Skip to content
Chumbaq
Join the waitlist

The winner is the message that sold

Most tools crown the variant with the most taps. Chumbaq holds back a control group and scores every variant on purchases and revenue, so you know the message actually made money.

Join the waitlistRead the docs

Ridge boots restock

Purchases within 72 hours of the send

  • Variant A48 hours only: 20% off trail gear36purchases
  • Variant BYour size is back: Ridge Mid boots68purchases
  • Control groupNo push sent24purchases

Variant B wins: 68 purchases and $4,182.00, against a control group that got no push.

  • Control group

    A slice of the audience gets nothing, so you see the real lift.

  • Scored on conversions

    Purchases and revenue pick the winner, not opens.

  • Several variants

    Test copy, images, timing or channel side by side.

  • Revenue per variant

    See what each variant earned against the control.

Variants and a control group in one campaign

Write two or more variants, choose the share of the audience that gets nothing, and send. A curious headline may win the taps while a clear offer wins the purchases; Chumbaq shows both and scores on the one that pays.

New push campaign

Draft
AudienceCarts left this week2,184
Variant AVariant BControl, 10%

Title

Your size is back

Message

The Ridge 2 is back in {{size}}. Two pairs left.

Buttons

Shop your size → /product/ridge-2, Not now

Send

At each user's best hour, quiet hours 22:00–08:00

Send a testSchedule

Sample campaign.

What would have happened anyway

Some users would have bought without a push. The control group tells you how many, so the lift you see is the lift the message caused.

Revenue by campaign

Last 30 days
  • Spring sale reminder Email$11,033.00

    187 sales

  • Welcome series In-app$7,614.00

    141 sales

  • Ridge boots restock Push$6,270.00

    104 sales

  • Back in stock WhatsApp$5,664.00

    96 sales

  • Order shipped SMS$1,902.00

    38 sales

Sample figures from Trailhead, a sample app.

Send the events that matter

Track purchases and other goals from your app with their value. Chumbaq ties them back to the message and the variant that came first.

main.dart, identify

await Chumbaq.instance.init(config);

// Who this is, once they log in.
await Chumbaq.instance.identify(
  IdentifyParams.builder().externalUserId('u_42').build(),
);

A/B testing, answered

How big should the control group be?
Big enough to measure: the dashboard lets you set the split. Small audiences need more time before a winner is clear.
Which channels can I test?
Push campaigns support variants and a control group today. The control-group method applies to journeys and other channels as they add variants.

Keep exploring

Bring your users back

Free up to 2,500 monthly active users. No card, no sales call.

Join the waitlistRead the docs