We increased daily ChatGPT citations from 200 to 1700+ for a brand new domain. This is the 3 step process we used to create content that gets cited and increases brand mentions. The most important thing to note is that we didn’t “over-engineer this”, we wanted to keep this process as simple and scalable as possible. Step 1: Setting up a prompt strategy. When we started tracking prompts, we weren’t trying to find a singular golden prompt that would unlock everything. We grouped our prompts into 3 categories A) Organic Prompts (80% of the prompts). All around the category, problems potential customers are facing, industry questions. But most importantly, no brand or competitor mentions. B) Branded Prompts (10% of prompts). Just too keep a temperature on sentiment around the brand C) Competitor Prompts (10% of prompts). To understand the sentiment around our competitors and where we need to increase content coverage to influence groups A & B. These groups are then set into Monitors in Promptwatch Step 2: Analyzing the Query Fan Outs. This is where we find the topics for our content pieces. Once we have a few days of data of responses, we use the Promptwatch MCP in Claude to analyze all of the QFOs for prompts in our organic monitor. Here we’re looking for patterns: What wording is being repeated, Average number of Fan outs, etc.. Step 3: We then write content based on these recurring themes in the QFOs and publish. Once they’re published we measure how long does it take to get crawled, how long does it take to get cited, and the content that gets cited first, we double down on. The content that doesn’t, we kill and move on. This continuous process is how we keep growing our daily citations.
Daniel Hale
1mo
I feel like this is kind of silly. Perhaps the query fanouts I'm seeing are different than yours. What I see them as is a bullshit jumble of additional terms mashed onto the actual main keyword. Take an example from a prompt we are tracking "commercial cleaning services CITY". Here is one fanout "best commercial cleaning companies CITY STATE janitorial services CITY list contact information". I have obviously replaced the actual location info with variables (CITY, STATE). Now you tell me how the hell you make a piece of content to target that fanout, because that screams of spam to me. Adding supplemental information to the actual page targeting that term - sure, that could potentially make sense. But making new pages targeting this dribble is a fools errand.
Cool! Did that lead to any noticeable growth in visitors, leads, or revenue?
Mihai Pintilie
1mo
Using query fan-outs as the source for content ideation is an interesting shift from traditional SEO workflows. Have you noticed that content built from high fan-out topics gets cited more often because of topic selection itself, or because the language ends up matching how models retrieve information?
Nazar Dziadyk
1mo
Gijs de Groot 1700 citations is great. But did any of it show up in pipeline? And did ChatGPT start saying nicer things about your brand, or did just the count go up?
Shubhankar Das
1mo
How long did it take to get from 200 to 1700? What was your backlink generation strategy? What's the average KD for the keywords you've targeted? If you're working on a niche where the average DA of competitors is in the low range, then it's possible to get citations in Google AI mode. ChatGPT is a whole different game that involves way more off-page SEO work than on-page work, like content creation. So, just curious to know how this was possible.
Tayyaba Ameen
1mo
I like the fan-out step here. teams still write GEO content from assumptions, not from how the model breaks down the topic.
Bhavay Gupta
1mo
Quick one, when citations jumped from 200 to 1700, was that new pages getting picked up or your existing ones getting cited more often?
Jose Thomas
1mo
If the brand appearing in ChatGPT answers truly moves the pipeline, I'd consider it a growth lever.
Some questions: 1. how much traffic does the site get? 2. do you think this approach is sustainable? 3. if you think it's sustainable, do you think it's defensible? 4. do you think this is something new? 5. do you understand that a small-mid sized business copying this approach is likely to get their site penalized by Google for scaled content abuse? 6. do you understand the damage that can do to their business? 7. do you think that looking at a line of "citations" going up for a while on a chart makes up for that? 8. do you care, or do you only care about your ARR? Bonus question: wen model collapse?