Use case · Pilot design
Design the pilot
both sides can judge
A pilot nobody can fail is a pilot nobody can pass. AnyCRM proposes the channel from the evidence, writes exit criteria as lines both sides sign, and finds the missing baseline before week one rather than at week twelve.
- Channel and scope from the evidence
- Exit criteria both sides sign
- The baseline found before week one
Deal economics
Sample dataOpen
AnyCRM. Deal economics.
Proposed
The pilot, proposed from the report, channel, scope and length. Channel: inbound chat on billing queries, The theme their own customers raise most. Scope: two intents, one language, Small enough to run, large enough to read. Length: twelve weeks, two review points, Week four and week eight
Exit criteria
What both sides agree counts as passing, written before it starts. Containment against the agreed line, measured weekly. Resolution quality, sampled and scored, human review. Escalation handover time, against their own current handling Shapes, not promises. Each line is the level the pilot has to reach, agreed by both sides in writing.
Risk
There is no baseline for the channel you chose. Billing chat is handled by a team that does not report separately, so week twelve would end in an argument about whether anything improved. No baseline.
Your call
Three moves offered: Add two baseline weeks before the pilot starts; Run on email, where they already report separately; Run in shadow for a fortnight and use that as the line. The plan rewrites itself around the choice.
Resolved
Two measurement weeks added at the front, Their team keeps working, the numbers get captured. Exit criteria re-expressed against that baseline, Every line now has something to be compared to. Week fourteen ends in a decision, not a debate. Baseline agreed. Exit criteria signed by both sides.
Deliverables
Pilot design, Channel, scope, length, review points and owners. Exit criteria sheet, Each line, how it is measured, who reads it. Review points, Week four and week eight, on both calendars. Filed on Larkspur Utilities, activity logged, both review points on the calendar.
Close
Design the pilot so it can be failed.
Larkspur Utilities, whose billing chat queue turned out to have nothing to measure against.
How it works
Four steps, and none of them is typing.
- 1
Their customers pick the channel.
Larkspur Utilities
Sample- From the report
Channel: inbound chat on billing queries
The theme their own customers raise most
Scope: two intents, one language
Small enough to run, large enough to read
Length: twelve weeks, two review points
Week four and week eight
The queue their own customers raise most is the one the pilot runs on, taken from the report already on the account.
- 2
Scope small enough to read.
Larkspur Utilities
SampleContainment against the agreed line
measured weekly
Resolution quality, sampled and scored
human review
Escalation handover time
against their own current handling
Shapes, not promises. Each line is the level the pilot has to reach, agreed by both sides in writing.
A defined set of intents, one language, a fixed length and two review points. Everything outside that is written down as outside it.
- 3
No baseline, no start.
Larkspur Utilities
SampleThere is no baseline for the channel you chose.
No baselineWhen the chosen queue is not reported separately on the client side, week twelve would end in an argument. The flag offers baseline weeks at the front, a different channel that is already measured, or a shadow period used as the line.
- 4
Both sides sign what passing means.
Documents
SamplePilot design
Channel, scope, length, review points and owners
Exit criteria sheet
Each line, how it is measured, who reads it
Sent for signature
The exit criteria sheet says what is measured, how, by whom, and what level counts as a pass, with the review dates on both calendars before anything is switched on.
The risk beat
When something turns, it tells you and shows you the moves.
Here the turn is: There is no baseline for the channel you chose. Three moves were offered and a person took one: add two baseline weeks before the pilot starts.
Risk beat · AnyCRM AI
Sample data- 1
Detection
There is no baseline for the channel you chose.
- 2
Explanation
Billing chat is handled by a team that does not report separately, so week twelve would end in an argument about whether anything improved.
No baselineRaised before the pilot is proposed - 3
Options
- Add two baseline weeks before the pilot starts
- Run on email, where they already report separately
- Run in shadow for a fortnight and use that as the line
- 4
Human choice
Add two baseline weeks before the pilot startsA person presses one. Nothing is sent or written until then.
- 5
Resolution
Week fourteen ends in a decision, not a debate.
Both sides signed the same definition of passing before anything was switched on, and the baseline behind it was measured rather than remembered.
Baseline agreedExit criteria signed by both sides
Where the answers come from
The record you already have.
A pilot is designed from what the account already tells you: which queue their customers complain about, what your delivery model covers, and what can actually be measured on their side.
The voice-of-customer read
Which queue their customers actually complain about
Your service lines
What your delivery model can stand behind for twelve weeks
The deal record
What has already been proposed and agreed
The output
The pilot design
Every line points back at one of these.
The client's reporting
Whether the chosen channel can be measured at all
Meeting transcripts
Constraints and dates the client has already named
Past pilots
What scope has actually been readable before
What you receive
Documents, not a dashboard.
A pilot design and an exit criteria sheet both sides can sign, with the review dates already on two calendars.
See the full CS Teams page- Tasks
The pilot design
Channel, scope, length, review points and owners
- XLSX
Exit criteria sheet
Each measure, how it is read, and who reads it
Baseline plan
Baseline plan
How the line is established before the pilot starts
- Tasks
Calendar review points
On both sides, before anything is switched on
Why BPOs need this
What we heard, in their words.
The problem
Week twelve is an argument about whether anything improved.
A delivery lead
The problem
The client's definition of success appeared in month three.
A sales director
The problem
We piloted on a queue nobody reported on separately.
An operations manager
Questions
The honest answers.
Who decides the exit criteria?
Both sides. The sheet is drafted from the evidence and signed before the pilot starts, which is the only point at which the definition is uncontested.
What if there is no baseline?
The design stops and says so, and offers three ways through: measure first, move to a channel that is already measured, or run in shadow and use that as the line.
Does the pilot design include pricing?
Pricing sits on Deal economics, where the price per outcome is built. The pilot design covers what is run, how long for, and what counts as passing.
Can the scope change mid-pilot?
It can, as a recorded change with a reason. What it cannot do is change quietly, because every weekly report reads against the criteria as signed.
Where the pilot meets the commercial shape: scenarios, the pricing inputs behind an outcome price, and the margin guardrails a pilot has to respect to be worth running.
Deal economicsGive your team the unfair advantage.
Full deal intelligence from day one. No credit card required.