Analytics
PULSE, content health and alerts.
Nova Guide reports which guidance is used and completed, which steps have stopped finding their controls, and, where an organisation chooses, how often a workflow is completed. Every figure is shown with the division that produced it, and no event carries patient data.
1
PULSE
Five measures of how guidance is used: Proficiency, Uptake, Loyalty, Satisfaction and Engagement. Each is a single division, printed beside its result.
| Letter | Measure | Formula | Notes |
|---|---|---|---|
| Proficiency | guides completed ÷ guides started | How often a guide that is started is finished. | |
| Uptake | live guides used at least once ÷ live guides | Counts guides that were live by the start of the period. | |
| Loyalty | people who returned ÷ people active in the previous period | Counted per person, never per workstation, and split between returning for a new guide and repeating the same one. | |
| Satisfaction | answers of “helpful” ÷ all answers | One tap at the end of a guide: Yes or No, with no rating scale and no comment box. | |
| Engagement | people who used more than one guide ÷ people who used any | Counted per person. |
- 1.1Formulas
- Every score is printed with its formula and its fraction, so any figure can be reconstructed by hand.
- 1.2Composite
- The mean of every measure that could be calculated for the period, equally weighted. It says how many of the five measures it stands on.
- 1.3A measure without data
- Keeps its place, says why it could not be calculated, and is left out of the composite.
- 1.4Small numbers
- Ratios over fewer than five people or events are withheld, because a ratio over so few can identify the individuals behind it.
- 1.5Comparisons
- Each figure is compared with the previous period in a sentence, such as “This is up from 5 starts in the previous 30 days.” A period without a comparison says so in words.
- 1.6Attribution
- A card shows what share of events could be attributed to an individual, which decides how much Loyalty and Engagement can say.
- 1.7Charts
- Every chart can be viewed as a table, and its colours are chosen to be told apart with common colour vision deficiencies.
- 1.8Per guide
- The analytics screen lists each guide with how its sessions ended and its median time, and each guide’s record shows how staff answered the helpfulness question.
2
Content health
Content health lists the steps that are failing, worst first, so the content team knows what to fix when an application changes.
- 2.1What it counts
- For each step: how often its control was not found, how often it was ambiguous, and how often it was matched through a weaker signal, with the lowest confidence seen.
- 2.2Order
- Worst first, for each application.
- 2.3Weak matches
- A step that still works but matches through a weaker signal is listed too. It is often the first sign that an update has changed the application’s screen.
- 2.4Overview
- The portal’s overview lists the steps failing to find their control across every application, beside open alerts and content waiting for review.
- 2.5Silent applications
- An application that is published but sending nothing is marked Silent, in amber. That usually means the extension is not deployed, or nobody has opened a guide.
3
Alerts
A step that keeps failing opens an alert for the people who maintain the guidance.
- 3.1When
- One step of a live guide fails to find its control five times within eight hours, counting both missing and ambiguous controls.
- 3.2Where
- On the portal’s overview and on the guide’s own page. One alert per step is open at a time.
- 3.3Who is told
- The alert is emailed once to the organisation’s admins and moderators who have signed in and have an address. The portal’s settings say who is emailed.
- 3.4Closing
- An alert closes when an admin or moderator marks it fixed, when a revision is published, or when the guide is unpublished or withdrawn. A step going quiet closes nothing.
- 3.5What does not count
- Weak and slow matches do not open an alert, and failures from a version that a revision has replaced open nothing.
4
Workflow analytics
Workflow analytics shows whether a workflow is being completed, with and without guidance. It is off until an administrator turns it on.
- 4.1Declared tasks
- An author records the steps of a workflow in the studio, as for a guide, and a reviewer approves it. A task shows nothing to anybody.
- 4.2What is recorded
- When each step of the task is reached, when the task is completed, and which screens are viewed, by path.
- 4.3What it reports
- How often the workflow is completed, the step where people stop, completion with and without a guide open, and the median time taken.
- 4.4Screens used
- Which screens of the application people are on, how often guidance was open there, and how long people stayed.
- 4.5Only the task’s steps
- Nova Guide watches for the task’s own steps and records nothing else that is clicked. Describing a clicked element on an EPR screen can mean recording a patient’s name.
- 4.6Switching it on
- Off until an administrator turns it on in the portal’s settings. It is a separate processing purpose, so it comes with its own note for your information governance lead.
5
What is recorded
Analytics in Nova Guide is built from a short list of events, defined so that patient data has nowhere to go.
Recorded
- Guide started, step completed, guide completed or abandoned
- Helpful and unhelpful answers
- Step not found, ambiguous, matched through a weaker signal, or found slowly
- Tooltip viewed; beacon clicked or dismissed; field hint shown
- Session started
- With workflow analytics on: task step reached, task completed, screen viewed
Never recorded
- Patient data of any kind
- Text from the application’s page
- Anything typed into a field
- Web address parameters; screens are recorded by path only
- Free text from staff: helpfulness is Yes or No
- 5.1Enforced by design
- The event definitions have no field that could carry page text, typed values or web address parameters.
- 5.2Pseudonymisation
- Staff identifiers are replaced on arrival with HMAC-SHA256, using a key held for each organisation.
- 5.3People and workstations
- Counted separately. Where Nova Guide cannot tell who is using a workstation, it counts the workstation and never claims a person.
- 5.4Preview
- Previewing a guide in the studio records nothing.
- 5.5Retention
- Events are kept for 24 months by default, then deleted by a nightly scheduled job. The period is agreed with each organisation.
How data is stored, protected and hosted is on the Security and data protection page.
See a guide recorded, approved and published.
A demonstration records a guide, takes it through review and publishing, and shows it as staff see it. It also covers how Nova Guide would be deployed and assured in your organisation.
- RecordingA guide recorded in Nova Guide Studio, click by click.
- Review and publishingApproval by a named reviewer, then the guide going live.
- On the screenThe guide as staff see it, started from the launcher.
- Withdrawal and PULSEWithdrawing a guide, and the analytics behind it.
- Your organisationDeployment, data protection and the clinical safety material.