From 35 to 259 AI citations in seven weeks: how CodHob took hold in Kenya's credit platform market

CodHob is a platform for launching complex technology products under your own brand. Its CRYSTAMO product targets the Kenyan market. The site was built on Tilda. That is good enough for a quick content start, but the platform has a ceiling on E-E-A-T signals, the very signals AI engines rely on when choosing sources for an answer.

The client's task was specific: when a user asks AI about a credit platform in Kenya, CodHob and its expert content should appear in the answer. Not a one-off mention, but measurable, growing dynamics that decision-makers inside the company and investors can both understand.

By April CodHob had neither an AEO program nor regular funnel monitoring. The starting point was zero.

Monitoring rested on two metrics that grow at different speeds and should not be confused.

• Citation is when an AI engine gives a link to a specific URL in its answer.
• Mention is when the brand is simply named in the text, without a link.

The first is easier to grow with content. The second requires the brand name to settle in the model's "memory" as a recognizable source.

We measured across a four-stage funnel (Awareness, Evaluation, Consideration, Decision) and 160 queries in the Kenyan B2B fintech market. By August the panel was expanded from three AI engines to five: Gemini and Google AI Overviews were added to the already tracked ChatGPT, Perplexity, and Yandex.

Publications went out on three platforms: the site's /news/ section, LinkedIn, and Medium. After the June measurement came a wave of publications on June 30 and a piece on data migration on July 8. Some of these URLs entered AI citations for the first time in the August snapshot.

We wrote content not "about everything" but for specific decision stages: implementation risks, key platform features, the deployment process, regulatory compliance, and launch readiness testing. That mattered, because general introductory articles about what a credit platform is were almost never cited.

Over seven weeks, from June 23 to August 10, total citations of codhob.sg across five AI engines grew from 35 to 259.

The number of pages from results grew from 9 to 16, and the number of #1 positions for the site's own URLs grew from 15 to 105. In the overview block the site became one of the sources for 47 of 158 queries, about 30% coverage.

Here the case has a fork that would have been easy to skip, but we deliberately count both figures.

Of the jump from 35 to 259, about a hundred citations came from the new measurement channels, Gemini and Google AI Overviews, which were not in the June panel. If we remove that effect and compare the same three AI engines that were measured in June, growth is 35 to 159, a factor of 4.5. That is the figure worth building forecasts on, and it does not depend on how many new panels get added to the measurement.

The most telling change is not even the total number of citations but where they appeared.

Decision queries look like "is the platform ready to launch," "go-live," and "onboarding." At this stage the user is already choosing a contractor, not studying the market in general. More than half of such queries now return an answer with a link to CodHob. That is the effect of content on implementation risks, deployment, and launch readiness, the topics that collected the most citations.

The pattern is simple: materials on risks, features, deployment, go-live, and compliance win. They are specific, factual, and tied to a decision. General overview articles almost never make it into citations.

As of August 10, CodHob is the second domain in the niche by citations, behind only LinkedIn.

If we remove the sites with user-generated content, LinkedIn and YouTube, CodHob is first among product domains in the niche, ahead of specialized credit platform vendors.

The case has a weak spot, and it is worth stating too. The site link appears in answers often, but the brand name is rarely spoken.

Perplexity is the platform that cites CodHob URLs most willingly (79 citations) but almost never names the brand. That is the next point to work on. Citing a URL and the model recognizing the brand are different layers, and the second grows more slowly than the first.

Three conclusions worth taking from seven weeks of work with CodHob.

• First, citation growth is not linear and depends on the coverage of the measurement panel. Compare on equivalent platforms, otherwise it is easy to pass off an expanded methodology as organic growth.
• Second, content tied to a specific decision stage, such as risks, deployment, and go-live, works better than general overview material and moves the bottom of the funnel the most, where customers ready to buy are.
• Third, citing a URL and mentioning the brand by name are two separate metrics that grow at different speeds. For CodHob the second is now the next growth point.