Continuous pulse + manager-actionable insights + open-text themes that preserve anonymity. Tally's response rates average 84% — vs the 32% industry baseline most annual surveys deliver.
People teams listening with Tally
Tally clusters open-text comments into themes by topic and sentiment — then enforces a minimum cluster size before showing anything to leadership. Below k=5 it's hidden, full stop. No exec ever sees a comment they can attribute.
"Rotation feels uneven across geos, and the runbook gap means 3 people carry the on-call load on weekends."
"People are actually using the $2k allowance this quarter. Took two cycles, but the friction is gone now."
"IC4 → IC5 jump still feels political. Calibration calendar isn't shared."
"Meeting load down meaningfully — Wed/Thu no-meeting days actually held."
Tally produces a private manager scorecard from each pulse — six dimensions of management quality plus retention risk, compared to org median and the manager's own trajectory. It's a coaching tool, not an HR weapon.
"Tally's manager scorecards surfaced a gap in our engineering org we'd been missing in annual surveys: 8 EMs scored below the 25th percentile on coaching. We trained 28 EMs over Q2 and Q3. Retention is up 14 percentage points YoY. The cost of one of those eng departures alone paid for Tally for 4 years."
Mercury runs monthly pulses across 6 functions. Their People team replaced a 92-question annual survey + 3-month action-planning cycle with continuous Tally signals and quarterly manager calibrations. Pulse response rate sits at 87% twelve months in.
Includes unlimited surveys, themes, and manager scorecards. No per-response charges, no add-on modules.
"We canceled our annual survey contract three months in. Eight pulses gave us more signal than the previous five years of 92-question epics combined."
"The manager scorecard turned my hardest performance conversation of the year into a data conversation. The manager saw the gap before I had to point it out."
"I'd been worried AI themes would feel intrusive. The k-anonymity rule and the way themes preserve nuance changed my mind. Our employees actually trust the tool, which is the whole game."
We enforce k-anonymity at query time (default k=5, configurable up to k=20). Below that threshold, a slice doesn't render — not for the manager, not for HR, not for an admin. We also strip the writing-style fingerprint from open-text comments before clustering (we don't preserve "voice"). For the most sensitive teams, you can require all themes pass a human review before publication. Customers in regulated industries (healthcare, finance) typically run at k=10.
Yes — Rippling, Gusto, BambooHR, HiBob, and Personio are bidirectional (segments + lifecycle events sync automatically). Workday is read-only via SCIM + an integration cluster (~3 days to configure for typical org). We pull org structure, tenure, function, geo, level — and use those as segments. We never write back to your HRIS without explicit per-field consent.
Default: monthly pulse, 6–8 questions, sampled smartly so no one is surveyed more than once per quarter on the same question. Most customers add a comprehensive quarterly check (14–20 items). You can run ad-hoc 1-question lightning polls anytime ("Did this all-hands land?"). Survey fatigue is real — we cap total questions per employee per month and flag at-risk individuals if engagement drops.
A combination of embedding-based clustering (we fine-tune a sentence-transformer on HR-domain text) and a smaller summarization model that generates the theme label. On Enterprise, the entire pipeline runs in your tenant's region — Frankfurt (EU), Virginia (US East), Sydney (APAC). No customer text crosses regions. We don't fine-tune any shared model on customer data; your themes never inform another customer's clusters.
Yes — but with guardrails. You can branch the question library by function, geo, or any HRIS attribute. Tally maintains a validated core (24 questions tied to the management index) that we strongly recommend leaving consistent across teams so cross-org comparisons remain meaningful. Custom questions ride alongside the core. On Growth+, you can author entirely custom indices with our research team's help.
30-day trial for orgs of any size. We'll run your first pulse on the demo call.