Product update
Kojable builds an algorithm to find editorial opportunities for AI search
We built an algorithm that ranks the specific articles, guides and comparisons where a publisher could add information buyers need. It starts with the questions buyers ask AI and checks what each page already covers. Your team gets the exact page, a proposed addition and the evidence to check before approaching the publisher. After an approved change goes live, an agreed retest compares whether your company appeared in the answers and which tracked pages were cited. This is part of Influencer Marketing for AI Agents.
Rank opportunities
Find specific articles, guides and comparisons where a publisher could add information buyers need.
Review a proposal
Your team gets the exact page, a proposed addition and evidence to check. The publisher decides whether to make the edit.
Compare later answers
After an approved change goes live, an agreed retest checks company mentions and citations to tracked pages.
What Kojable built
When a buyer asks AI to compare tools or explain an implementation step, the answer may link to a useful guide. The guide might cover the category well and leave out one fact the buyer needs to judge whether a product fits. We look for opportunities to add that detail.
-
Start with the buyer question.
Capture the AI answers and record the citations they show. Research other relevant articles, guides and comparisons on the same topic.
-
Check the exact page.
Verify its identity and read what it says. Look for missing or incomplete information that matters to the buyer. Identify the author or publisher where we can verify it.
-
Rank the proposed contributions.
Assess the page’s relevance to the question, its credibility and audience, and whether an addition would be useful and practical for the publisher to make.
A page can qualify through additional research, even if it didn’t appear in the recorded citations. An exact-page citation needs evidence that the recorded link identifies that page. How Kojable verifies source identity
What your team receives and approves
Each ranked proposal gives your team the page URL, the buyer question, the gap in coverage, the suggested addition, where it could go and the evidence to check. It also includes the author or publisher where available.
In one anonymised B2B programme, Kojable delivered 79 ranked proposals across 56 domains to Company B’s delivery partner. Of those, 69 proposed adding a mention and 10 proposed improving an existing mention. Company B’s proposal workbook
One proposal concerned the integration section of an independent implementation guide. The guide explained how tools worked together. A buyer who needed to review changes before applying them still needed to know where Company B’s product could fit in that workflow.
-
The proposed addition.
A short example in the integration section showing how a proposed change could go through review or approval before it was applied.
-
The evidence to review.
Customer-approved product documentation and a verified example of the workflow. The proposed wording about governance needed to reflect what the product could support.
-
The approval decision.
The team checks the wording and evidence and chooses whether to take the proposal to the publisher. The publisher decides whether to make the edit. Company B’s worked example
What you see after retesting
When follow-up is in scope and relevant material goes live, we ask the same buyer questions under comparable conditions. The comparison shows where the company appeared and which exact tracked pages appeared in the final citations. How Kojable compares AI answers before and after a change
Company A. We prioritised 129 third-party pages. In the matched September–November 2026 comparison, the company’s visibility score for questions that did not name it rose from 2.42% to 12.12% across 11 questions and five AI and search surfaces. Across the full 12-question panel, the number of distinct URLs from the 129-page list that appeared in citations rose from one to six, using the same October-selected list on both dates. ChatGPT cited one newly observed target page in November. Its answers to unbranded questions named Company A zero times in both periods. Company A’s results and citation method
Company B. In the matched September–October 2026 comparison, the company’s unbranded visibility score rose from 50.00% to 61.31% across seven questions and four AI surfaces. The campaign team reported that ten separately approved external pages were live before the October retest. Two overlapped with the 79 proposals. None of those ten exact URLs appeared in the final citations across 236 retained observations from both dates. Company B’s matched results · Publication and citation findings
How four AI search workflows compare
All four offerings measure how brands appear in AI answers and which sources AI cites. They help teams act on that information in different ways.
Later Creator AEO
Run and measure creator campaigns. Choose creators using citation and campaign data, then create or repurpose content. Later’s Creator Evidence Index can publish source-linked answers from campaign videos.
Track mentions, citations and competitive visibility to guide the next campaign.
Kojable AI answer alignment
Review a specific change. Find an information gap in an AI answer. See the exact page section, the suggested addition and its supporting evidence, including for an existing independent article.
After publication, an agreed retest compares what AI says and which exact pages it cites.
Lighthouse AI visibility and page fixes
Prioritise site edits and plan pitches. See which publishers and authors AI cites. Get ranked fixes for your own pages and briefs or drafts for questions where your brand is missing.
Mark a fix as shipped and Lighthouse reruns the same prompts on a schedule.
Muck Rack PR and AI visibility
Find and pitch journalists. See cited outlets, journalists and exact URLs. Build media lists, personalise pitches and send outreach.
Track pitches and earned placements and review how the brand appears in AI over time.
Questions about the workflow
Can a page qualify if it didn’t appear in the measured citations?
Yes. Additional research can find pages about the same buyer question. A page can qualify if the publisher has a credible way to add useful information and the proposed wording has supporting evidence.
Can Kojable propose improving an existing mention?
Yes. An article may already name a company and give buyers little detail about how its product fits. A proposal can add that context at the relevant point in the article, with evidence the team can verify. Company B’s migration article example
What makes a retest comparable?
Use the original questions, the same AI surfaces and the agreed measurement rules. Once the change goes live, record material differences in the testing conditions so the team can interpret what it sees. Kojable’s retest guide