Case study: Moodi,
four databases to one.
Moodi is a media intelligence platform that reads video frame by frame, audio track by track and images for colour and composition, and tells brands how an audience will feel about what they publish. In a single month that is millions of new rows and relationships. Moodi ran it on four separate database engines; it now runs it on Data Oil.
The outcomes, as Moodi reports them
One engine where there were four
| Measure | What changed |
|---|---|
| Database costs | 40% lower overall operational database expenses |
| Infrastructure charges | 80% lower total cloud infrastructure spend |
| Stack complexity | 80% less architectural and schema complexity |
| Database footprint | From four separate database engines to one |
| Delivery speed | 40% more customer projects completed |
These are Moodi's figures, from its own case study, not Data Oil's measurements.
Before
Four engines, joined in the application
A relational database for metadata, accounts and billing, a document store for media tags, a search engine, and a store for relationships. Linking a second of audio to the colour palette of the frame it sits under meant joining across them in Moodi's own code, keeping four copies in step, and assembling the context for its AI models from four places before every prompt.
With Data Oil
Rows, documents, search, vectors and relationships in one place
The relational schema, the JSON documents and the search index moved into one engine, with semantic search over the same data rather than beside it. The pipelines that copied data between databases went, and so did the joins across them. The context for Moodi's own models and for third-party models is retrieved in one place, and people reviewing results feed their corrections straight back into the same data.