Frase

Has someone asked why your brand isn't in ChatGPT answers yet? Score one page 0-2 on five things: the full answer in the first 40-60 words, specific headings, a fact or example every 150-200 words, inline links to sources, a visible date. Fix the lowest total first.

Worried your product pages are too salesy for AI to quote? One AI citation study compared pages AI cited with top Google results it ignored. About 3 in 4 of both were businesses selling something, and similarly salesy. What differed: the cited pages answered the question first.

Set Cloudflare to block AI training? Since its Sept. 15 default change, that block also covers crawlers that do search and training together. Cloudflare says that includes Googlebot, Bingbot and Applebot, unless you opted out beforehand. Have you checked what yours covers now?

Google says it has added more inline links to AI Mode and AI Overviews, and is showing them higher up in answers. Before checking if any are yours, write down where you stand: 20-50 buyer queries, whether you're cited, and who's cited instead. Then you can tell if a fix worked.

A brand-name prompt is the easiest AI visibility check to run, and it only covers buyers who already know your name. Discovery happens on topic questions, where nobody has typed your name yet. Build most of your tracked list from those, and note who gets cited when you don't.

If a buyer sends Meta's Muse agent to research you, its visit to your pricing page looks like theirs. With no published user agent, robots.txt has no name to match and analytics will likely count it as the buyer. It reads text. Put prices in HTML and label every form field.

Wondering if blocking AI search crawlers is normal? We read the robots.txt of 98 B2B software vendors from US buyer searches. None block OpenAI's, Perplexity's or Anthropic's search crawlers site-wide (robots.txt only, not firewalls). If yours does, check it was deliberate.

Can you count on ChatGPT's referral tags? OpenAI added UTM tags to ChatGPT links in June. Since GPT-6 they appear to be gone, and links appear to send no referrer. OpenAI hasn't confirmed it, and it may be a bug. If you measure AI visibility by clicks, track citations too.

Think "User-agent: *" plus "Disallow: /" keeps every bot out? A bot with its own section ignores the * rules completely. So if your robots.txt has a GPTBot section anywhere, your "block everything" line doesn't cover it. When did you last open yours?

Want your content out of OpenAI's training but still in ChatGPT search? Separate crawlers. Block GPTBot (training), leave OAI-SearchBot (search) alone. Of 98 B2B software vendors we checked, 2 block a training crawler in robots.txt. 6 more state ai-train=no without blocking.

Did your ChatGPT traffic fall around October 9? One analysis of ~500 pages: ChatGPT visits down about 31k a day, direct up about 26k a day on the same pages. GPT-6 links appear to send no referrer or UTM tag. Check direct on your ChatGPT landing pages before you report a drop.

How long should a featured snippet answer be? Aim for 40-50 words, roughly 250-300 characters. Shorter loses context. Longer gets truncated. Put the question in an H2 or H3 and answer it straight underneath, before any background. How long is the answer on your best page?

Won a featured snippet and not sure it’s earning anything? Compare CTR on queries where you hold it with queries where you don’t. That shows whether it sends traffic or gives a zero-click answer. Then check which of those pages AI engines cite. Frase tracks cited pages by URL.

Does most of your content sit at the top of the funnel? A lot of brands flood the awareness stage with blog posts and leave consideration thin. The pages buyers need in between: comparisons, use-case breakdowns and clear answers to "is this right for me?"

A common problem: blog posts with plenty of traffic that barely move anyone toward a decision. Check what it points to next. A post that answers the question and gives no clear next step is a dead end, and readers leave the funnel there. It sounds obvious but a lot fail this.

OpenAI have announced watermarking in the EU when drafting content with ChatGPT. But will it hurt rankings? OpenAI says the mark doesn't measure how much a human contributed and says nothing about whether the page is high quality, and that's the bit Google's systems assess.