Home / Method
How Stradem evaluates an opportunity
The method starts with a customer segment, sizes the opportunity and values it. It keeps what we expect apart from what a test shows, and notes what each action needs before it can start. Each section states its assumptions.
Segments and opportunities
Finding opportunities in customer behavior
A segment is a group of customers or accounts that share something, such as single-product accounts that use the product every week. An opportunity is a segment with an action, an expected gain and a cost. Segment Twin builds one from the other in four steps. It starts from your segments or finds them in your data.
- Define the segmentAgreed aggregate counts show how many accounts sit in the segment, and the usage, order or renewal measure that defines it. Where your team has propensity scores, the results for each score band enter as evidence.
- Find the barrierA segment that trails its peers or its plan points to a barrier: friction in the product, a missing offer or a gap in coverage. Patterns suggest barriers. Confirm them with customer context and the people responsible before choosing an action.
- Propose the action and show where the gain comes fromEach action carries an expected gain. The source is an earlier test, a pilot, a benchmark or an assumption.
- Value itThe expected gain and the cost give a value and the margin earned for each dollar of cost. How sure we are and how ready the team is sit beside it.
Different customer problems need different actions
| Segment | Possible barrier | Action to test | Where the gain comes from |
|---|---|---|---|
| New accounts that stall before setup is complete | Friction in setup | Onboarding fix | A pilot on 400 accounts |
| Single-product accounts that use the product weekly | No second-product offer | Second-product offer | One earlier offer to similar accounts |
| Large accounts with no seller | No coverage | Added sales coverage | An assumption, with no earlier test |
Segments from the worked example. How sure we are of a gain reflects the quality of the evidence, how relevant and recent it is, and how much uncertainty remains. Each rating states its source.
Measurement
Keep what we expect apart from what a test shows
A forecast says what we expect before launch. A reported result says what happened afterward. A tested gain says what the action added, compared with a comparison group. Growth Twin keeps the three apart and labels each one.
| Measure | Question | Source | What it supports |
|---|---|---|---|
| Forecast | What do we expect? | The model and the expected gain | A decision to run a test |
| Reported | What happened? | Reporting for the period, or a before and after view | A view of scale and trend. Other changes in the period sit inside the same figure. |
| Tested gain | What did the action add? | A comparison group that stayed out of the action, with its design | A decision to scale, refine, stop or hold |
Reading the result of a second-product offer
The offer in the worked example forecasts a gain of +2.5 points. After the test window, the offer group converted at 11.1% and the comparison group, which stayed out of the offer, converted at 9.0%.
The reported result is 11.1%. The tested gain is the 2.1 point difference between the two groups, with 3,000 accounts in each. Its 95% range runs from +0.6 to +3.6 points. The range uses the standard method for comparing two conversion rates, rounded to one decimal place. That tested gain and range are read against break-even and the target for action. The test measures conversion over eight weeks, and the margin from those conversions arrives over the following 12 months.
Goals and prior years
A goal states an intention, and a prior year sets a baseline. A comparison group estimates what would have happened without the action. A customer who is likely to buy may buy without an offer, so a propensity score shows who is likely to buy and the tested gain shows what the offer added. The client work describes past operating results.
Ready to start
What can hold an action back
Some actions need something that does not exist yet: capacity, an owner, data, a platform feature or a plan for who covers the accounts. The comparison notes it beside each option, so a high-value action that is not ready waits behind the best one that is.
| What is missing | Example | What closes it | Who closes it |
|---|---|---|---|
| Capacity | The product team has no capacity this quarter | Reserve sprints in the next planning cycle | Head of product |
| Delivery owner | The work has a sponsor and no one assigned to deliver it | Name one owner for delivery and one for the measure | The sponsor |
| Data | The usage measure exists only in a monthly report | Produce the measure weekly | Data lead |
| Platform | The platform cannot keep a random group out of the offer | Turn on the feature, or assess whether another comparison design is feasible | Platform owner |
| Coverage plan | No rule assigns sellers to the accounts | Define the plan and route existing sellers to a first group | Head of sales operations |
Can we start now? is the third question
Every option has to earn the margin per dollar of cost that the team needs, and the options that do are compared on value. Whether it can start now decides which one goes first. Closing a gap has its own cost and owner. When a gap closes, the team looks at the option again, because the cost, the timing or the value can change.
- An option that returns less than the team needs goes back for rework, however ready it is
- An option that reaches the required margin to cost but is not ready waits behind the best one that is
- When a gap closes, the option returns to the comparison with updated numbers
Assumptions
What sits behind the numbers
Every figure in an example or a calculation rests on stated assumptions. A decision log keeps them beside the result.
- Margin per outcome is what one extra outcome adds over the period, after product and service costs. Program cost is counted separately.
- The period is the first 12 months after launch unless stated.
- Where the outcome is an early step, such as finishing setup, the link to paid or kept accounts is an assumption. Where it is feasible, we measure the later outcome directly.
- Each team sets the return it needs. The example uses 3.5 times.
- Test results include an uncertainty interval supported by the design and data. Forecasts show their assumptions and scenarios, and when a model supports a statistical interval, we state its method. The rule for the decision uses the result, the safeguards and how ready the team is.
Questions about the method
Where do the segments come from?
Your team defines the segments it already manages. Segment Twin proposes more from agreed aggregate data by comparing outcomes, size and trend. Where your team has propensity scores, Stradem uses the results for each score band to help prioritize opportunities.
What happens when the gain is an assumption?
The opportunity enters the comparison with a Low rating for how sure we are, and the test is built to confirm or reject the assumption before a rollout.
Why write down the forecast before launch?
A forecast written before the result keeps the target and the rule for the decision fixed. The results test the idea the team chose.
What if random assignment is unavailable?
We assess whether the available data supports a credible comparison. We use an appropriate design where one is feasible and state its assumptions next to the result. Where the data supports only the outcome, we report the outcome and say what limits it. Every result is labeled as a forecast, a reported result or a tested gain.
Is a reported result ever enough?
A reported result shows scale and trend, and it informs the next idea to test. A decision to scale an investment rests on the tested gain and its range.
Who decides whether an action is ready to start?
Your named owners decide. Stradem lists what is missing from the information your team provides, and each owner confirms the item and the fix.
Does a gap change the value of an action?
A gap can change the cost, the timing and the expected value of an action. The team looks at the opportunity again when those change.
Can closing a gap be part of the plan?
Yes. The cost of closing a gap enters the comparison, so the team sees the value of the action, including the cost of making it ready.