Case Study

Building the Digital Asset Constellation in One Day

A ChatGPT Conversation Became a Warehouse Query. The Warehouse Query Became 13 Live Pages Across 9 Sites.

Krisada / Claude 3 views
Proof Point

Three things stood out.

Measurement Frame

What was measured.

Build Date 2026-08-04
Sites Touched 9
Content Pieces Published 13
Distinct Site Architectures Encountered 3
Baseline 90d Impressions 9,903
Baseline 90d Clicks 28
Baseline 90d Ai Crawler Requests 13,781
Warehouse Events Logged 4
Why It Matters

The short version.

Krisada brought a Kodi conversation proposing a Digital Asset explainer series and a Digital Asset Observatory.

Instead of writing content on spec, the plan got checked against real GSC and AI-crawler data in the Digital Karma Data Warehouse first.

The data held up. What shipped was 13 pieces of content across 9 sites, using the actual architecture each site already had, not a template forced onto all of them.

Context

What was the system?

The starting claim was simple: digital assets are entering the news cycle through legislation coverage, and the portfolio already had the properties to own that topic early.

That is a easy claim to make and a hard one to verify.

The warehouse made it checkable instead of a guess.

Methodology

What changed?

Pulled 90-day GSC and AI-crawler data for the 7 sites in the original proposal, plus a 180-day scan of every 'digital asset' family query across the full portfolio.

The cluster sat at 9,903 impressions and 28 clicks over 90 days, average position in the 40s to 70s.

Nobody in the portfolio ranked for this topic yet.

But the same 7 sites logged 13,781 AI-crawler requests in that window, ClaudeBot and GPTBot leading on most of them.

The sites were already being read. They just were not being found.

That gap became the thesis: publish real content on top of AI visibility that already existed, and see if search visibility follows.

Built a 3-tier plan (AssetClassLeverage.com as the public hub, DatasetSEO.com as the data layer, AISymantix.com as the AI-behavior research layer), then executed it.

Each of the 7 original sites turned out to run a different content architecture.

Three were straightforward JSON-article sites.

Three were marketplace or product templates with no blog system at all, one needed a one-line router change, one used an existing generic page template, one was cloned from a static-page pattern already on the site.

Read each site's actual router and template code before writing anything, rather than assuming one format would fit all seven.

A ninth site, SmartDigitalInvesting.com, got added mid-build after Krisada flagged it as growing on its own.

The warehouse confirmed it: impressions climbed from 3 to 798 a month between February and July.

That site's report template turned out to be shipping placeholder text for every report it had ever published.

Fixed that as part of adding its first real one.

Findings

What did the evidence show?

Three things stood out.

First The data changed the plan before any content got written.

The original proposal treated AssetClassLeverage.com and AISymantix.com as the two candidates for hosting the Observatory.

The warehouse showed AssetClassLeverage.com already had the only glossary, directory, and category scaffolding of the seven sites, and that DatasetSEO.com had already claimed the measurement-layer role in a post from three days earlier.

The architecture assigned itself once the data was in front of it.

Second Assuming one content format across nine sites would have broken at least three of them.

The fix cost was small everywhere, once the actual router was read first.

Guessing would have cost more.

Third What looked like a naming conflict, two sites, AssetClassLeverage.com and AIAssetLeverage.com, competing for the same 'digital asset' queries, was not a conflict at all.

Krisada runs multiple properties after the same search terms on purpose.

Redundancy crowds out competitors more effectively than one property holding the position alone.

The two sites kept distinct angles, investment framework on one, AI-build methodology on the other, which is what makes the redundancy additive instead of duplicate.

Key Takeaway

What can be reused?

Publishing content because a topic feels timely is a guess.

Publishing content because the warehouse already shows AI systems reading your sites on that exact gap is a plan.

The difference between the two is one query against data you already collect.

Evidence Library

Keep Browsing the Case Library

Move from this evidence file back into the full proof system.