This platform has three tiers: the questionnaire, open to everyone; the dashboard, which takes the report viewer passphrase; and the builder and data console, which take the administrator one. Either passphrase works here and opens whichever tier it belongs to.
This is a front-of-house lock, not security. The questionnaire is a single HTML file, so anyone who opens it in a text editor can read both the passphrase and any data loaded into it. Keep confidential response files out of the copy you circulate.
The sentiment dashboard shows scored results for the whole population. Enter the passphrase to continue.
A front-of-house lock, not security — the same caveat as the administrator passphrase, and for the same reason: there is no server behind this file to check a credential against. What it does buy is that results shown here are already suppressed below the minimum reportable cell, so an unlocked dashboard screen still cannot expose an individual.
Each answered item is rescaled (v-1)/4*100. A dimension score is the mean of its answered items. Reverse-keyed items are flipped before rescaling. 'Prefer not to answer' is excluded from the mean, not treated as neutral.
Pipeline Row. This is the CSV line this response would contribute, in the column order of the response template on the Data tab.
These carry through the questionnaire, the templates and the analysis.
{Worker} becomes whatever the workforce is called.cycle column so pulses can be stacked.No peer-reviewed study identifies an optimal response rate. The research literature deliberately declines to set one, and it is worth knowing why before quoting a number to a client.
What this means in practice. A census employee survey carried inside an organization, with visible leadership endorsement and time set aside to complete it, routinely reaches far higher than the academic averages above. Targets in the 70–80% range are common in professional practice and in scopes of services. They are contractual commitments and quality signals, not validity thresholds derived from research, and they should be described that way to a client.
Set the field above to whatever this engagement is held to. Treat falling short of it as a trigger to state a caveat and examine who did not respond — which is the part the research actually supports.
Three tiers, each opening a different screen. Set a passphrase pair for this engagement rather than shipping the same two words to every client — and read what these locks can and cannot do before telling a client the file is protected.
The firm’s mark stays to the left of the title; the client’s sits to the right of it. Two colours are enough — every other shade in the palette is derived from them, so the tints, rules and chart ink stay in step instead of leaving a client to specify nine values that have to agree with each other. Logos, colours and any brand font are embedded in the file itself, so a branded copy still opens with no internet connection.
.woff2, .woff, .ttf or
.otf is embedded and then offered in the lists above, so the client’s own
typeface renders on any machine. Check the licence permits redistribution before embedding
it. The client’s colours carry through the platform chrome, the dashboard and the leadership delivery build. Three things deliberately do not follow them.
Question text is stored with placeholders rather than fixed nouns, so the same
instrument can be pointed at a city, a hospital, a hotel or a firm without anyone retyping
questions. Pick a sector to set everything at once, then adjust individual terms below.
Column names in the exports do not change — department stays
department — so data from different clients still stacks.
The list of the questionnaire offers. This is client
data rather than wording — a sector preset changes whether the question says
“department” or “service line”, never which ones exist. One per line as
code | Label; omit the code and one is generated from the label.
department column of every data file. Changing a code on a live engagement
orphans any data already collected under the old one, so fix the list before fielding.Tick an item to put it on the survey. The questionnaire, the progress counter, the scoring, the response template and the analysis all follow this selection — nothing has to be kept in step by hand. Every section also takes questions the client writes: a client question is asked, scored, exported and reported exactly like a bank question, and its column code is checked here rather than discovered to clash after the survey has been fielded.
Both templates are generated from the survey as you have built it, so the columns always match the questions being asked. Rebuild the survey and the templates change with it.
One row per respondent. Rated items take 1–5 or
blank for “prefer not to answer”. Multi-select answers are separated by
semicolons. Two example rows are included — delete them before loading real data.
Peer medians per dimension, peer organizations, and the peer quartile band —
the only figures on the dashboard that cannot come from your own responses. Leave a
value blank and that mark simply stays off the chart.
Files are read in the browser. Nothing is uploaded, and nothing is kept when you close the page.
Response File — drop a CSV here, or click to choose one.
no response rows loaded
Roster File — drop a CSV here, or click to choose one.
no roster rows loaded
Benchmark File — drop a CSV here, or click to choose one.
no benchmark file loaded
This file’s headers do not match the response template, which is normal — survey platforms export question text rather than item codes. Each expected column is matched to a detected one below. Check the matches before applying.
Writes a standalone dashboard file for the client — the same dashboard, carrying aggregated and already-suppressed cells and nothing else. Withheld values are not written into the file, so it cannot disclose them to anyone who opens it in a text editor. No respondent-level row and no verbatim comment is present. It opens in any browser offline and is safe to email or archive.
Three checks on the response file, reported rather than applied. Every one of them has a false-positive story — a crew whose experience really is uniform will straight-line honestly, and a deliberately blank demographic block is the consent promise working. So nothing is dropped unless you say so, and if you do, the figure that leaves the firm has to say so too.
Which dimensions move together with one chosen outcome, and how much of what can be explained each one accounts for. Run on the whole population: the model needs every response it can get, and a driver model fitted to a single would both break the minimum-cell rule and fit noise.
Dimension means across everything loaded, plus any responses entered on the Survey tab. Groups below the minimum cell of 5 are withheld.
