We open every source ourselves
A person or one of our agents opens each link before it goes on a page. The link has to lead to the exact article or report, not a homepage or a list of articles. It has to carry a date. It has to say what we claim it says.
If a link fails any of those tests, we don’t use it.
We name who produced the numbers
A survey run by a company that sells something is still a survey, and we will cite it. We tell you who ran it, so you can weigh it yourself.
We keep a dated copy
When we check a source, we save a copy of the page and record the date and time. We also ask the Internet Archive’s Wayback Machine to save a public copy. If the original page later changes or disappears, the record of what it said on that day still exists.
Our saved copies are for our records. We don’t republish other people’s articles.
We check whether a site allows archiving
Some sites ask not to be archived. Before we request a public copy, we look in three places: the site’s own instructions to crawlers, the page itself, and the public lists of sites that have been excluded from the Wayback Machine. Those lists change often, so we check them again every time we run our source checks, not once.
If a site has opted out, we don’t request a public copy. If a site is excluded after we have published, the archived link stops working. We remove that link from the answer and say that no public copy is available.
What you see on each answer
- The source named in the sentence.
- A link to the exact article.
- A link to the archived copy, when one exists.
- The date we last checked it.
Sources behind a paywall
We look for the free original first. Most paywalled articles report on a survey, a press release or a government figure that anyone can read, and that is what we cite.
When a paywalled article is the only source, someone with a subscription reads it and confirms the claim. We link to the article and mark it “subscription required.” We never use a paywalled article as the only support for an answer, and we never go around a paywall.
How we use predictions
We follow public prediction sites. Some are prediction markets, where people trade on how a future event will turn out. Others are archives that track forecasts made by analysts and public figures. When a prediction bears on a topic we cover, we log it with its date, its source and the date it can be checked.
When we cite one, we say whose prediction it is. A market price is what traders thought on that day. It is not a forecast of ours. We read these sites and we don’t trade on them.
We also log our own predictions with a date, and we report how they turned out, right or wrong.
When a link breaks
We re-check our links on a schedule. If a source has moved, we update the link. If it has changed or gone, we say so on the answer, point to the archived copy, and review whether the answer still stands.
When we get it wrong
If you find a source on this site that doesn’t support what we wrote, tell us at team@salescitation.com. We will check it, fix it, and note the change on the page.