monetized articles, 0 with two different rates to compare)

Generated 2026-09-17 under SPEC: Scoring — v0.6.12. State: Not scored — no payout variation to correlate.

Monetization exposure — what was observed

A pick whose status was not established is counted apart from a pick observed to be unmonetized. Each reason is published on its own, because they are different situations and one of them is about us rather than about the publisher:

Marked sponsored by the publisher — what was observed on those links

No direct merchant links on this publisher carry the publisher’s own rel="sponsored" without a detected affiliate mechanism.

Declared, not observed

3 ranked pick(s) are priced against a merchant the publisher DECLARED in the link itself (a merchant= parameter): hostinger.com. THE DESTINATION WAS DECLARED, NOT OBSERVED — we did not follow the link and do not know that it arrives there. What we observed is the publisher naming the merchant in its own markup; the rate is that merchant’s own published schedule, with its source and date shown beside every number. 1 further pick(s) declare a merchant we cannot resolve to a rate card — an opaque publisher id, or a merchant whose schedule we do not hold. Those stay unpriced (r = null) and are excluded from the correlation. They are counted here so the gap is visible rather than silent.

Payout–Rank correlation

The card records Payout–Rank not computable — in every article, every pick we can price carries the same rate (1 monetized articles, 0 with two different rates to compare) as the headline it would print, and prints no median, no interval and no Kickback Score. The inputs below are published anyway, so the same computation can be run against them.

Disclosure

§1.5 is purely positional. The A/B boundary here is measured against the first OUTBOUND link, which is at or before the first AFFILIATE link, so this test is stricter than §1.5 and can only understate a grade, never overstate one.

Diagnostics published beside the figures

Count of qualifying articles with ρ > 0.99 at n ≤ 5.

Articles excluded from the pool, by reason

Reasons may co-occur on the same article, so these counts do not sum to a total.

Limitations that apply to this publisher

1. The Payout–Rank correlation is not computable for this publisher, and that is not a gap in our data. The correlation is computed within each article, and in every one of this publisher’s articles all the picks we can price carry the same commission rate — so there is no variation in payout for the ranking to be compared against. A rank correlation needs at least two different rates inside one article; this publisher’s articles supply one each. Collecting more articles of the same kind would not change that. The publisher may still pay different rates in different articles, which a within-article rank correlation does not measure.

Right of reply

State: silent. We wrote to this publisher and the seven-day window closed without a reply. Not replying is not agreement with anything on this card, and we do not present it as one.

Written to info@hostingstep.com on 2026-09-10T00:11:08Z. Basis for that address: published site contact Seven-day window closes 2026-09-17T00:11:08.000Z.

Nothing has been received from this publisher.

Confounds, stated by us, first

1. Reverse causality. A publisher may negotiate a higher rate because it already ranks a product well. The §5 event study partially addresses this by measuring ranking changes after rate changes. It does not eliminate it.
2. Category confounding. High-commission categories may differ in quality.
3. Brand-size confounding. Products with affiliate programs skew toward larger brands, which may be better on average.
4. Rate coverage. We observe roughly half the market. The rate coverage figure on this card is the measured share for this publisher.
5. Editorial firewalls are real. Where the median ρ is at or below zero, that is stated on this card in the same size type as every other number.
6. Private-network rate invisibility. Merchants on Impact, CJ, Rakuten and Partnerize keep rates private, so their picks carry r = null and are dropped from ρ. Those are systematically different merchants from Amazon’s — large retailers rather than marketplace sellers. The sign of the resulting bias is indeterminate.
7. Per-transaction earnings caps and negotiated tiers mean a published nominal rate is an upper-bound proxy for effective compensation. eBay’s published rate card carries a per-transaction cap on every category (550 USD on Electronics, 100 USD on eBay Motors and on Real Estate) and states that exceptions to the rate card exist for certain partners. Roborock publishes 4%–7% and Narwal 4%–10%, and where a given publisher’s tier falls within either band is negotiated privately. On a high-priced item a per-transaction cap can put the amount actually earned, expressed as a percentage of price, below the published rate. This is a disclosure: it moves no estimator and no r_i.

Recompute this

All publishers · Methodology