AI search is a bottom-of-funnel game
98%
of buying searches named a brand. Informational searches? Just 34%
How Prismic uses Prismic for AI Search
From invisible to #1 recommended brand
We pointed Prismic at our own blind spot in AI search and in 30 days, we outranked established players in the ABM space. This is how we did it
EveryoneknowsAIsearchmatters.Whatnobodyagreesoniswhattoactuallydotoshowup,sowedidtheexperiment
We ran the exact loop we sell, on a problem we actually had.
No hypothetical brief. We picked one category where we had real AI-visibility gaps - ABM landing pages - and tracked it daily across ChatGPT, Perplexity, Google AI Overviews, and Microsoft Copilot.
The test was deliberately narrow: three prompts a real ABM marketer would actually type, run every day, on every platform, for just over a month. We recorded whether Prismic was mentioned, where it ranked, and which URLs got cited.
We then worked backwards from those citations to figure out what content we needed to build.
The logic behind the prompts we chose:
How would someone actually find this product? We wanted;
- One prompt where someone is mid-task and might surface us incidentally,
- One short direct recommendation search,
- One scenario-based search that mirrors how a real buyer evaluates tools.
We prioritized creating the right content over volume
Every piece of content had one of two jobs: earn citations by matching the format and intent of what AI platforms were already surfacing, or teach the AI how Prismic compares to the tools it was already recommending.
The 4-week execution that changed our AI rankings
Before the first page went live, we spent time getting the foundation right: setting up daily tracking across all four platforms, running the research, and building the content list. Once that groundwork was done, the execution moved fast.
WEEK 1
Foundation
- Published “Prismic vs alternatives” comparison pages, targeting the competitors we saw cited most in the research phase
- Drafted a broader pipeline of guide and recommendation content
- Refined internal linking and page structure
WEEK 2
Intent alignment
- Shipped “best tools for ABM landing pages” - the exact listicle format we saw winning citations in the research phase
- Expanded listicles into adjacent use cases
- Updated key pages with structured formatting
WEEK 3
Authority
- Published dedicated “alternatives” pages to sharpen differentiation against the top-cited competitors
- Added longer-form instructional resources to increase depth
- Interlinked all ABM-related pages into a topical cluster
WEEK 4
Consolidation
- Closed remaining content gaps flagged by the weekly tracking runs
- Strengthened topical coverage across the cluster rather than adding new topics
- Shifted to a monitoring phase
Here’s what 30+ days of the right content actually did.
Not projections. Not estimates. Numbers from the tracking tool, pulled at the end of the experiment.
Four findings that will change how you brief your content team
One content format dominated AI citations
51%
of all non-homepage citations went to comparison listicles (more than other format combined)
Your website is your best PR agency
73%
of all AI citations went to owned content, ahead of big publishers, blogs, and social
Quality AI-generated content will get cited
14 days
is how long it took our AI-generated content to get cited across four major AI platforms
Here’s how you can replicate the workflow from start to finish with Prismic
Connect your data, get recommendations
Prismic pulls in the prompts and rankings from your AI visibility tool, and tells you exactly what content to create or update to improve your visibility.
Produce the recommended content
Prismic’s content agents research what’s ranking and what other content are missing, then draft a page in your voice, from your knowledge base. New content or updates to existing pages.
Review, edit, publish, repeat
You approve and ship. Your visibility score updates. The next gap feeds straight back into the same pipeline. You are always in control.
Finally, some advice to anyone looking to replicate this (with or without Prismic):
AI visibility monitoring is a non-negotiable
Every day, the tracking tool (Profound and PeecAI) ran all three prompts and logged whether Prismic appeared, at what position, and which URLs were cited. Without this, we’d have been guessing.
Your content must always have a unique POV
For each page on the list, we fed the Content Agent our actual competitive research docs, sales call transcripts, messaging guidelines, and brand voice docs. Generic AI content won’t cut through.
There should always be a human in the loop
An editor reviewed every draft for accuracy, added specific examples, and tightened the keywords, then published directly into Prismic’s CMS. No dev touch required.
Publishing speed matters more than most people realise
No developer touched this experiment at any stage. Pages went live the same day they were approved — that gap between “approved” and “indexed” is where most teams lose the compounding advantage.
Now run the same playbook for your category
The same workflow that took us from invisible to #1 is now the product. Get a quick demo to see how Prismic;
- Pulls signals from your AI tracking tool, fan-out queries, and keyword research automatically
- Recommends a prioritised list of content to create based on what’s actually being cited
- Drafts pages grounded in your own product knowledge, not the open web
- Gets content live the same day it’s approved