Answer in brief
Before sharing a viral post, stop and ask whether you know the source and where the claim began. Look up who published it, search for better coverage of the same claim, and trace the quote, picture or video to its first appearance, checking the date on the way. If a few minutes of searching cannot confirm it, do not pass it on as fact.
A viral post reaches you with its context cut away
What reaches you on a busy day is rarely a whole story. It is a headline without the article beneath it, a clip without the minutes before it, a screenshot without the page it was cut from. The fragment travels because it is striking, and by the time it lands in your feed or your group chat you can no longer see who made it, when, or for whom.
Passing it on makes you its latest source. For a business owner that is not an abstract risk: a post from the company account carries the company’s name, and a correction published the next day has to find the same readers on its own. UNESCO’s page on media and information literacy lists two figures: 56% of internet users frequently use social media to stay informed about current events (UNESCO/Ipsos, 2023), and two in three digital content creators do not systematically fact-check information before sharing it online (UNESCO, 2024).
Those figures explain why the size of an account says little about whether anyone checked the claim. They also point to the remedy, which is small. You do not need a newsroom to fact-check a viral headline. You need a short routine that restores the missing context, and one already exists: four moves, two quick technical checks and a rule for the cases that cannot be settled.
SIFT: four moves a digital literacy teacher wrote down in 2019
The routine comes from Mike Caulfield, an educator who works on online digital literacy. On 19 June 2019 he published a short text on his blog Hapgood called “SIFT (The Four Moves)”, noting that he and his colleagues had been teaching the approach to students since 2017. The name is an acronym: Stop, Investigate the source, Find better coverage, Trace claims, quotes and media back to the original context.
Its design is deliberately modest. Caulfield describes it as a short list of things to do when looking at a source, each tied to one or two effective web techniques. Notice what the list leaves out: it does not ask you to study the page itself for signs of quality, since a polished layout and a confident tone are easy to produce. Every move sends you somewhere else to recover context: who is speaking, what others report, where the material began.
The method also scales to the purpose. Caulfield writes that quick and shallow investigations make up most of what people do on the web, and that if you only want to repost something it is probably good enough to find out whether the publication is reputable. Deep checking of individual claims is for the things that really matter to you. What follows is the shallow version: a filter to run before pressing share, not an investigation.
Stop first: the urge to share is the signal to check
The first move costs nothing and is the one most often skipped. When you meet a post, Caulfield’s instruction is to ask whether you know the website or source of the information, and what the reputation of both the claim and the site is. If you do not know, his advice is blunt: do not read it or share it until you know what it is.
The moment to apply it is the moment you least want to. Material that makes you angry, delighted or proved right is the material your thumb moves fastest on. The International Federation of Library Associations and Institutions (IFLA) has an eight-step infographic, “How to Spot Fake News”, based on a 2016 article by FactCheck.org, and one of its lines makes the point: check your biases, and consider whether your own beliefs could affect your judgement. A strong reaction is not evidence that a claim is false. It is a reason to spend a minute before passing it on.
Stop has a second meaning that keeps the routine short. Checking can turn into an afternoon of tangents, so Caulfield advises pausing to remember your purpose whenever you feel overwhelmed. Decide in advance how much time a share is worth to you. A few minutes is a sensible ceiling for an ordinary post, and reaching it without an answer is itself a result.
Leave the page to find out who is speaking
The second move is to investigate the source, which means knowing what you are reading before you read it. Caulfield’s examples are deliberately balanced: an article on economics by a Nobel prize-winning economist, and a video on the benefits of milk produced by the dairy industry. Neither label settles whether the content is right, but both change how you should weigh it. He puts the cost at about sixty seconds.
The technique is to leave. One of IFLA’s eight steps says to click away from the story and investigate the site, its mission and its contact details; another says to run a quick search on the author to see whether they are credible and whether they are real. Open a new tab, type the name of the outlet or the account, and read what others say about it. What a site says about itself tells you least.
With screenshots and forwarded messages the source is not the friend who sent them, and usually not the account that posted them. It is whoever first made the claim. Look inside the image for a name, a logo, a handle or a date, and search for those. If nothing identifies an origin, the item is unsourced, however many people have passed it along. IFLA adds one more question for this stage: a story that is too outlandish might be satire, so check the site and the author before taking it literally.
Judge the claim by looking for better coverage
Often the source hardly matters, because what you care about is the claim. Caulfield’s third move covers this case: set aside the item that reached you and look for trusted reporting or analysis of the same claim. His illustration is an article announcing that koalas have just been declared extinct. The quickest route to the truth is not to study the organisation that published it but to find the best available source on koalas, or to scan several and see what the experts agree on.
In practice, type the core of the claim into a search engine in plain words and read the results page as a whole before opening anything. Check who else is reporting it, whether the reports all trace back to a single origin, and whether careful outlets are hedging. Then open the strongest one and read past the headline. IFLA’s steps say the same: headlines can be outrageous in an effort to get clicks, so read the whole story and follow the links to supporting sources to see whether they actually support it.
A specialised search can shorten this step. Google describes its Fact Check Explorer as a search engine for fact checks: according to the company’s training page, you type in a keyword or upload an image, and the results name the organisation that carried out each check and how it rated the claim, with a link to the full text. Read the check itself rather than the one-word rating. An empty result means only that no listed organisation has published on the claim, not that the claim is true.
Trace the quote, the photo and the clip back to the start
The fourth move deals with material that is genuine but cut loose. Caulfield lists the typical cases: a video of a fight that does not show what happened before it, a real photograph with a misleading caption, a claim about a new medical treatment resting on a study that may not say what the post says. The answer in each case is to follow the item back to where it first appeared and look at it whole.
For pictures the tool is a reverse image search. Instead of words you give a search engine the image itself, and it returns pages where the same or a similar picture appears. Look for the earliest appearance and the fullest caption. For video, pause on a distinctive frame, save it as a screenshot and search with that. Fact Check Explorer accepts an uploaded image as well, which shows whether fact-checkers have already examined it.
For quotations, put the exact words in quotation marks, add the speaker’s name and look for the full speech, interview or document. A quote that exists only on image cards, with no recording or transcript behind it, should be treated as unverified. For numbers, find the report they come from and read the sentence around the figure, because a statistic can lose its unit, its year or its qualifier on the way to a headline.
An old story reposted today is still an old story
Dates deserve a check of their own, because a story can be accurate and still mislead. IFLA’s wording is that reposting old news stories does not mean they are relevant to current events. A real report about a storm, a product recall or a price rise, recirculated years later without its date, reads as though it were happening now. Nothing in it is false, and the people who share it are wrong all the same.
Look for the date in several places: the publication line of the article and any note saying it was updated, the web address, and the upload date of a video, which is not always the date of the events it shows. A screenshot may leave the date outside the frame, which is one more reason to find the page it was taken from. If a search shows the same story circulating months or years earlier, you have your answer.
The same logic settles many arguments about images. A photograph cannot be older than the event it is said to show. If a reverse image search finds the picture on a page published before that event, the caption is wrong, whatever the picture actually depicts. This comparison of two dates needs no knowledge of the subject, and it is the quickest test to run on a dramatic picture.
When the check runs out: wait, ask or share nothing
Some items cannot be settled in a few minutes, and that is a normal outcome, not a failure. During a developing event the first reports are incomplete and sometimes wrong, and no amount of searching will produce coverage that does not yet exist. The rule for these cases follows from the first move: an item with no identifiable source, no better coverage and no traceable original is unverified, and what is unverified is not shared as fact.
You have three honest options. You can wait, since a few hours often bring the reporting that was missing. You can ask: one of IFLA’s steps is to ask the experts, meaning a librarian or a fact-checking site, and a private message to the person who sent you the item, asking where it came from, costs less than a public correction later. Or you can let it pass. If your work obliges you to respond to a rumour, say what is known and what is not, and link to the best source instead of the viral post.
If you have already shared something that turns out to be false, correct it where you posted it, in plain words. Remove or edit the original with a note saying what changed, and tell the people you sent it to directly. In a team, write the routine down: who may publish, which checks come first, and who decides when a case is unclear. A habit that takes a minute is easier to keep than a reputation is to repair.
Practical checklist
- Pause before reacting and ask two questions: do I know this source, and do I know where this claim started?
- Open a new tab and search for the name of the site, account or author to read what others say about them.
- Search for the claim itself in plain words and compare at least two independent reports before trusting the post you received.
- Run the picture, or one frame of the video, through a reverse image search and note the earliest date on which it appears.
- Find the publication date of the original item, and if nothing is confirmed within a few minutes, do not share it as fact.
Questions and answers
How long does it take to check a viral post before sharing it?
The routine is built to take a few minutes, because one search will often show either solid coverage or none. Mike Caulfield, who devised the SIFT method, writes that sixty seconds spent finding out where an item comes from helps you decide whether it is worth your time. Give yourself a fixed limit, and treat anything still unclear at the end of it as unverified.
What do the four letters of SIFT stand for?
Stop, Investigate the source, Find better coverage, and Trace claims, quotes and media back to the original context. Mike Caulfield published the acronym in 2019 as a short list of actions, each linked to a simple web technique. For a simple repost, he suggests, finding out whether the publication is reputable is probably enough.
How can I tell whether a viral photo or video is genuine?
Run the picture, or a screenshot of one frame of the video, through a reverse image search and look for its earliest appearance. A genuine picture can also mislead when the caption gives it the wrong place or date, so compare the date of the oldest copy with the event it is said to show. If the image was online before the event, the caption is false.
Is it acceptable to share a post with a note that it may be untrue?
It is better not to. Your caveat is attached to your post, not to the claim: a forward, a quotation or a screenshot can drop your words and keep the headline. If the subject matters to your audience, describe in your own words what is known and what is not, and link to the most reliable source you found, leaving the viral item out.
What should I do after sharing something that proved to be false?
Correct it in the place where you posted it, in clear language and as soon as you know. Edit or remove the original with a short note explaining the change, so that people who saw the first version understand why it has gone. If you sent it to particular people or groups, tell them directly and say where the accurate information can be found.