Customer
The team completed an agreed work task
What gets in the way
A purchase has happened, but value has not
RA LAB / HOW THE LAB WORKS
One problem. One test. A clear decision.
We find where customers fail to get value. The team and AI test one change. Evidence guides the decision: scale, iterate or stop.
- 01Problem
GRI findings + customer evidence
- 02Plan
One accountable role + one change
- 03Test
Human × AI + a comparison
- 04Decision
Value + cost + quality
An interactive guide to the method. The examples below are test plans, not completed experiment results.
01 / CUSTOMER GROWTH LOOP
From first interest to repeatable profit
The customer journey continues after the sale. We connect first value, the next need and business economics.
B2B / 02
Promise → a first verified outcome
Team
Deliver first value and verify it with the customer
Accountable role
Solution Expert + Business Analyst → Customer Success
AI + human review
Suggest onboarding steps and flag common errors; a specialist verifies the outcome with the customer.
What counts as evidence
The agreed task is completed and confirmed by the customer
Primary metricTime to first value + activation rate
When we move on
The customer confirmed a useful outcome
Data and the next-stage criterion
- What we record in CRM
- Start, first-value event, date, confirmation, costs
- Link to economics
- Onboarding and support costs before value
AI helps prepare and check a scenario. The team remains responsible for the offer, quality and decision.
02 / HUMAN × AI × BUSINESS
The customer journey continues after the sale
These are functions, not mandatory job titles. In a small team, one person may hold several roles. Every handoff has an accountable receiver.
Qualification → discovery → solution → first value → adoption → retention → renewal / repeat → expansion
- 01Presale
Qualification
Confirms the need and passes selection criteria to Sales.
- 02Sales / Growth Lead
Solution selection
Agrees the promised outcome and terms, then hands the job to the expert.
- 03Solution Expert
Working solution
Checks fit and readiness of the use case.
- 04Business Analyst
Evidence and process
Records the baseline, value event, data and test criteria.
- 05Customer Success
Value and adoption
Helps deliver first value, resolve friction and return when needed.
- 06Account / Revenue Owner
Repeat business and economics
Owns renewal, repeat purchases and useful expansion, with contribution checks.
We pass on the customer job, promised outcome, constraints, verified event and next step. We agree a response time before starting. The receiver confirms understanding; we check outcome quality and that nothing was lost in the handoff.
03 / TEST → EVIDENCE
We test a hypothesis before claiming an effect.
Below are three templates. Baselines have not yet been measured. We set the minimum useful change and sample size before starting. Small-sample findings remain provisional.
01 / FIRST VALUE
Will guided onboarding deliver value sooner?
A short route to an agreed outcome will reduce time to first value without reducing quality.
View details
- Baseline
- Measure current activation rate and median time to value; include customers who never reach value.
- How we compare
- Compare randomly assigned eligible groups receiving guidance or the current process. If randomisation is not feasible, use comparable cohorts and state the limits.
- Observation window
- Agree the task cycle before starting; then account for support and refunds. Twenty-one days may be a test setup cycle, not evidence of retention.
- Cost and quality
- Team time, AI and rework; activation, errors, complaints and drop-off must stay within agreed limits.
- Decision rule
- Scale only if time improves meaningfully, activation remains sound and costs are acceptable.
02 / REPEAT PURCHASE
Will customers return when a new need arises?
Consent-based contact at the natural time of a new need will increase the share of repeat paid purchases.
View details
- Baseline
- Measure repeat purchases in the original cohort over equal periods, excluding refunded purchases and tracking renewals separately.
- How we compare
- Compare eligible customers with and without additional contact. Account for purchase terms and seasonality consistently.
- Observation window
- Observe a full new-need cycle. For long cycles, plan 90–180 days or longer; confirm the period from actual data.
- Cost and quality
- Contact and incentive costs, net contribution, complaints, opt-outs and absence of pressure to buy.
- Decision rule
- Continue if incremental repeat business covers costs and preserves relationship quality.
03 / USEFUL EXPANSION
Will a bundle solve the customer job better?
A relevant addition will increase contribution per eligible customer without unnecessary purchases or a worse experience.
View details
- Baseline
- Measure contribution, order mix, discounts and refunds before testing. Revenue is not profit.
- How we compare
- Compare eligible customers offered a bundle or the usual offer. Measure all included customers, not only buyers.
- Observation window
- An agreed purchase cycle plus the return period; one successful order is not enough.
- Cost and quality
- Variable costs, discounts, rewards, service, refunds and displacement of other purchases.
- Decision rule
- Scale with positive incremental contribution and verified value; otherwise revise the bundle or stop.
01
Scale
A useful change is supported by a comparison; costs and quality meet the agreed limits.
02
Iterate
There is a signal, but evidence is limited or the scenario needs work. Change one condition and retest.
03
Stop
Value is unsupported, economics do not work, or quality and consent limits have been breached.
04 / MEASUREMENT
Define first. Measure second.
Before a test, we agree the event, denominator, period, data source and costs. A sign-up, meeting or proposal alone does not establish first value.
Customer events and states
- New Customer
- First valid paid purchase.
- Activated
- Agreed first value reached.
- Active
- Useful action within the defined need cycle.
- Renewed
- An expiring contract was renewed.
- Repeat Buyer
- A new paid purchase; not a duplicate renewal.
- Reactivated
- Return after an agreed inactivity period.
- Expanded
- Incremental contribution from expansion verified.
- Churned
- Customer loss established by an agreed rule.
States and events can overlap: a customer may remain active while expanding a purchase. Employee bonus categories do not define customer cohorts.
Metrics in plain language
- Activation Rate
- Share of the original eligible cohort reaching agreed first value within a defined window.
- Time to First Value
- Time from the agreed start to the value event. Read the median alongside activation rate and incomplete journeys.
- Cohort Retention
- Original-cohort customers with a useful action in the next agreed cycle / the original eligible cohort.
- Renewal Rate
- Genuinely renewed contracts / eligible contracts expiring in the same period.
- Repeat Purchase Rate
- Customers with a new valid paid purchase / the original buyer cohort over the same window. Track renewals separately.
- Expansion Margin
- Incremental contribution versus control or a comparable baseline after variable costs, discounts, incentives and displacement.
- LTV / CAC
- Customer contribution over an explicit horizon / full acquisition cost. Projected LTV is labelled as an assumption, not a fact.
- CAC Payback
- Time for cumulative contribution to cover acquisition costs, accounting for actual cash receipt timing.
This public guide contains no customer data. A real test uses only necessary data with appropriate access; personal data and CRM contents are not sent to website analytics.
RA LAB / NEXT STEP
Propose a problem to test
AI helps prepare and check a scenario. The team remains responsible for the offer, quality and decision.
What the model is based on
This is a proposed RA LAB model connecting the customer journey, accountability and evidence of value. Implementation, training, renewal and expansion experience informs the process; it does not establish an AI effect. Implemented projects can be reviewed separately.