Sentiment

What the engines say about you when they name you.

Being recommended is half of it. The other half is how ChatGPT, Perplexity, Gemini and Google describe you when they do: warmly or grudgingly, for your quality or against your price, and whether the facts they repeat are true. Citeify scores that tone 0 to 100, keeps their exact words, and checks every claim against your own site.

Run a free Check See the plans, from £39/mo

A worked example

Northwind Coffee, as the engines talk about it.

A fictional roaster, rendered by the same Sentiment section a customer’s report uses. Read down: the score and the verdict, what the tone attaches to, how the engines frame its competitors, the exact sentences, and the claims with what the site actually says beside them.

Sentiment

What the engines say about you when they name you: the tone, what it attaches to, and their exact words. Paid runs also ask two fixed reputation questions.

The engines named Northwind Coffee 14 times, mostly positively (64% positive, 7% negative). They credit quality and expertise and mark Northwind Coffee down on price. 1 claim they state as fact conflicts with your own site. Tone 79/100.

79/100

Sentiment

14 mentions: 9 positive, 4 neutral, 1 negative. Credited for quality and expertise. Marked down on price.

In their words

Verbatim from the engines' answers, unedited.

“Northwind is a small-batch roaster praised for freshness, though pricier than the big brands.”
Perplexity
positive
buyer question
“best specialty coffee subscription uk”
“Reviewers consistently mention the roast dates on every bag and the knowledgeable tasting notes.”
AI Overviews
positive
reputation prompt
“What do people say about Northwind Coffee?”
“Northwind Coffee appears to be a reputable independent roaster with a strong local following.”
Gemini
positive
reputation prompt
“Is Northwind Coffee reputable?”
“Northwind Coffee offers a subscription, but it is on the expensive side compared with Beanline.”
ChatGPT
negative
buyer question
“coffee subscription uk under £20”
“Northwind Coffee, a roaster based in Bristol, sells single-origin beans online.”
Perplexity
neutral
buyer question
“where to buy single origin coffee beans uk”
“Customers describe the team as helpful when a delivery goes wrong.”
AI Overviews
positive
reputation prompt
“What do people say about Northwind Coffee?”

What they state as fact

Claims the engines make about you, checked against your own site. A conflict is worth correcting at the source; the engines read your site too.

Roasting since 2014← confirmed on your site
Based in Bristol← conflicts with your siteYour site: “Roasted in Bath, Somerset.”
B Corp certified← confirmed on your site
Ships whole bean only← not on your site

What moves it

The engines assemble what they say from reviews, press, directories and your own pages. A run of new reviews, a mention in a round-up, or a competitor’s campaign shows up in the tone within a month. So does a wrong fact, once it is repeated.

What to do about a wrong claim

Correct it at the source. If your site says Bath and the engines say Bristol, the engines are reading an old directory. Fix the directory, state the fact plainly on your own site, and the next run shows whether it took.

Why it is measured, not guessed

Every plan asks each question five times on six engines and adds two reputation prompts, so the score rests on dozens of mentions and carries a history. The free Check asks once, which is enough to see whether you are named but not enough to read a tone.

FAQ

Common questions.

Every run keeps the sentence in which an engine names you, plus one either side. A classifier reads each mention and calls it positive, neutral or negative, and names the attribute it turns on: price, quality, expertise, trust, service, range, location or availability. The score is the balance of positive against negative across all mentions, on a 0 to 100 scale where 50 is even. Under five mentions there is no score, because a number from three sentences is not a measurement.

Two places, and the report labels each quote with which. Buyer questions, where you are named unprompted in an answer about your market, and two fixed reputation prompts asked on every plan: what people say about you, and whether you are reputable. Reputation prompts are where the engines repeat what they have read about you elsewhere.

Anything an engine states about you as fact, a founding year, a location, a certification, is checked against your own crawled pages. It is confirmed when your site says the same, conflicting only when your site says something different, and left as unverified when your site is silent. A conflict is worth correcting at the source, because the engines read your site too.

The free Check runs 5 questions once each on four engines, which is enough to see whether you are named but not enough mentions to read a tone from. Sentiment is on every plan, where each question is asked five times on six engines and the reputation prompts are added.

What the engines say about you when they name you.

Northwind Coffee, as the engines talk about it.

Sentiment

What moves it

What to do about a wrong claim

Why it is measured, not guessed

Common questions.

How is the tone score measured?

Where do the mentions come from?

What does "checked against your site" mean?

Can I see sentiment on the free Check?

Find out which questions you are losing, and to whom.