Site Analysis Data Lakehouse
Broad, Deep, and Qualitative Analysis
It’s easy enough to do small-scale, purely deterministic analysis in a spreadsheet with a twist of manual analysis (which we’ve been doing for decades now) or to point a chatbot at a website to answer some simple qualitative questions (new to all of us and it feels so easy!). But that doesn’t answer broad qualitative or large scale questions.
Chimera is a data lakehouse, that pulls together the unstructured pool of content from the site (to allow ad hoc analysis like pattern extraction or LLM categorization) along with highly structured data (across graph DB, vector DB, relational DB, summary analysis in files, and memory DB) leveraged by data pipelines that allow processing at scale (such as passing tens of thousands of documents at once for LLM processing).
Considering big digital presence changes? Contact David Hobbs Consulting.
Answer Questions About Your Digital Presence
-
How coherent is our digital presence across all our websites?
Look at technology use, logos, and other factors across the site and also determine likely other domains that belong to the broader digital presence.
-
We just identified an issue on one of our pages. Across our digital presence, how often does it occur?
-
How should our content be treated in a migration? How much manual effort will there be?
Make decisions about content using rules.
When To Use Chimera
Whether you’re an agency, consultant, or owner of a digital presence will of course influence how you think of the lifecycle of your business with respect to websites, but here are some of the possible steps in your work where Chimera may make sense.
-
Prospecting
- Evaluate a potential prospect against qualities of their website that your organization provides high value
-
Proposing Projects
- If you work inside an organization, then help build the case for projects you would like to implement
- If you provide services to organizations, then use Chimera to help develop your proposals
-
Discovery
- Use Chimera to dig into the realities of the current site to help inform the project
-
Planning
- Make decisions about your content, using rules
-
Relaunch
- Use Chimera to evaluate basic quality (and potentially that old URL to new URL redirects work correctly) after launch
-
Ongoing
- Develop dashboards against your quality rubric over time, slicable by owners of the content
- Chase down the prevalence of issues that are raised
Domains of Analysis
- Business Development
- Content Strategy
- IA & UX
- Migration & Transformation
- Ongoing Digital Presence Operations
- SEO
- Taxonomy
- AI Planning
Use Cases
- Classification of content
- Summarization
- Immediate semantic search
- Qualitative analysis, such as applying quality rubrics
- Sprawl and digital presence coherence analysis
- Evaluating a site against your firm’s unique way of addressing content
- Migration planning and making decisions about content based on rules
- Developing compelling reports to influence change, including dynamic visualizations
How AI and Probabilistic Algorithms Are Used
The first probabilistic algorithm in Chimera was a near-text duplicate detection algorithm added in January 2019. LLMs are now woven into Chimera in several ways — although you can also use Chimera with very limited LLM involvement:
- Connect your own chatbot or use Chimera Chat to interact in English
- Chimera uses an LLM to evaluate whether a crawl is potentially spiraling out of control
- Large-scale ways of batching content for qualitative evaluation
- RAG search across your content inventory
- Chatbots can help you develop patterns to scrape data out of pages — rather than having to hack things like XPath and Regex yourself
What Does It Mean To Be a Data Lakehouse and Why Does It Matter?
Especially nowadays, the temptation is to assume that you can just go into a chatbot and ask it any question. But we also all know that LLMs just make things up. By having a deeply structured database along with unstructured data — especially a cache of the digital presence — ready for analysis that hasn’t even been defined yet, you’re able to better ground LLMs.
By having a variety of database types and tables optimized for different needs, much of the analysis can be faster. Graph databases reveal link relationships and site structure. Vector databases power semantic search and similarity detection. Relational databases handle the structured, quantitative data. Together they give you a foundation that a generic chatbot simply cannot match.
Content Chimera isn’t magic. David Hobbs Consulting built this tool to support broader processes of major digital presence change, which includes people, processes, and models — and also to apply the skills acquired over decades of data analysis experience. Contact David Hobbs Consulting to help use Content Chimera to drive your complex transformations forward.
How Does Chimera Compare?
| Capability | Screaming Frog & Spreadsheets | Generic Chatbot | Content Chimera |
|---|---|---|---|
| Simple, technical analysis | ★★★ | ★★★ | ★★★ |
| Qualitative analysis on single page | — | ★★★ | ★★★ |
| Large scale single-pass analysis | ★ | — | ★★★ |
| Weave in data from other sources reliably | ★ | ★ | ★★★ |
| Ability to iterate on analysis | — | ★ | ★★★ |
| Qualitative analysis on entire digital presence | — | — | ★★★ |
| Over time analysis | — | — | ★★★ |
| Deep context of data passed to LLMs | — | — | ★★★ |
| Visualization | — | — | ★★★ |
| Interactive reports and dashboards | — | — | ★★★ |