A pulse survey is short and recurring: cadence, question count, examples, the anonymous vs confidential decision, and the tools compared across waves.
A pulse survey is a short, recurring questionnaire — typically three to ten questions — sent to the same group on a fixed cadence so change can be read over time rather than once a year. The point is not brevity for its own sake. A pulse works only when each wave can be matched to the same respondent, because the finding you want is that a person moved, not that an average did. Cadence makes it a pulse; identity makes it useful.
The pattern most teams inherit is a fresh anonymous form each wave, dropped into a new sheet. It produces four averages that look like a trend line and cannot support one. If the people answering in wave three are not the people who answered in wave one, the movement between them may be a change in who responded rather than a change in anyone's experience.
Key takeaways
Every pulse program eventually meets the same fork. Make the survey anonymous and people answer honestly, but no wave can be joined to another, so the program can never say a person improved. Attach names and the waves link, but candor drops on exactly the questions — psychological safety, manager trust, whether the training was worth it — that the pulse exists to ask.
Sopact calls that fork The Anonymity Trade, and treats it as a false choice created by the tooling rather than by the method. The trade only exists when identity and reporting are the same decision. Separate them and both are available: every response binds to a persistent Contact ID so waves link, while reporting is confidential rather than anonymous — results are never shown below a group threshold, and no individual answer is exposed to a manager.
That distinction is a data-model property, not a policy one. A form-centric tool creates a new response record per submission with no stable identity between them, so anonymity is the only way to promise privacy. A record-centric tool holds one participant record that accumulates waves, so privacy can be enforced at the reporting layer while the linkage survives underneath. The identifier design that makes the link hold is on longitudinal survey; writing the questions themselves is on survey design.
The pulse survey is a reaction to the annual engagement survey, which arrived once a year, took six weeks to report, and described a workforce that had already moved on. The first era of tooling was the annual instrument itself: long, census-style, and analyzed by a consultant.
The second era was the engagement platform — Culture Amp, Lattice, Officevibe and others — which made recurring short surveys routine and normalized the cadence. These are competent products and they solved the frequency problem. Most are built around anonymous aggregate reporting, which is a deliberate and defensible design choice for HR engagement work, and it is also the reason a participant-level trajectory is usually not available.
The third era is the repurposed form builder. Google Forms and Typeform are used for a large share of pulse programs because they are free or already in the stack. They collect the responses and stop there: each wave is a separate form with a separate sheet, and joining them is manual work someone does in a spreadsheet.
The one test that separates the eras: ask whether the tool can show you a single named participant's answer to the same question across all four waves, without an export. If it can only show you wave averages, you have a survey tool. If it can show the trajectory, you have a pulse program.
A workable pulse is three to ten questions on a cadence matched to how fast the thing you are measuring can move — weekly for an active cohort, every two to six weeks for a training program, quarterly for organization-wide sentiment. Faster than the underlying change and you measure noise; slower and you find out too late to act.
Keep the core questions identical across waves. A pulse earns its value from comparability, so the two or three anchor questions should be worded the same every time, with any rotating questions clearly separated from them. One open-ended question per wave is usually enough — it is where the explanation for the number lives, and coding it is covered on qualitative data.
A concrete shape: a 320-participant workforce cohort running four waves — intake, week six, week twelve, and 90 days after exit — with the same three confidence questions and one open comment each time. That produces 320 trajectories rather than four averages, and it is what makes it possible to see that the week-twelve dip is concentrated in one site. The stage below runs a single wave the traditional way and as a loop.
Anonymous pulse surveys cannot be joined across waves; confidential pulse surveys link every wave to a persistent participant ID while withholding individual answers from reporting. Read the last column — the difference is not privacy, it is whether a trajectory exists at all.
Vendor capabilities below reflect each product's public documentation and typical configuration, and they change; verify current behavior before buying.
| Tool | Built for | What happens between wave 1 and wave 4 |
|---|---|---|
| Google Forms | Free ad-hoc collection | Separate form and sheet per wave; joining is manual |
| Typeform | Well-designed one-off forms | Per-form responses; cross-wave linkage not the model |
| Culture Amp | HR engagement programs | Recurring waves with anonymous aggregate reporting |
| Lattice | Performance and engagement | Engagement pulses tied to HR records, aggregate reporting |
| Officevibe | Team-level manager pulses | Frequent short pulses, anonymity by design |
| Sopact | Program cohorts tracked over time | Every wave on one persistent Contact ID; confidential, not anonymous |
The engagement platforms are good at what they were built for, and anonymous aggregate reporting is a reasonable choice for organization-wide HR sentiment. It is the wrong choice for a program cohort, where the question is whether a specific participant improved. That is The Anonymity Trade resolved the other way: confidential reporting over one persistent Contact ID, so the trajectory exists and no individual answer is exposed.
A pulse report is the summary produced after each wave of a pulse survey: the current scores, the movement since the previous wave, the themes from the open-ended responses, and the subgroups that moved differently from the average. A useful one is issued within days of the wave closing, because a pulse report that arrives a month later is an annual report on a faster schedule.
The difference between a pulse report and a survey summary is the second column. A survey summary states this wave's numbers; a pulse report states the change and attributes it — which cohort, which site, and what people said about it. That requires the waves to be linked, which is why the reporting format and the data model are the same problem. Turning wave results into a decision is on survey analysis, and pairing a pulse with an intake and exit instrument on pre-and-post surveys.
The entire premise of a pulse is acting between waves. If wave two is still being cleaned when wave three goes out, the cadence is decorative. Reading responses as they arrive is what makes the interval actionable. That is the premise of the Loop, Sopact's method for continuous impact intelligence: collect clean at the source, analyze the moment data arrives, improve while you can still act.
The Loop is also what keeps a pulse comparable. The same anchor questions, the same codebook for the open comment, and the same participant identity every wave — so a movement between waves is a real movement and not a change in method. That standard has its own chapter in reliability and reproducibility.
One method, three moves that never stop
Then the cycle runs again, a little sharper each wave. Read the method: the Loop methodology →
The fastest way to test a pulse design is to write one wave and check whether it can be joined to the last one. Each prompt below pastes into Sopact Sense's Assistant, or reasons through with your team; the arrow above each links the Academy walkthrough that shows the expected output and the tips.
Academy walkthrough → Build a four-wave pulse instrument
Design a pulse survey for this program: [PASTE PROGRAM + COHORT SIZE + DURATION]. Give me three anchor questions worded identically for every wave, one rotating question per wave, and one open-ended question. Recommend a cadence and justify it against how fast the outcome can actually move. Flag any question that cannot be compared across waves. Return a table: Wave / Question / Anchor or rotating / Why it is here.
Academy walkthrough → Test whether your waves can be joined
Here are two waves of pulse data: [PASTE WAVE 1 + WAVE 2]. Tell me what share of wave 2 respondents can be matched to a wave 1 response, which matching key is doing the work, and where the match fails. Then tell me whether the observed change is a real movement in matched respondents or an artifact of a different mix of people answering. Return: Match rate / Key used / Failure causes / Matched-only change vs raw change.
Academy walkthrough → Read the open comment while the wave is warm
Theme the open-ended responses from this pulse wave against my codebook: [PASTE CODEBOOK + RESPONSES]. Give the theme distribution, the two themes that moved most since the previous wave, and three verbatim quotes per moving theme. Do not invent themes that are not in the codebook — mark anything unclassifiable as UNCODED.
Academy walkthrough → Fix the identity problem before wave one
Review this pulse collection setup for cross-wave linkage: [PASTE FORM FIELDS + HOW IT IS DISTRIBUTED]. Identify every reason a wave-3 response might fail to join to the same person's wave-1 response, including email changes, shared devices, and anonymous submission. Then give the minimum changes that make linkage reliable without making respondents feel identified. Return a table: Risk / Why it breaks linkage / Fix.
Each walkthrough is short and practical: what to do, the prompt to run, the output to expect, and the tips that keep it reliable.
Watch: collecting quantitative and open-ended responses together at the source — the setup that makes each pulse wave joinable to the last.
A pulse survey is a short, recurring questionnaire — usually three to ten questions — sent to the same group on a fixed cadence so change can be tracked over time instead of once a year. The defining feature is not length but repetition against a stable group. Sopact's position is that a pulse only works when each wave binds to the same participant under a persistent Contact ID, because otherwise you have a series of averages rather than a trajectory.
A pulse report is the summary issued after each wave of a pulse survey: current scores, movement since the last wave, themes from the open-ended responses, and which subgroups moved differently from the average. The word 'pulse' in pulse report refers to the recurring cadence — it is a reading taken at an interval, not a one-time study. Sopact issues it within days of a wave closing, because a pulse report delivered a month later is an annual report on a faster schedule.
Three to ten, with two or three anchor questions worded identically every wave. A five-question wave answered by 80% of a cohort is more useful than a forty-question wave answered by 30%, because response bias does more damage to a pulse than a small sample does. Sopact recommends keeping the anchor questions fixed and clearly separating any rotating question so comparability across waves is never in doubt.
Match the cadence to how fast the thing you are measuring can actually move: weekly for an intensive active cohort, every two to six weeks for a training program, quarterly for organization-wide sentiment. Faster than the underlying change and you measure noise; slower and the finding arrives too late to act on. Sopact treats the interval between waves as the working period — the pulse exists so something can change before the next one.
Confidential is usually the better choice for program cohorts. Anonymous responses cannot be joined across waves, so no participant trajectory exists and the program can never show that a person improved. Sopact calls this dilemma The Anonymity Trade and treats it as avoidable: bind every response to a persistent Contact ID for linkage, and enforce privacy at the reporting layer by never showing results below a group threshold. Genuine anonymity remains the right call for whistleblowing or high-risk disclosure.
An engagement survey is typically an annual, census-style instrument covering many topics; a pulse survey is short, recurring, and narrow. Engagement platforms such as Culture Amp, Lattice and Officevibe made recurring pulses routine and generally report anonymously in aggregate, which suits organization-wide HR sentiment. Sopact is built for the other case — a program cohort where the question is whether a specific participant improved, which requires the waves to link.
A pulse survey is defined by short length and frequent cadence; a longitudinal survey is defined by following the same subjects over a longer horizon, often with longer instruments and wider gaps. They share one requirement — the same person identified at every wave. Sopact treats the identifier design as the shared foundation and covers it in depth on the longitudinal survey page, while this page covers cadence, brevity, and the anonymity question.
You can collect the responses, but each wave is a separate form with a separate response sheet, so joining wave three back to wave one is manual spreadsheet work repeated every cycle. That is workable for two waves and breaks down by the fourth. Sopact binds each wave to a persistent Contact ID at collection, so the join the analyst usually rebuilds by hand already exists.
Keep two or three anchor questions on a consistent scale — for example confidence in applying a specific skill, clarity about what is expected, and whether support is available when stuck — plus one open-ended question asking what would most improve the next few weeks. A concrete example: a 320-participant workforce cohort running waves at intake, week six, week twelve, and 90 days after exit, with the same three questions each time. Question wording and scale design are covered on Sopact's survey design page.
Attrition is the main threat to a pulse program, and it is rarely random — the people who stop answering are often the ones whose experience is deteriorating, which biases the trend upward. Keep waves short, send them in the channel the cohort already uses, and report the matched-only change alongside the raw change so the reader can see the difference. Sopact tracks who has and has not responded per wave against the Contact ID, so follow-up targets the right people rather than the whole list.
Next: get the identifier design right on the longitudinal survey page, or turn a closed wave into a decision on survey analysis.
