Constructing an independent search intelligence moat

Enterprise SEO teams are typically constrained by expensive, heavily rate-limited third-party APIs that obscure raw historical rankings behind high SaaS paywalls.

SERP Sweden was engineered as an in-house OLAP intelligence platform for the Swedish market, flattening 52GB of nested SERP data into columnar MergeTree tables capable of instant multi-dimensional aggregation.

Columnar MergeTree storage & sub-50ms autocomplete

Storing 295 million records with 291 columns required strict schema discipline. By sorting primary keys on (keyword, location) and using LowCardinality string encodings for search intent and result types, memory footprint was reduced by over 80%.

The frontend leverages Next.js 16 App Router and Turbopack to query ClickHouse sort key prefixes directly, delivering sub-50ms typeahead autocomplete without dedicated search middleware.

Semantic topical maps & Wikipedia entity graphs

Rankings without semantic context provide little strategic direction. SERP Sweden integrates continuous MediaWiki API pagination, resolving thousands of internal and external entity links to construct Koray Tuğberk Gübür-style topical authority maps.

This allows technical SEO architects to visualize topic clustering, discover content voids, and calculate domain-level Share of Voice across Swedish search queries.

What success looks like

SERP Sweden demonstrates that a solo engineer can build enterprise-scale data infrastructure rivaling major SaaS providers when leveraging high-efficiency columnar databases and modern frontend runtimes.

The result is total independence from third-party API rate limits and unmatched speed in Nordic search market intelligence.

← Back to all systemsRead field notes →