What it is
Readiness Check scores how well a client's chatbot answers the questions people actually ask it. It sends the chatbot a fixed set of 100 real questions, grades every answer, and returns a score plus a ranked list of what to fix.
Nothing about the chatbot or the client's content changes when a check runs. It only asks questions and grades the answers that come back.
Why clients will care
Clients keep asking two things: how is my chatbot doing, and how do I make it better. Until now there was no number to point at and no clear next step.
It gives them a baseline score, so improvement becomes something they can measure.
It shows exactly which content is thin, and what fixing it is worth.
At onboarding, it confirms their AI resources are working before they go live.
After onboarding, re-running it catches content that has gone stale.
What gets tested
Each chatbot profile is tested against 100 real world questions covering the areas people ask about most. The set is fixed, so the same questions are asked on every run. That is what makes two runs comparable and lets a client see whether their work moved the score.
If the account uses department modules, each module has its own set of 100 questions and its own separate results.
How an answer is graded
Every question has a checklist of what a good answer should contain, split into two parts:
Information coverage: the key facts the answer should include, such as eligibility, deadlines, and required documents.
Actionability: the next step the answer should give, such as a form, a phone number, or an office to contact.
Each checklist item is graded met, partly met, or missing. It is not pass or fail, so an answer gets credit for what it does cover. This is also why an answer can be factually correct and still score poorly: if it never tells the person what to do next, it loses the actionability half.
How the score is built
Question scores roll up into a department score, and department scores roll up into the overall score. Every number comes from the same question level grading, so the overall score, the department scores, and the individual questions always agree with each other.
The score bands are the same at every level:
80 to 100 is a Strength.
65 to 79 is Moderate.
Under 65 is a Weakness.
Any single question scoring under 50 is counted as a critical issue.
What a client gets back
An overall readiness score, plus separate scores for information coverage and actionability.
A score for every department, ranked weakest first, so it is obvious where the gaps are.
For any question: the answer the chatbot actually gave, which checklist items it met and missed, and which sources it used.
A run history for each profile, showing whether the score moved and by how much.
The Improvement Plan
This is the part clients will act on. It turns the score into a ranked to-do list of the questions holding the score back.
It includes only questions scoring under 80, grouped into high, medium, and low priority.
Every item shows what it is worth in points. A point is what the overall score would gain if that answer scored full marks, so the same fix is worth more in an area people ask about a lot. Clients can see the payoff before they do the work.
Sensitive topics, such as safety, are flagged and listed first within their priority group.
Each item leads straight into creating the resource that fixes it, with the right department already filled in.
Re-running the check
A score only changes when the underlying content changes. Once a client adds or updates a resource, the check becomes available to run again, and the new run shows whether the change worked. Running it against unchanged content would only return the same score, so it stays unavailable until there is something new to measure.
What clients can control
Clients choose which departments, issues, and questions are in scope for a run, so the score reflects the areas their chatbot is actually meant to handle. They can also adjust the checklist a question is graded against.