The Toolbox Store
A store of small, sharp capabilities you add to Google Sheets, Excel, Power BI, or Tableau — not yet another platform to migrate to. Each one is an export of a real, tested toolbox capability: built once, never rebuilt. Posted prices, honest availability, no quotes.
5 Plug-in Packs · 2 Field Kits · 21 drop-ins · 6 available now
How the store works
A is an ingredient: one sharp capability you drop into the tool you already use. A is a recipe that bundles several packs into an ordered workflow for a whole task. Every capability is an export of a live toolbox spoke — the same engine behind our concierge products — so you’re never the first to run it.
We label the way we’d want it labelled to us: available means it ships today; made-to-order means the engine is live and packaging it for your host is a few days; roadmapmeans we haven’t built it yet and won’t pretend otherwise. When something is made-to-order, the card carries an honest estimate with a range — never a bare number. Or, when it's a whole-org job, hand it to a we run on your data, org-wide.
Sheets, Excel, Power BI, Tableau. No migration, no new platform to learn.
Each capability exports a tested toolbox spoke — one core, with a thin host adapter.
Available, made-to-order, or roadmap — labelled straight, with error-barred estimates.
The store
Filter by the problem you’re solving, the tool you work in, or how ready a capability is. Each card shows what it solves, where it runs, and — for recipes — the packs it bundles.
7 of 31 capabilities · Statistics
Ingredients
Single-purpose capabilities that drop into the analytics tool you already use. Each one is an export of a live toolbox capability — built once, never rebuilt.
Split a mean pay gap between two cohorts into the part explained by measurable factors and the part that isn't (Oaxaca–Blinder).
Honest estimate: Packaging into a host adapter: ~2–5 working days (the decomposition endpoint is live).
Solves
exports calculus spoke
Run a seeded, reproducible Monte-Carlo simulation over your own distributions and a formula — uncertainty as a range, not a single number.
Honest estimate: Packaging into a host adapter: ~2–5 working days (the Monte-Carlo engine is live).
Solves
exports forecasting spoke
Pre-register a hypothesis, randomly assign treatment deterministically, and measure lift with a bootstrap CI — tamper-evident, so the result is credible.
Honest estimate: Packaging into a host adapter: ~3–6 working days (pre-registration + assignment + lift endpoints are live).
Solves
exports program-evaluation spoke
Boil a collaboration network down to interpretable measures with benchmarks — reach, cohesion, key-person risk, siloing — not an unreadable hairball graph.
Honest estimate: Packaging into a host adapter: ~3–6 working days (the ONA-measures + connectivity engine is live).
Solves
exports network-analysis spoke
Pick a defensible study design, check your assumptions, draw a reproducible sample, and run the right two-group test — with the canon cited.
Honest estimate: Packaging into a delivery surface: ~4–8 working days (the rigor catalog + compute endpoints are live).
Solves
exports research-methods spoke
Build a latent-factor model, validate it on holdout with a fairness check, and only publish the new version once it passes — then score subjects against it.
Honest estimate: Packaging into a delivery surface: ~5–10 working days (build/validate/publish/score endpoints are live).
Solves
exports factor-models spoke
Look up a published Bayesian prior for a construct→outcome relationship, fuse it with your own observation, and size the value of more data — never a fabricated number.
Honest estimate: Packaging into a host adapter: ~3–7 working days (the connector compute is live; rich prior coverage depends on the upstream Principia store).
Solves
exports principia-connector spoke
Describe your problem in plain words. We’ll tell you honestly whether we have a drop-in for it today, can package one in days, or should point you elsewhere.