Log inExperience Cascand

Learn

Test whether Cascand is helping, against a real control group.

You can test whether Cascand, or a change you make, is working. Choose the outcome that counts, such as replies, meetings or opportunities, add a guardrail such as opt-outs or bounces, set up a control group, and let Cascand compare the groups. It says how much the evidence can show, and refuses any verdict the numbers cannot support.

Experience Cascand →Pricing

How it works

How an experiment runs.

  1. Step 1

    Define the experiment and the outcome.

    Name it, state the hypothesis, and choose what counts as success: a reply, a meeting booked or an opportunity created. Add a guardrail if you want one, such as opt-outs or bounces. Three kinds of test are built in: hold Cascand's recommendation back from the control group, give the treatment group's drafts a different style direction, or change the follow-up timing for the treatment group.

  2. Step 2

    Assign groups at random.

    Assign by hand or let Cascand randomize. There is a minimum size per group, so a result is never read from too little.

  3. Step 3

    A guardrail watches.

    If the treated group breaches the guardrail, the experiment stops there and its result can never be confirmed. Nothing is confirmed automatically.

  4. Step 4

    Read what it found.

    Each result says plainly what it can show, and anyone assigned to a group but never actually contacted is counted as unreached rather than as a result either way. Cascand confirms nothing from a breached guardrail, a lopsided split, too little exposure or too small a sample. A valid result becomes a proposal that an administrator on your team adopts or declines once, on the record.

In your hands

What you control.

  • Who is in each group. Cascand never starts, reassigns or acts on an experiment; you do.
  • What counts as success, and what counts as harm.
  • What you conclude, and whether a result is adopted.
What it needs

Enough people to randomize across, and time for outcomes to arrive.

Questions

Questions people ask.

What does an experiment measure?
The outcome you choose, such as replies, meetings or opportunities created, compared between the groups, with a plain statement of how much the evidence can show.
Can I test my messages against each other?
Yes. Give the treatment group's drafts a different style direction, or a different follow-up timing, and compare the outcome. Or test Cascand's recommendations against a control group.
Does Cascand claim it caused a result?
No. It shows how strong the evidence is and does not claim a cause it cannot show.
Will an experiment hurt my pipeline?
Choose a guardrail, such as opt-outs or bounces. If the treated group breaches it, the experiment stops there and its result can never be confirmed.
What happens when I adopt a result?
The decision is recorded with who made it and when. Adopting is a deliberate human step, never something an experiment does to itself.

Keep exploring

Where to go next.

Next

See it with your own work.

Tell Cascand about your business and see what it would build. A person on our team reviews every application.

Experience Cascand →Pricing