Apple's Quiet Scoreboard and the Batch of Open Source Analytics Tools

Two things happened in sports technology this year that are worth putting next to each other. Apple shipped a sports app that does one thing. A group of developers, consultancies and database vendors shipped tools that try to make analytics less dependent on a handful of cloud vendors.

Slate published a piece on 10 June arguing that Apple Sports, launched in the winter of 2024, is a rare case of an app doing exactly what it promises. The piece lists what the app does not have: no news headlines, no streaming guides, no TikTok-style scroll. The author says he uses it for NHL, NFL, college football and college basketball. During the Stanley Cup Playoffs, he notes, he had to turn off automatic Pittsburgh Penguins updates because the app was faster than his streaming package.

Data behind the simplicity

The mechanics are worth spelling out, because they explain why a score-checking app feels different from what ESPN or MLB have become. Apple Sports surfaces team names, logos, records, game times and scores, plus standings. Tap a game and you get a box score, a leaderboard or rosters. The app covers MLS, NWSL and roughly 30 other soccer leagues, including Ecuadorian Serie A and second-division Bundesliga, plus men's and women's tennis and the LPGA Tour. Live Activities can push scores to the lock screen. Betting odds appear on the scoreboard but can be switched off in settings. Advertising is described as occasional, mostly banners pointing to Apple TV.

Slate quotes an Apple executive's original pitch from the 2024 announcement: "We created Apple Sports to give sports fans what they want, an app that delivers incredibly fast access to scores and stats." That is the whole product statement. There is no agentic layer, no summarizer bot, no image generation.

Whether that restraint scales is a different question. The app consumes sports data rather than producing it, and its value depends on licensing deals with leagues and data providers that Slate does not discuss. It also depends on Apple's willingness to keep a product with no obvious revenue attached to it running. Neither of those is guaranteed.

The other side of the market

Meanwhile, the tooling layer underneath sports and enterprise analytics is getting more crowded. On 13 September, a project called plainoldanalytics appeared on GitHub under the handle poundifdef. It is a self-hosted, bolt-on web analytics package for Go applications. Its author positions it as similar in spirit to Umami, Plausible or PostHog, but embedded in the app itself. The core package is storage-agnostic: importing it does not pull in DuckDB or any other backend. Developers pick a storage package, or implement the Storage interface and pass it to plainoldanalytics.New.

The repository shows a built-in dashboard mounted at a path of your choice, a middleware wrapper, and an optional browser session recorder. Traffic flushes to disk roughly once a second, and the README tells developers to call Close() on shutdown to flush anything still buffered. Adapters exist for gin and chi. The documentation warns that the middleware should be scoped to a group, so the analytics dashboard itself is not recorded as traffic.

On 8 September, a software consultancy called Adaca published a different take: adaca-analytics, a self-hosted dashboard for Google Analytics 4 and BigQuery, running on Cloudflare Workers. Daily rollups from the GA4 Data API or a BigQuery export land in a D1 database the user owns, while realtime stays on Google. The README describes six dashboards out of the box, a four-step builder, saved segments, read-only share links, comparison against previous periods, and weekly and monthly summaries by email or Slack. Setup requires a Google service account with Viewer access and a deploy to Cloudflare, which forks the repository and provisions D1 and KV.

The project is MIT licensed. Adaca says it is a software consultancy that builds custom software and embeds senior engineers in client teams.

Architecture arguments

Starburst published an analysis on 11 September of two architectures for agentic data analysis, and its framing applies to both of the tools above. The first option is to extract data from source systems and hand the raw files to an agent. Starburst calls this "option 0" and argues it is impractical outside unusual circumstances, because agents that charge per data item become prohibitively expensive when you send terabytes. The second option is to give the agent direct access to the database and let it write its own SQL, iteratively refining queries while the database engine does the processing. Starburst notes that agents can overwhelm database systems with speculative queries, but says the industry is focused on supporting agentic workloads.

That is the trade-off the new tools are quietly making. Adaca keeps the heavy aggregation in a database the user controls. Plainoldanalytics keeps traffic in the application's own process. Neither sends raw data to a third party by default. In a market where the loudest sports tech headlines are about AI targeting and prediction systems, the more interesting shift may be toward smaller tools that refuse to do more than one job.