Recense vs Displayr

What stays the same

• Real survey statistics. Cross-tabulation, sig testing, weighting, derived variables — all first-class.
• SPSS .sav as a primary input. Same starting point, same fidelity on labels, weights, and missing-value handling.
• AI agents in the workflow. Both tools have agents. The wiring is different (see below).

What changes

• Computation runs in your browser. No upload, no server queue, no slow large-dataset round-trips. Tabulation is in milliseconds.
• Agent parity through a designed-for-it Surface protocol. Every screen exposes a Document, an Action vocabulary, and a Snapshot. Agents read and write the same surfaces a human does — the canvas is one shared notion of truth.
• Bring your own model. Connect your Anthropic, OpenAI, Google, or Fireworks key. Switch providers as the frontier moves. No waiting for the platform to ship support for a new model.
• No R underneath the easy surface. Power workflows in Recense don't fall back to a scripting language. Derived variables and formula cells live in the canvas; the agent can build them too.
• Local-first by architecture. Your survey data does not have to leave your browser to be analysed.
• Self-serve and transparent pricing. Sign up, drop in a file, work. No demo call required.

Cost

Displayr Professional is roughly £2,200–2,500 per user per year. Recense Starter is £30 / month — about a fifth of the per-seat cost — with a Pro tier still well below the Displayr line. See the pricing page.

Migration

Bring your .sav files. Variable structure, labels, missing-value handling, and weights are preserved. Displayr's R-based custom calculations don't port directly — but most of what they do is either built-in (sig testing, weighting, nets) or expressible as a derived variable or formula cell. The agent can help reproduce specific custom outputs on request.

Decks built in Displayr's reporting layer don't move. If your deliverable today is a heavily-templated PowerPoint with hundreds of pre-built charts, plan a project to rebuild that as a Recense canvas-as-report. For most agencies that's a couple of days of work and a better artifact at the end.

Where Displayr still wins

Displayr ships some advanced methods (full conjoint, latent-class segmentation, correspondence analysis) that aren't yet first-class in Recense. These are on the roadmap, but if your work runs on them today, that's a timing question worth being explicit about.