Discussion
  • AI Search
  • Technical SEO

Comparison sites paid by commission: how transparent do ranking methods need to be to hold up in search and in AI answers?

Published 1 reply

Discussion starter: a programmatic SEO question, posted with an answer from Niraj Raut to open the thread. If you have dealt with this on a site, reply with what you saw, especially where it differs.

Disclosure: Niraj Raut, who posted this answer, runs the SEO consultancy linked at the end of it.

1 reply

  • admin
    Admin
    Original poster

    No search engine or AI platform publishes a transparency threshold for commission-funded rankings, so the bar that actually binds you is consumer law, with Google’s review guidance sitting on top of it. In practice, a reader should be able to tell four things at the point where the ranking appears: that you are paid and by whom, how much of the market you compare, what determines the order, and whether payment changes the order or the placement. If commission influences the order, that has to be visible next to the list, not buried in a footer or an About page.

    The important distinction is that disclosing you earn commission is not the same as being transparent about how you rank. A site can carry a perfect affiliate disclosure and still mislead if the list is presented as “best” or “cheapest” while payment drives the order. For search, the stronger lever is not more disclosure text but visible evidence that you evaluated the products. For AI answers, there is no documented rule at all, which makes the legal and trust basics more important, not less.

    Three things that get bundled together as “transparency”

    Disclosure

    Telling readers you have commercial relationships. Necessary everywhere, and the part most sites already do, often only in a footer.

    Methodology

    Explaining the criteria, their rough weights, the data behind them and how much of the market you cover. This is what lets a reader check your ranking.

    Independence

    Whether payment changes inclusion, order or placement. This is where regulators focus, and where most real risk sits.

    Most comparison sites are strongest on the first and weakest on the third. The problems start when the page’s framing (“the best”, “cheapest”, “we compare the market”) implies independence that the ranking logic does not deliver.

    The legal floor, and why disclosure alone did not save Trivago

    The clearest case study for comparison publishers is Australian. In April 2022 the Federal Court ordered Trivago to pay $44.7 million in penalties after finding its algorithm prioritised offers from booking sites paying the highest cost-per-click fees, while Trivago represented that it would help users identify the best deal or cheapest rates. According to the ACCC, a higher-priced room rate was selected as the top position offer over cheaper alternatives in 66.8% of listings. The ACCC chair said the penalty “sends a strong message not just to Trivago, but to other comparison websites, that they must not mislead consumers when making recommendations”.

    The lesson is not “add a disclosure”. It is that the ranking must match the claim the page makes about it. If your default order is weighted by commission, you cannot frame it as price-led or quality-led.

    Other markets approach the same problem from different directions. EU consumer law, as amended by the Omnibus Directive, requires platforms that let consumers search across traders to disclose the main parameters that determine ranking and their relative importance, and treats undisclosed paid ranking as an unfair practice. The UK’s Digital Markets, Competition and Consumers Act gave the CMA direct fining powers for consumer law breaches from April 2025. In the US, the FTC’s Endorsement Guides require clear disclosure of material connections. None of this is legal advice, and finance, insurance and energy comparison often carries extra sector rules, so a lawyer in each target market should sign off the methodology wording.

    Watch for this:

    “Featured” or “Recommended” as the default sort, with partners pinned to the top and non-partners either missing or pushed down, is the pattern that combines weak disclosure with an independence problem. If you keep a commission-influenced sort, label it as such on the sort control and offer a neutral default.

    What Google’s guidance rewards on the search side

    Google’s review guidance asks for evidence of first-hand evaluation, an explanation of what sets each product apart, benefits and drawbacks, and a reason with supporting evidence whenever you call something “best” for a purpose. It also suggests considering links to multiple sellers so readers can choose where to buy. Google’s people-first content guidance frames this as explaining “how” content was made: for product reviews, how many products were tested, what the results were and how the tests were run. Paid and affiliate links should carry rel="sponsored" (or nofollow), and Google’s thin affiliation spam policy covers pages that add nothing beyond merchant descriptions.

    Notice what this guidance does not ask for: commission rates or a scoring formula. Google’s documented position is about demonstrated evaluation work. A methodology page helps in search mainly because it forces the evaluation work to exist and be visible, not because a methodology URL is a ranking signal. No public evidence shows that a methodology page by itself changes rankings.

    Two further points apply to specific setups. If your comparison section lives on a third-party publisher’s domain, the site reputation abuse policy is relevant. Google’s August 2026 update lists factors such as presentation consistency, quality disparity, authorship clarity and duplication across sites, and it changed how enforcement works in the EEA. Separately, in early 2026 several practitioners reported visibility losses on self-promotional “best of” lists where brands rank themselves first, and industry coverage asked whether Google was targeting that pattern. That remains a field observation, not documented policy, and it concerns vendors ranking themselves rather than third-party comparison sites. The overlap is that both are rankings whose order serves the publisher.

    What changes when AI systems summarise your rankings

    Google’s documentation says there are no additional requirements to appear in AI Overviews or AI Mode beyond being indexed and snippet-eligible, and its May 2026 AI guide says seeking inauthentic mentions is not helpful. No AI platform documents how it treats affiliate disclosures or ranking methodologies. What follows is reasoning from how these systems work, not a documented mechanism.

    • Your ranking gets separated from your disclosure. A synthesised answer can cite your page for “X is the best option” without carrying over the partner labels or the market-coverage caveat. Your brand ends up attached to a claim stripped of its context. The fix is to put the qualification inside the claim itself: “Of the 14 lenders we compare, X had the lowest fees in our September review” survives extraction. “X is the best lender” plus a disclosure three screens away does not.
    • Rankings that diverge from other sources may be echoed less. Answers draw on several sources. If your top pick is one that independent reviews rate poorly, a synthesised answer has less reason to repeat it. This is plausible but not measured publicly.
    • Dated, factual data is more reusable than opinion. Fees, features, eligibility rules and “last checked” dates are the parts of comparison pages most likely to be quoted. Stale data then gets repeated with your name on it, which is a brand and compliance risk, not just an SEO one.

    Levels of transparency and where I would aim

    Level What is published Legal exposure Search and AI position
    1 Generic affiliate disclosure in the footer High if the page claims “best” or “cheapest” Hard to distinguish from thin affiliate pages
    2 Disclosure next to the list, plus a list of criteria Moderate: the reader still cannot tell whether payment affects order Better, but claims remain unverifiable
    3 Criteria with rough weights, market coverage (how many providers, how many are partners), paid slots visually separated, data dates Low in most markets, subject to local advice Strong: claims are checkable and survive extraction
    4 Level 3 plus per-product scores, testing evidence, excluded products and why, and a methodology change log Lowest Strongest, but costly to maintain accurately

    I would treat Level 3 as the minimum for any ranked list and Level 4 for money pages in finance, insurance, energy and health, where both regulators and Google’s quality raters apply a higher standard. Practitioners reasonably differ on Level 4 for low-stakes categories: a 50-product kitchen gadget roundup may not justify per-product scoring, whereas a home loan table almost certainly does.

    Hypothetical example:

    An Australian broadband comparison site lists plans from 40 providers, 12 of them commercial partners. Its “Best NBN plans” page defaults to a “Featured” sort that pins partner plans to the top, and non-partner plans appear only if a user scrolls and expands the list.

    A defensible version changes the default sort to monthly price for the selected speed tier across all 40 providers, shows a “Go to site” button for partner plans and a plain “Visit provider” link for the others, moves paid placements into one labelled “Sponsored” slot above the table, and states under the heading: “We compare plans from 40 providers. 12 pay us a commission when you sign up. Payment does not affect the order below.” The page title also stops claiming “Best” unless the methodology behind “best” is published.

    The commercial trade-off is real: revenue per visitor on the default sort may fall. That is a business decision to measure, but it is not one you can trade against legal compliance.

    What to check and how to measure it

    • Does every ranked list carry a one-line disclosure and market-coverage statement directly above or beside it?
    • Are paid placements visually and structurally separate from the ranked list?
    • Can a sceptical reader reproduce your top three from the published criteria and data?
    • Does any page claim “best”, “cheapest” or “whole of market” that the method does not support?
    • Are affiliate links marked rel="sponsored", and are data “last checked” dates visible?
    • When you ask AI assistants “how does [your brand] make money” or “is [your brand] independent”, is the description accurate?

    The on-page changes can be tested properly on the conversion side, because users can be randomised: compare click-out rate and revenue per visitor between the current and transparent versions. The search and AI effects are harder to isolate. Use a before and after comparison on a group of ranking pages against untouched ones, Search Console’s generative AI impressions by page, and a fixed set of prompts run repeatedly, noting whether you are cited and whether the answer’s recommendation matches yours. If AI answers cite you but recommend differently, that is a signal your ranking reads as out of line with other sources.

    A practical standard to hold yourself to

    Ask two questions of every ranked page. Could a sceptical reader, or a regulator, reproduce your top three from what you publish? If commission influences the order or placement, is that visible at the point of the ranking? If both answers are yes, you are as transparent as anything documented for search or AI answers requires. Beyond that point, further gains come from evidence of real evaluation, not from longer disclosure text.

    Need help with this on your own site?

    Niraj Raut works with businesses in Nepal, Australia, the UK, Europe and the US on technical SEO, ecommerce SEO, local SEO and AI search.

    See SEO services

Add a reply

Replies are open to approved members. Log in or apply to join.