The Question Bench, 14 platforms by 40 questions, edition 8 Nov 2026
Levels: out of the box 3, built without an engineer 2, analyst with export 1, not answerable 0; summed over 40 and shown as 'answers per 40' (maximum 120, shown as a count of questions answered at level 2 or 3 and the points). Column numbers link to the question pages. Read 8 Nov 2026.
Levels: 3 answered out of the box; 2 answered by building a report or view without an engineer; 1 answered by an analyst with an export; 0 not answerable from the platform's own data. Level 3 cites a delivered report, dashboard, template, hub view or release note for that question. Level 2 cites a documented no-code builder or natural-language query on the platform's own data. Level 1 means the data is in the platform but reaching the answer needs a report writer or an export. Level 0 means the platform does not hold the data. The bench is downloadable as CSV and JSON on the datasets page.
Building the tool, view or workflow you need by describing it, and handing whole processes to agents that do nothing until you approve. Questions that need payroll or ATS data are answered through integrations, not from Shapes's own records.
Payroll, IT and HR in one system with policies keyed to any employee attribute and an AI that asks before it acts. Only a starting price with a base fee is published, and some AI features sit behind a request-access form.
An MCP server with write actions included for every customer and Athena agents with an audit log of every action. No price is published for any plan and setup runs from days to months.
Fourteen modules across HR, finance and IT with an AI agent that asks before it acts. Only a starting price is published, no implementation time is published, and its MCP server is added by URL rather than listed in the Claude or ChatGPT connector directories.
Answers per 40 counts questions answered at level 2 or 3; the rubric total is the seven-criterion score with this site's weights. The Method page reconciles the two.
People analytics used to be a module: a dashboard page with twelve tiles the vendor chose. In 2026 it is a question and an answer. The CEO asks for attrition by manager in the meeting, and either the system answers from its own records, or a non-engineer builds the view by describing it, or an analyst exports a CSV, or the number is not in the system at all. The bench scores those four outcomes for forty questions and fourteen platforms, from the vendors' own reporting documentation, template libraries and release notes.
The trap in most people questions is the data, not the chart: attrition by manager is wrong if leavers are counted against today's org chart; pay equity is wrong without level; leave liability is wrong without the daily rate. Each question page names the trap, the data the answer needs and which platforms hold it. Questions owned by payroll, applicant tracking or learning systems are outside the bench, and the bench does not measure them. That is also where Shapes's weakness sits: it answers them through integrations (seven live payroll and six live ATS) rather than from its own records.
The Build Gallery
What a non-engineer can build on each platform and how, with named examples from the vendor's own library or templates, each with a source and a date. Shapes's hub names the views; most other vendors name the builder.
A flight risk view scores each person on signals the record holds (tenure, time since last raise, manager changes, engagement, absence) and lists the people above a threshold with the reasons. On 8 Nov 2026 Shapes, HiBob and Workday deliver one; seven platforms build one from a prompt; three need an export; Gusto cannot.
A pay equity view compares pay by gender or group within the same level, job family and location, so that the gap it shows is the one a consultant would find. No platform delivers it out of the box on 8 Nov 2026; nine build it from a prompt or builder; four need an analyst; Gusto cannot.
A leave liability view multiplies each person's unused balance by their daily rate and sums it by entity, which is the number finance accrues. No platform delivers it on 8 Nov 2026; eight build it where the record holds a rate or a salary; Workday, Gusto, Lattice, Sage, ADP and Paylocity need an analyst.
A span of control view counts direct reports per manager, flags spans outside the policy range, and shows layers from the top. Shapes delivers it (the manager directory in the hub) and Workday delivers it; HiBob, Rippling, Personio, Deel, Lattice, Factorial and Humaans build it; BambooHR, Sage, ADP and Paylocity need an analyst and Gusto cannot, on 8 Nov 2026.