How it works
We measure how AI answers about you. Then we write.
Three steps, one plan. Methods from the peer-reviewed GEO study (Princeton, Georgia Tech, the Allen Institute and IIT Delhi — ACM KDD 2024), measured against real AI answers every month.
Step 1
Book a call
Before the call we send you a Website Consistency Check of your own site. On the call — 30 minutes — we ask the AI engines about your business while you watch: prompts like "Who is the best [your industry] in [your city]?" for a local business, or "Which [organization] should I choose?" for a national one.
What they say is your starting point. The notes are yours whether or not we work together.
Step 2
We measure and write
Every month we ask all six engines the same questions and record what each one said: which engines named you, which named a competitor instead, and which sources they cited. An answer is quoted where the engine gave one; a run that fails is shown as failed, never as a zero.
Then we write what the research favours — the content methods the Princeton study tested — for your business:
- Posts for the platforms we set together, and blog drafts when the blog is in your mix
- Profile text, in month one
- Structured data (a Schema Pack) and page specs, with each website review
- Q&A page content written from your confirmed facts
Step 3
You publish. You see what changed.
You, or your web person, publishes the content and implements the specs. Under the base plan RankForward does not post, paste or change anything on your properties — that is Managed Execution, a priced tier limited to three clients.
From cycle two the report compares like for like, engine by engine, where the samples allow. You see what changed; nobody promises what will.
Who this is for
Local service businesses
Customers ask AI "who near me?" — a roofer, a plumber, a blinds company in one city.
National organizations
Customers ask AI "which one?" with no city attached — memberships, education providers, brands.
We take clients nationally. Multi-location businesses: ask on the call.
What we do. What you do.
We do
The measurement, the report, and the writing: posts, blog drafts, profile text, schema and page specs — ready for your final check.
You do
- Publish the posts and blog drafts you choose to use, on your own profiles and website
- Implement the specs we write — schema, Q&A page content, page changes — yourself or through your web person
- Confirm the facts about your business we may quote; we quote only what you have confirmed
Want us to do the posting and pasting? Managed Execution is $2,750 a month all in, limited to three clients, and everything still goes through your approval — with the profile or CMS access you choose to grant for the agreed work.
Why you can't easily DIY this
The research is public. The measurement is the work.
Anyone can read the paper. What takes the time is asking six engines the same questions every month, keeping every answer, and writing to what they actually said — then doing it again next month so the comparison means something.
That is what the plan buys: the measurement, repeated, and the writing done for you.
6
AI engines measured, every cycle
9
content methods from the Princeton study
The evidence
The study behind the methods
Our content methods come from "GEO: Generative Engine Optimization" (Aggarwal et al., ACM KDD 2024 — Princeton University, Georgia Tech, the Allen Institute for AI and IIT Delhi). The researchers built a benchmark of ten thousand questions and measured, on a held-out set, how often a changed page was cited by generative search engines.
In their test, the strongest lifts came from adding quotations, statistics and credible citations to a page, and citing sources helped the lower-ranked pages most; old-style keyword stuffing made pages less visible. Those are the study's results, dated 2024, about the methods in the researchers' benchmark — not a prediction for any one business. What your business sees is what we measure, every month.