A journalist who wants to quote your poll gets about 10 seconds to decide whether she can. She clicks the link under the percentage. If the page on the other end shows a sample size, a recruitment source, a time window, and the full distribution before she scrolls, the number stays in her draft. If it shows a headline and a chart image, the number comes out. A self-selected poll of 412 readers with a public methodology page will beat an undisclosed survey of 10,000 in that pass every time, because the page gets screened before the percentage does.
The question you write is the smallest replaceable part of the whole apparatus.
What a Citable Poll Looks Like
Run a search for "state of [your industry] 2026" and open the top result. Odds are good you'll find a poll with no published N, no recruitment source, and no dates. There's a percentage, a headline, a screenshot of a chart. A reporter can't hand any of that to an editor.
A citation is a link, and the link has to land somewhere worth defending.
AAPOR is unusually plain about where the standard sits. Its Transparency Initiative treats the disclosure itself as the deliverable:
"AAPOR's Transparency Initiative is designed to promote methodological disclosure through a proactive, educational approach… In doing so, AAPOR makes no judgment about the approach, quality or rigor of the methods being disclosed." (AAPOR Transparency Initiative)
Pew hands reporters the mirror-image advice: "Lack of transparency is a red flag, as is polling from campaigns or advocacy groups that do not regularly release results" (Pew Research). What the fact-checker actually tests is reachability. Can the methods be reached at a URL, are they dated, do they sit in the same scroll as the number. It's the same screen that sits behind original research as a ranking signal: you can't pass it by paraphrasing whatever already ranks, and a reporter runs it faster than Google does, with a harder cut at the end.
Four Disclosures Every Citation Needs
AP Stylebook sets the cite-or-cut rule for most U.S. newsrooms. "Stories based on public opinion polls must include the basic information for an intelligent evaluation of the results," and it names the requirements: who paid for the poll, how many people were interviewed, who was interviewed and how they were selected, mode and timing, sampling error margins, and the questions asked in the order they were asked (AP Stylebook polling entry, public mirror).
You don't have to chase all seven. Four carry the rest. When I screen a poll in our own niche, a missing sample size is the thing that stops me before I ever reach the percentage.
Sample size. One integer per period. A range like "several hundred" looks like something is being hidden. Publish the number: N=412.
Recruitment source. Plain sentences. "Self-selected LiquiChart blog readers" tells the reporter exactly which universe the percentage describes, and which ones it doesn't.
Time window. A start date and an end date for every period. A lone publication date strands the citation the moment the calendar moves past it.
Raw distribution. Every option's percentage, published together, so the headline number sits next to the numbers it beat.
The trade press runs on zombie statistics: numbers everyone repeats, nobody can trace, each one having outlived whatever page once explained it. The four disclosures are what keep your number out of that pile, and what let a reporter defend it in an editorial meeting.
The second reader matters more than the first. "The best polls are transparent about their methods," Kristen Olson, director of the University of Nebraska-Lincoln Bureau of Sociological Research, told Journalist's Resource in a post about what reporters check when they vet citation provenance on deadline (Journalist's Resource). The screen detects transparency because transparency is what survives a fact-checker reading closer than the writer did.
One Question Per Poll
Quote a 12-question survey and every paragraph needs a caveat. Which question, in what order, after what routing, with what incentive. A single-question poll gives the reporter one number with one denominator and one window.
LiquiChart polls are single-question by design. The option set does the analytical work a survey would spread across follow-ups, which makes how you design and seed the options the real architectural decision. Wording is one input. The options, the recruitment line, and the period structure are the rest of what a reporter evaluates.
Adding a second question "for demographic context" costs more citation value than it buys. You hand the reporter a caveat she can't quote around, resting on a sample that was never built to support the cut.
Recruitment Disclosure Over Demographics
Publishing demographic breakdowns your sample can't defend is the fastest way to get cut from a draft. A self-selected sample has no probability frame. It can't produce a representative slice by age, role, or industry, however precise the percentages look.
AAPOR's standard for non-probability samples asks for a disclosure list: who sponsored the research, who conducted it, how the sample was created, how the data were weighted, what procedures ensured data quality (AAPOR). A demographic crosstab appears nowhere on it. Publish one anyway and you've minted a zombie statistic, because the crosstab keeps circulating long after the fact-check kills the page it came from.
AP goes further than AAPOR here. Its line on self-selected polls: "Polls based on submissions to Web sites or calls to 900-numbers may be good entertainment but have no validity. They should be avoided because the opinions come from people who select themselves to participate. If such unscientific pseudo-polls are reported for entertainment value, they must never be portrayed as accurately reflecting public opinion and their failings must be highlighted" (AP Stylebook polling entry, public mirror). AP's condition for reporting a self-selected poll at all is that its failings be highlighted, and a plain recruitment line on a public methodology page is what that highlighting looks like in practice.
"Self-selected LiquiChart blog readers, N=412, May 1-15 2026" is the whole disclosure. The permanent methodology page for the poll embedded below turns that sentence into a link a fact-checker can actually click.
Boring Questions Win Citations
The clever framing wins the engagement bracket. It screenshots well, it travels on LinkedIn, and none of that reaches the desk where citations get decided. An editor defends a citation by clicking through to the methodology page, and cleverness adds nothing on the far side of that click. Worse, the screenshot a clever poll earns is frozen: a chart is a claim about the world, and a screenshot is that claim with the date torn off, wrong the day the data moves and unable to correct itself.
I watched one of our cleverer polls pull LinkedIn shares for a week and never get cited once. Shares and citations pay out of different economies. Our experiment on whether polls move time-on-page points the same direction: the engagement spike expires, while the dataset and its methodology page are still quotable a quarter later.
The four answers below describe four different artifacts a reporter will eventually open, so name yours honestly.
The first three answers each put a number into the world under a different publishing setup, and the fourth stays out of the data layer entirely. None of the four produces a first-scroll methodology page on its own. Whichever one you picked decides whether a reporter ever finds that page, or thinks to look for one.
The chart below carries the same question across five reporting periods. A screenshot from any single period would mislead a reporter quoting it months later; the live chart keeps every period's number attached to its dates.
Publish the Full Distribution
The reporter quotes the winning percentage. Then the editor asks what the other 60% said. If you never published the full distribution, the reporter has to invent context to answer, and invented context is the first thing a fact-checker deletes.
Put the raw distribution on the methodology page in the same scroll as the sample size: four options, four percentages, summing to 100, with the period stamped above them. Our claim attribution study found the same screen deciding which claims survive review: the ones an editor can re-derive from the full denominator without leaving the page.
The lead percentage is what gets pasted. The editor reaches past it within 24 hours.
Build for the Six Month Recheck
A citation passes through three checkpoints. The reporter pastes the percentage on Tuesday. The editor verifies it on Friday. The fact-checker reruns it six months later, when the topic resurfaces. A static blog poll clears the first checkpoint and starts dying at the second.
Pew's May 2026 analysis of AI-era survey fraud raises what disclosure now has to carry: "Bogus respondents are survey-takers who make no effort to answer questions truthfully and instead try to complete surveys as quickly as possible to get monetary rewards… This tendency of bogus respondents has led to false conclusions and news organizations needing to retract stories based on opt-in polls" (Pew Research). In 2026 the methodology page is doing more work than it has ever done, and the page is where that work has to show.
Writing for the long cycle means building the surface the poll will be cited from, months after the surface it launched from has scrolled away. Tracking audience sentiment over time describes what that surface has to do: hold the original period's number intact while new periods accumulate, record every leader change as a dated event, and resolve to a URL that still reads as defensible when the editor reopens the tab.
Every answer above is also an answer to a question the publisher did not realize they were being asked: what surface holds the percentage three months after the reporter quotes it. The four options sort into two camps that behave nothing alike, snapshots that disclose a number once and tracking that keeps disclosing it, and a reporter's editor can only verify against the second. The publisher who treats a poll as a finished number and the publisher who treats it as a running one walk away with the same percentage and opposite odds of keeping the citation.
What an Uncitable Poll Costs
You will never see the cost. A fact-checker who pulls your number in October is deleting a citation that, from where you sit, was never visibly written. The reporter doesn't email to explain. The percentage rides in the draft for one revision, and then the link that should have pointed at your methodology page points at a competitor's, because theirs existed. The silent swap is the failure worth watching for most, and it stays invisible to anyone who never built a surface to check.
A poll question with no methodology page underneath it is a tweet with buttons.
Create a poll with the methodology page built in.
Someone in your niche is getting quoted this October. The editor settles one question: whose page sits under the percentage.