What does this site earn, and from whom?

Nothing, from nobody. As of 2026-08-06 there are zero affiliate links on this site, zero display advertising units, zero sponsored posts, zero paid placements in the Ledger, and no page carries a commercial interest of any kind. The rest of this page is the arithmetic behind that, because “we take no money” is a sentence anyone can type and the numbers are what make it checkable.

§1The high commission rate is not the tell, and repeating that it is would be a lie we benefit from

Costs in commission rate

~0 — 60–75% is the channel norm, not a premium

Costs in conversion

the entire difference; a scan converts, an explanation does not

Costs to the reader

0 — the recommended tool is usually already installed

The convenient version of this page would say that PC optimisers pay several times the software norm, that a 70% commission is the signature of a bad product, and that Working Set therefore refuses an unusually large pile of money. That version is false in every part, and we held it ourselves until a research pass in 2026 checked it against the vendors’ own published rates.

Here are those rates, read on 2026-08-06 on each programme’s own page, with the URL for every one of them numbered in §7. The right-hand column is our own classification of the subject matter, labelled as ours.

The third column says what a programme sells, not what we think of it. It marks which vendors sit inside the subject this publication examines and which do not, and it is not a verdict on any product named in the table. This page carries no per-product finding about any of these companies, and none should be read out of the rows: verdicts are issued one claim–surface pair at a time in the Ledger, against a named registry key, API or file, with the evidence attached to the row.

Read the first three rows against the next three. Backup and partition vendors — software nobody argues about — pay 60 to 70%, which is the same band as the optimisers. The reason is the sales channel, not the product: low-cost electronic software delivery sold through resellers such as Cleverbridge, Avangate/2Checkout and FastSpring runs at those margins, because the marginal cost of a licence key is zero and the affiliate is doing the acquisition. A 70% commission describes how the software is sold. It does not describe whether it works.

The row that finishes the argument is CCleaner at up to 35% — a below-average rate published by the single most recognised brand in this entire category. If commission rate were the signal, that number would point the wrong way.

So the honest accounting is this: refusing to promote a free-scan-to-paid-fix funnel costs this site almost nothing in commission rate. The rate on the products we would be willing to name, in the rare cases where a paid product is the right answer, is in the same band as the rate on the ones we would not. Anyone claiming otherwise — including anyone claiming it on our behalf — is repeating a tidy falsehood, and it is a falsehood that happens to flatter us, which is exactly the kind we are obliged to check hardest.

What separates a product worth refusing from an ordinary one is the funnel, and the funnel is a pattern rather than a company: a free scan that reports a number, a diagnosis the reader cannot falsify, and a paid remedy for a problem stated in units nobody defines. That pattern, not the percentage, is what the Ledger is built to make legible, one claim–surface pair at a time. Where a regulator has acted on it, the order is named and dated on the page that cites it, and where we could not read the order ourselves we say so instead of gesturing at it.

§2What refusing the money actually costs is conversion, not rate

Costs in rate

~0 percentage points against a defensible vendor

Costs in conversion

a scan result converts a frightened reader; a mechanism does not

Costs in traffic

not established — nothing published yet, no traffic to measure

Here is the real economics, and it is a more interesting thing to say than the version it replaces. Scareware converts a frightened user far better than honest advice converts a curious one. A page that tells a reader their machine has 2,317 errors and offers a fix for $49.95 is selling relief from a state it created thirty seconds earlier. A page that explains what the registry is, what a cleaner is counting, and why the count is arbitrary sends most of its readers away reassured and buying nothing. Both pages can pay the same 70%. Only one of them triggers a purchase.

The category is built on that asymmetry, and the numbers are published by the people running it rather than inferred by us. RealDefense, which operates the affiliate programme behind several consumer PC-utility brands, advertises to prospective affiliates that “Top partners earn $100,000+ per month” at a “~6% average conversion rate” and “$5+ EPC”, read on its own affiliates page on 2026-08-06. A 6% conversion rate on a consumer software purchase is not what an explanation converts at. This is why the supply side of every question this site answers is dominated by ranked-roundup pages and by AI-generated filler — the funnel pays for the content, and the content exists to feed the funnel.

We stop at the published figure rather than going one step further. An earlier draft of this paragraph carried an enforcement claim about the firms that run these funnels. It is not on the page, because ftc.gov press pages returned HTTP 403 to our fetcher on 2026-08-06, and an enforcement claim whose order we have not read is not one we publish, however well it would have finished the paragraph. Where this site does cite such an order it names it and dates it; see the standing preamble to the Ledger.

None of that is available to a publication whose commonest verdicts are NO-OP, DEPENDS and UNVERIFIED. Uncertainty does not convert. That is a cost, we are paying it deliberately, and it is the only cost on this page we cannot put a number on.

And the answer is very often a free tool that pays us nothing. Autoruns and the rest of Sysinternals, services.msc, Resource Monitor, Performance Monitor, Task Manager, tasklist /svc, DISM — every instrument this site recommends is either shipped with Windows or published free by Microsoft, and none of them can be monetised by anybody. Measure it yourself is a tool page with no downloads on it at all. That is the editorial point and it is also the structural reason this site never needs an affiliate link to a utility.

§3What is permanently disqualified, and it is not a close call

Costs to accept one

the Ledger, which is the only asset here that compounds

Costs to disclose it instead

not available — no disclosure repairs this category

Costs to refuse

0 known offers as of 2026-08-06

Affiliate revenue from PC utilities, cleaners, optimisers, driver updaters, registry tools, “activation” tools or consumer security software is permanently excluded. This site’s subject is wrapped installers, coercive upsells and unfalsifiable meters; a commission from that category would be disqualifying on its face and would poison the Ledger. There is no amount of disclosure that fixes it, which is why it is a prohibition rather than a disclosure obligation.

Also permanently excluded: sponsored reviews. Vendor-supplied review copies — we could not accept one without acquiring the first-hand experience claim this site is built not to make. “Partner” pages. Selling links to third parties. A sponsored Ledger row, a paid verdict change, or a “featured tool” slot. And a consulting funnel, which sells the same promise to fix a slow PC that this publication exists to examine.

§4Why there is no display advertising either, in numbers

Costs in served impressions

30–45% of desktop sessions serve nothing (modelled)

Costs in inventory

our working vocabulary is the brand-safety blocklist

Costs recovered

~$2.50 per 1,000 pageviews, published, on the one network that publishes one

Four structural drags. Three of them carry a published number and the fourth is labelled as modelling, which is the difference this page is obliged to keep.

The demographic is the worst possible one for display. Ad blocking runs at 32.5% of US internet users overall, on GWI’s Q2 2025 estimate. The device split is older and worse-attested: AudienceProject’s 2020 survey put desktop at 37% against 15% on mobile, and no newer survey of that split appears to have been published, while Blockthrough’s more recent device estimate is a narrower 27% on desktop against 22% elsewhere. This audience is desktop-heavy and technically literate, so the true figure for it sits above whichever of those is the general population number. We do not have a measurement for this site and cannot have one before it is serving; the 30–45% in the cost line above is modelling from those two surveys and is marked as such.

The working vocabulary of this site is the brand-safety blocklist. Malware, virus, crash, infected, hijack — these are the words the articles need. Integral Ad Science and Reuters published a joint test in which 54% of the URLs in the Reuters news section that contextual targeting had cleared as brand-suitable would have been demonetised by a typical keyword blocklist, against 4.27% of the lifestyle section. A wire service’s brand-safe news reporting loses more than half of its inventory to the mechanism; a site whose subject is those words has no reason to expect better.

The advertisers who most want this audience are the vendors this site writes against. Google’s own AdSense documentation states the ceiling: “You can block a maximum of 500 URLs.” These vendors rotate domains faster than a list of that size can track. The likeliest outcome of running display here is a registry-cleaner advertisement served on a page explaining why registry cleaners are a NO-OP.

The money is small. The two ad networks built specifically for technical readers are the honest comparison, and they sit on opposite sides of the transaction. EthicalAds tells publishers to expect “around $2.50 per 1,000 pageviews”. Carbon Ads publishes an advertiser-side rate card of $6 CPM for its developer audience and $5 CPM for its creator audience — what the buyer pays, before the network’s share, and on invited inventory rather than open signup. Modelled, not measured: $2–$8 session RPM on AdSense alone is our own arithmetic from those rates and the blocking figures above, it rests on a traffic number this site does not have, and it is not a measurement of anything. There is a structural problem on top of the rate in any case: an ads.txt file publishes a publisher ID at the domain root, in public, for anyone to read.

Adding it up, running display advertising here would trade the site’s entire reason for existing for a few dollars per thousand sessions. Not at launch, and on present numbers not ever.

§5If an affiliate link ever appears here, this is exactly what it will look like

Costs to qualify

a verdict of HARDWARE, reached before any link exists

Costs in placement

at most 1 link, on at most 2 articles, in the concluding paragraph

Costs if any condition fails

the article ships with no link

There is one place on this site where buying something is the honest answer. Articles about memory, boot and disk sometimes conclude HARDWARE: the bottleneck is a mechanical disk or four gigabytes of RAM, and no setting fixes it. If this site is ever monetised, that paragraph is the only candidate, no earlier than six months after launch, under all of the following conditions, every one of them binding.

The disclosure sits immediately above the link, in the same visual block, in body-copy size, in running text. Not in the footer. Not as a link to this page. Not on the about page only. Not behind a hover, an expander or a click, because a disclosure that requires an action to reveal is not clear and conspicuous. The wording is the plain one — that we receive a commission on purchases made through links in the post — placed where a reader reaches it before the link and cannot proceed past it without reading it.

At most one contextual link, on at most two articles, in the paragraph that concludes HARDWARE. Never in a roundup, never in a sidebar, never in a table of ranked products, never inside the Ledger. No experience claim attaches to it: we would link a category or a specification, cite manufacturer specifications and named third-party measurements, and state that we have not tested the parts. No wrapper, no shortener, no redirect and no intermediate hop. And if any single condition cannot be met, the article ships with no link at all.

Any page on which we hold or acquire a commercial interest of any kind carries a labelled box stating it, placed after the evidence and before the verdict — never at the top, never hidden. As of 2026-08-06 that box appears on exactly zero pages:

§6How this site links out, which is a monetisation policy wearing a different hat

Costs per outbound link

0 hops — direct to the primary source

Costs in hosted binaries

0 bytes, permanently

Costs in wrappers and shorteners

0 — every link is the real address

Primary sources and vendor pages are linked directly, never through a wrapper, a mirror, an affiliate redirect or a shortener. Download-catalogue listings are cited by name and archived URL as evidence of how a product propagated, marked as such, and are never presented as a place to obtain software. We link no binary, host no binary and mirror no binary, and no file whose bytes are executable exists in any deploy of this site.

Every outbound link points at the address it appears to point at, and you can read that address in the status bar before you click it. There are no interstitials and nothing counts a click on the way through.

Ranked, if revenue is ever genuinely required: leave the site unmonetised and treat the Ledger as what it actually is, the thing this publication is for; then newsletter sponsorship, the one channel ad blocking and keyword blocklists do not touch; then the single hardware slot described above. Nothing else.

§7Sources & method: every number on this page, with the page it came from

Costs in fetches

15 sources, read 2026-08-06

Costs in numbers modelled rather than measured

2, both labelled in the body

Costs in claims deleted for want of a source

3

Method, plainly: every affiliate rate in §1 was read on 2026-08-06 from the programme’s own page, in the order the table gives them, and each URL is below. Those URLs are plain addresses cited as evidence of a published rate; none of them carries an affiliate identifier, and following one earns this site nothing. Nothing on this page was measured by us. This site has served no traffic, so it has no session RPM, no impression count and no block rate of its own, and the two places where a number is our own arithmetic rather than somebody’s published figure say so in the sentence that carries them.

Three claims that stood on an earlier draft of this page have been removed rather than sourced. A commission band for Bitdefender, which publishes none we could read at its affiliate address. A rate card of $0.50–$1.10 CPM attributed to the developer-audience networks, which the published card in source 15 contradicts. And an enforcement claim about the firms that run the free-scan funnels, which is the open question below.

Open question, 1. Whether the specific enforcement history commonly attached to firms in this category is what it is reported to be. It is not asserted anywhere above, because ftc.gov press pages returned HTTP 403 to our fetcher on 2026-08-06 and an enforcement claim we have not read the order for is not one we publish. The evidence that would close it is the order itself, with its file number and date, read by us. If you have it, the address is the contact form.

1. MiniTool Software Affiliate Policy — “up to 70% commission on referrals”, tiered by monthly volume; “90 Days Cookie Period”. Accessed 2026-08-06
2. AOMEI Affiliate — “AOMEI standard affiliate commission is 30%. We are willing to increase your commission to 70% if you do proactively marketing.” Accessed 2026-08-06
3. EaseUS Affiliate Program — “up to 60% commission rate, plus extra bonuses”; attribution “last click, up to 90 days”. Accessed 2026-08-06
4. Auslogics — Partners, Affiliates — 70% commission, across a product list including BoostSpeed, Registry Cleaner and Driver Updater. Accessed 2026-08-06
5. IObit Affiliate Program — “Start at 60% and earn up to 75% commission rate”; “120 days’ cookie tracking”. Accessed 2026-08-06
6. CCleaner Affiliate Program — “up to 35% commission on all sales of consumer and business products”; “The referral window for CJ is 45 days.” Accessed 2026-08-06
7. iolo — Partners — read in full; states “industry-leading commission rates” and publishes no percentage, which is the basis for the publishes no rate cell. Accessed 2026-08-06
8. RealDefense — Affiliates — “Top partners earn $100,000+ per month”, “~6% average conversion rate”, “$5+ EPC”. Accessed 2026-08-06
9. Malwarebytes Affiliate Program — “Up to 30% commission for each sale”; “45-day referral period”. Accessed 2026-08-06
10. ESET — Join Our Affiliate Programme (UK store) — “up to 30% commission on a wide range of products”; 45-day cookie window. Accessed 2026-08-06
11. Backlinko, Ad Blocker Usage and Demographic Statistics — the 32.5% US figure, attributed there to GWI for Q2 2025; the desktop 37% / mobile 15% split, attributed to AudienceProject’s 2020 survey with the note that no newer data exists for it; and Blockthrough’s 27% desktop / 22% other-device estimate. Accessed 2026-08-06
12. AdExchanger, Keyword Blocking Demonetized More Than Half Of Reuters’ Brand-Safe Stories, 2026-04-06 — the Integral Ad Science and Reuters test; 54% of contextually cleared news-section URLs, 4.27% of lifestyle. Accessed 2026-08-06
13. Google, Block advertiser URLs in AdSense — “You can block a maximum of 500 URLs.” Accessed 2026-08-06
14. EthicalAds — Publishers — “Generally publishers see around $2.50 per 1,000 pageviews (CPM), assuming most of your traffic is from the EU & North America.” Accessed 2026-08-06
15. Carbon Ads — Media Kit: Pricing & Targeting — $6 CPM developer audience, $5 CPM creator audience; advertiser-side pricing. Accessed 2026-08-06