If you’ve built an FAQ page and still see the same handful of questions arriving in your inbox, chat widget, and contact form, you’re in familiar company.
The page is there. The accordions work. The questions are dutifully grouped by category, under ten of them, the way most guides recommend. And visitors keep asking anyway.
One thing worth checking is whether your answers only work while they’re sitting on your page. Pulled into a chat reply, a People Also Ask box, or an AI Overview, an answer can fall apart, because it quietly assumed the reader had already read three other things first.
To get a sense of how often that happens, I audited 83 FAQ answers across real WordPress sites in two industries. Small sample, found through search, so treat the numbers as directional.
Below is the three-axis test I used to check every one of those answers, with real before-and-after rewrites and a template you can adapt for your own page.
It should give you a quick way to look at your own warranty or pricing answer and see whether it holds up on its own, and which part to fix if it doesn’t.
How to Write an FAQ Answer: The Short Version
An FAQ answer works if a stranger can read it alone, with no page around it, and walk away done. Learning how to write FAQ answers that hold up is mostly about making each one self-contained.
Run every answer you have through three quick checks:
- Does the answer arrive in the first two sentences? If the reader has to get to sentence four before they find out what it costs or how long it takes, it fails.
- Does it depend on something outside itself? “Follow this link,” “see above,” “the details are in your contract.” Each one leans on something the reader may not have in front of them, so it fails.
- Does it say anything specific at all? “High-grade materials that meet industry standards” tells nobody anything, and trimming it wouldn’t help. Short isn’t the same as specific. That fails too.
Here’s why this matters more than it used to. Your answers already get read away from your page, in chat replies, People Also Ask boxes, and AI Overviews, and an answer that fails one of those three checks breaks the moment it travels.
Coming up: what I found auditing 83 answers on real FAQ pages, which checks break most often, and a template you can copy for rewriting the answers that fail.
Why Visitors Still Ask Questions Your FAQ Already Answers
Your answers were written for someone already on the page, already scrolling the list, already holding everything that came before. Strip that context away and the answer falls apart.
On our team’s read at WPChat, the questions people re-ask after reading an FAQ page tend to look the same. They’re the ones where the writer assumed the reader knew where they were standing on the site.
One answer mentions “our standard package” without naming which package. Another says “as mentioned above.” A third answers a pricing question with “it depends on the scope we agree.” All three look perfectly reasonable sitting in an accordion. Pasted into a chat window, they read like a shrug.
The pattern underneath: the visitor did read the FAQ. They just didn’t leave with anything they could act on, so they asked a person instead. The page collected the traffic and none of the credit.
There’s a tougher version of this argument coming out of content design, and it deserves a straight answer instead of a dodge. If a question gets asked often enough to earn an FAQ slot, your main content already failed. The pricing page should have handled the pricing question. The booking page should have covered the deposit.
Then comes the second problem, which is what happens when you fix it by copying. FAQ copy lifted from an About or Services page loses the context it was leaning on, and then it rots quietly while the original page gets updated around it.
The State of Iowa’s writing guidance makes the same case from the other direction, warning against building FAQ entries for content that already lives elsewhere on your site.
That objection is mostly fair. But deleting the page doesn’t delete the questions. It just moves them into your inbox, one at a time, for you to answer by hand.
What Happened When We Audited Real FAQ Pages
Here’s the sample, in plain words, before any numbers. I checked 83 individual answers across four WordPress sites. I found the pages through search, which means they skew toward the better-written end of the web, so the real-world failure rates are almost certainly higher than what you’re about to read.
I confirmed WordPress on each site through the generator meta tag or wp-content paths in the page source.
The results split by sector, and the split is the whole story. These two industries fail in opposite directions, so a blended average would hide the only useful finding here.
| Sector | Sites | Answers coded | Not portable | Dominant failure mode |
|---|---|---|---|---|
| Wedding photography | 2 | 47 | 36% | Answers that never resolve. Long, warm, no decision at the end |
| Home trades | 2 | 36 | 39% all-in | 8% length and context, plus 31% vacuous |
The wedding photographers over-answer. Their FAQ writing is genuinely lovely, full of reassurance about how the day will feel, and then it just stops without telling you what anything costs or when the gallery arrives. The reader finishes the paragraph warmer and no closer to booking.
The trades under-answer, and that’s where the quotes start.
One site describes its work as using “high-grade materials that meet industry standards” and, further down the same page, answers a warranty question by repeating the question verbatim in the answer field, and nothing else. Another answers a pricing question by saying the cost “varies depending on its unique aspects.”
Read those again. Not one of them could be improved by cutting words. The problem isn’t length, so trimming can’t reach it. That’s what vacuity means in practice, and it’s a different disease from length.
Now the failure mode I find most interesting, given its honest weight. Across all 83 answers, 11% depended on context outside themselves: “Follow this link,” “the bank details are on my contract,” “further than 2 hours from DN17.”
That’s the smallest of the three failure modes by a wide margin. I’m not going to build this article on it, and you shouldn’t rebuild your page around it either. It’s the most fun to spot and the least common to find.
One last observation, and I want to be careful about how much weight it carries. One site in the sample was clearly agency-built from a template, and it had almost no context-dependence at all, plus the highest vacuity rate of any page I audited. The boilerplate was clean because whoever wrote it had never done the work. That’s one page, not a pattern.
The Three-Axis Rewrite Test (With a Template You Can Copy)
Three questions, asked of one answer at a time. Every answer that failed my audit failed at least one of them, and most of the bad ones failed only one, which means the fix is usually smaller than it looks.
Work through them in order. Axis 1 catches answers that never land. Axis 2 catches answers that lean on something off-screen. Axis 3 catches answers that sound complete and carry no information.
Then copy the template at the end and use it on the next answer you write.
Axis 1: Does the answer land in the first two sentences?
Read only the first two sentences of your answer. If the visitor still doesn’t have the answer, the answer fails.
This is the wedding photography failure mode from the audit, and it’s the reason 36% of those answers weren’t portable. The writing is warm, the tone is right, and the resolution never arrives.
One note before the examples, and it applies to every Before/After pair in this section. The “before” text is a real answer from the audit, lightly cleaned. The rewrites are mine and they’re illustrative: the prices, brands and timeframes in them are invented to show the shape a good answer takes, not findings from the audit.
Here’s a real shape from the sample. The question was how long the couple waits for their photos.
Before:
We know how excited you’ll be to see your gallery, and we treat every wedding with the care it deserves. Every image is hand-edited by me personally, never batch-processed, because your day was unique and your photos should reflect that. Editing times do vary through the season and we’d always rather give your images the attention they need than rush them out. Rest assured, we’ll be in touch the moment your gallery is ready.
Four sentences. Nothing a couple can put in a calendar.
After:
Your full gallery arrives within 8 weeks of your wedding date, and you’ll get a preview set of 20 to 30 images within 5 days.
Weddings between June and September sit at the longer end of that window, because that’s our busiest editing stretch. If you need a few images sooner, for thank-you cards or a venue feature, tell us before the day and we’ll prioritize those.
The warmth survived. The waiting time arrived first.
That’s the shape most FAQ writing guides converge on, and it’s worth following: a direct answer in one to two sentences, then two to four sentences explaining what changes it. Answer first, drivers second.
For length, the answers that worked in my audit ran roughly 40 to 90 words. Genuinely complex topics like pricing tiers needed 120 to 180. Below 40 you’re usually skipping the drivers, and above 180 you’re usually writing an article inside an accordion.
The State of Iowa’s guidance already covers the narrow version of this rule: open with a definitive yes or no. That’s exactly right, and it only reaches yes/no questions. Axis 1 is the same instinct applied to every question type. If the question asks for a number, the number goes first. If it asks for a date, the date goes first.
Axis 2: Does the answer stand up alone?
Copy the answer out of the page and paste it into a plain text file. If it stops making sense with nothing around it, it fails.
Iowa’s guidance says this outright: fully answer the question, don’t just link to a different page. That rule was already written down by a government style guide, and it’s the right rule.
What I want to add is a test for it, and a wider net, because links are only one of the things an answer can lean on.
Across all 83 answers, 11% depended on something outside themselves. Three real examples, and what happens to each one when it travels.
“Follow this link.” On the page, the link is right there and the reader clicks it. In a chat reply the link often doesn’t survive the copy, and in a search snippet there’s nothing to click. Name the destination and give the information: “Book directly at yoursite.com/book, or call 01724 000000 between 9am and 5pm.”
“The bank details are on my contract.” True, and useless to anyone who hasn’t signed yet. Most people asking how to pay a deposit are deciding whether to book at all. Rewritten: “Your deposit is 25% of the total, paid by bank transfer. We email the account details with your booking confirmation.”
“Further than 2 hours from DN17.” This is my favorite one in the whole audit, because it fails in a way the business owner will never notice.
DN17 is a UK postcode district. The post town is Scunthorpe, in North Lincolnshire. The person who wrote that answer types DN17 every day, so it reads as obvious to them. A customer in Leeds trying to work out whether they’re inside the travel radius has no idea.
Rewritten: “We cover anywhere within about a 2 hour drive of Scunthorpe, North Lincolnshire. Beyond that we can still come out, with travel charged at 45p per mile.”
The rule underneath all three: name the thing instead of pointing at it. Postcodes, package names, “as mentioned above,” “our standard rate,” and internal shorthand all point. Names carry.
Axis 3: Does it say anything specific?
Take your answer and swap your business name for a competitor’s. If it’s still true, it’s vacuous, and it was telling your visitor nothing before you swapped it.
This was the largest failure in the audit and the one nobody warns you about. In the home trades sample, 31% of answers were vacuous, nearly four times the rate of the other two failures combined.
It’s also the axis that answers the question I get most often about FAQ writing: how do I tell if my answer is vague or just short?
Short is not the problem. “Yes, we’re insured” is four words and it passes cleanly, because the visitor asked a yes or no question and got a yes. There was no other information available to give.
Compare that to “We take safety seriously and are fully covered, so you can have complete peace of mind.” Longer, warmer, says the same thing, and it would be true pasted onto any competitor’s site in the country. That one fails.
Here are the three real ones from the audit, run through the test.
“High-grade materials that meet industry standards.” Swap the business name and it stays true of every roofer, plumber and builder trading today. Nobody advertises low-grade materials that flout industry standards, so the sentence carries no information at all.
“The cost varies depending on its unique aspects.” Every job varies. That’s not an answer to a pricing question, it’s a description of pricing.
A warranty question answered by repeating the warranty question. The answer field restates the question and stops.
Now the rewrites. The pattern is simple: replace the adjective with the number, the brand name, or the timeframe.
Before: “We only use high-grade materials that meet industry standards.”
After: “We fit Marley concrete tiles as standard, which carry a 30 year manufacturer guarantee. If your roof is listed or in a conservation area we source reclaimed clay pantiles instead, which adds around 20% to the materials cost. You’re welcome to ask for a different brand and we’ll price it.”
Same claim. It now says something a competitor down the road couldn’t copy without changing it.
Before: “The cost varies depending on its unique aspects.”
After: “A full re-roof on a standard 3 bed semi runs between 5,500 and 8,500 pounds, including scaffolding and waste removal. The biggest swings are roof pitch, whether the timbers need replacing once we strip the old covering, and your tile choice. Rotten battens are the most common surprise and add 400 to 900 pounds. We quote in person and the quote is fixed, so the number you get is the number you pay unless you change the spec.”
That one runs about 90 words, and it earns them. Pricing is exactly the kind of question where the 120 to 180 word range is fair game.
Before: a warranty question with the question repeated as the answer.
After: “Our workmanship warranty is 10 years from completion, and it’s insurance-backed, so it stands even if we stop trading. Materials are covered separately by the manufacturer, usually 25 to 30 years on tiles. Storm damage isn’t covered by either, because that’s a claim on your buildings insurance.”
Notice what happened to the length. Two of the three rewrites got longer, and every one of them got faster to read, because the reader stops re-reading a sentence when it finally contains something.
This is where I’d push back gently on the standard advice to keep answers as brief as possible. Brevity applied without a specificity test just deletes the numbers, and the numbers are the answer.
The answer template
Copy this and fill it in. It’s the same structure behind every rewrite above.
Question (H3, written the way a visitor actually says it)
- Sentences 1 to 2: The answer. A yes or no first if it’s a yes/no question. A number first if it’s a number question. A date first if it’s a timing question.
- Sentences 3 to 5: The drivers. What changes the answer, and by how much. Give the range or the amount, not just the fact that it varies.
- Sentence 6 (optional): One next step, named in full. “Book at yoursite.com/book” or “call 01724 000000,” not “click here.”
Then run the self-check, which takes about five seconds per answer.
Cover the question with your hand and read only the answer. If it still makes sense on its own, it’s portable, and it will survive being pulled into a chat reply, a People Also Ask box, or an AI Overview.
If you have to uncover the question to understand the answer, you’ve found your next rewrite.
How many questions should an FAQ page have?
As many as you have real questions, and no more. Towson University’s guidelines recommend keeping a page under 10, which is the most-cited number in FAQ advice, and it’s a reasonable instinct: a page of 40 questions usually means nobody has decided which ones matter.
But the ceiling is a rule about the page, not about any answer on it, and my audit suggests it tells you very little. The one page in my sample that actually met the 10-question ceiling had the highest vacuity rate of the four: 40% of its answers said nothing. The 26-question and 27-question pages scored 27% and near zero.
Four pages isn’t enough to claim that longer FAQ pages are better written, and I’m not claiming it. What the sample does suggest is that hitting the question count tells you nothing about whether the answers work. You can sit comfortably under 10, group everything neatly, and still fail the three-axis test on every entry.
So use the number as a prompt rather than a target. If you’re well over 10, ask whether the extras are questions customers actually send you or questions you invented to fill the page. If you’re under 10, that’s not a pass mark. Run the three axes anyway.
Where a Portable Answer Set Actually Gets Used
Your answers get read in three places, and only one of them is your FAQ page.
The first is the page itself, where a visitor scrolls, opens an accordion, and reads with your categories and your headings sitting right there.
The second is a search surface. A featured snippet, a People Also Ask box, an AI Overview. Your answer gets lifted out and shown next to answers from four other sites.
The third is a chat widget, replying to a visitor in real time.
That third one is the strictest test you will ever give an answer. The visitor never sees your page. They never see the question sitting above the answer. They never see the related questions on either side that quietly filled in the gaps.
They typed something in their own words and got your paragraph back, alone, with nothing around it.
An answer that doesn’t resolve, or that can’t stand on its own, dies in that moment. It was leaning on the page, and the page isn’t there.
This is exactly where WPChat fits, and it’s worth being precise about what it does.
Its Smart FAQs feature takes a visitor’s natural-language question and matches it against a library of FAQs you wrote in advance, using AI token-based search to find the closest one.
That library is the artifact you just built with the template above. The plugin isn’t writing your answers. It’s delivering them.
Which is why the rewrite has to come first. A vague answer can get matched perfectly and still leave the visitor with nothing, because matching the right FAQ and giving a useful reply are two different jobs.
The plugin handles the first one. Your writing handles the second.
If you’re weighing it up, size the plan to the number of FAQs you actually have rather than the number you plan to write someday.
| Plan | FAQ library size | Smart Search Tokens | Who it fits |
|---|---|---|---|
| Basic | 30 FAQs | 10,000 | A single service business with one offer |
| Plus | 100 FAQs | 25,000 | A store or multi-service site with real category spread |
| Elite | Unlimited | 50,000 | Large catalogs, multi-location, or ongoing library growth |
Check both columns before you pick. FAQ count is only half the sizing question, and the tokens are what get spent every time a visitor asks something.
A busy site with 25 FAQs can hit the token ceiling long before it gets anywhere near the FAQ ceiling, so look at your traffic as well as your library.
All plans come with a 14-day money-back guarantee, so you can put your answer set in front of real visitors and see which ones hold up on their own.
FAQs About Writing FAQs
What is a FAQ format?
A FAQ format is a question written as a heading, followed by a short direct answer, repeated down the page. There are three common presentations: a numbered list of questions, headed sections that group related questions, and an accordion that hides each answer until clicked. The format is the same in all three. Only the packaging changes.
What does an FAQ look like?
A finished FAQ answer looks like this: “Yes, we cover the whole of North Lincolnshire and up to 20 miles outside it. Anything further than that is quoted separately, and we’ll tell you before we book the job.” Question as a heading, answer in two sentences, no reliance on anything else on the page.
How long should an FAQ answer be?
Most FAQ answers should run 40 to 90 words, and complex topics like pricing models can stretch to 120 to 180. But word count is a symptom, not a target. A 30-word answer that says nothing still fails, and a 150-word answer full of specifics still passes.
Should there be an apostrophe in FAQs?
No, there should not be an apostrophe in FAQs, because it’s a plural and not a possessive. Write FAQ’s only when something belongs to the FAQ, as in “the FAQ’s opening line.”
The Takeaway
A good FAQ answer is one that still works when it’s read alone.
That’s the whole test. Take the answer off your page, strip away the heading above it and the questions around it, and see if a stranger could act on what’s left.
The three axes are how you check quickly. Does the answer land in the first two sentences? Does it stand up without anything else on the page? Does it say something specific enough that a competitor couldn’t paste it onto their own site?
Fail any one of those, and the answer only works in the one place it’s sitting.
Here’s what I’d do this week. Pull your five most-asked questions, run the test on each one, and rewrite the ones that fail.
Then stop. Five portable answers are worth more than thirty that fall apart the moment they leave the page.
Once those five are portable, they can do a second job. A chat widget can serve them to visitors in real time, matched to whatever a person types in their own words, without you touching the keyboard.
That’s what WPChat’s Smart FAQs feature is for. Pick the paid plan that fits the number of FAQs you actually have, start with Basic if you’re under 30, and put your rewritten answers in front of real visitors.
