Skip to content
BloopBloopBloop Docs

Best practices: measuring and optimizing your program

A program you do not measure is hard to improve. Start from the results, find the weakest step, change one thing, and test that change before you keep it. This page shows how to do that, and which Bloop reports to use at each step.

Program overview in referral analytics, where to start reading your results

Optimizing is a loop: measure, find the gap, change one thing, test, keep the winner, and repeat. Small, steady changes add up. A change made without a test can cost you reward budget with no way to tell what worked.

Read the funnel backwards

Section titled “Read the funnel backwards”

Start with the results, then work back to the behavior that produced them:

  1. Outcomes: who joined, who made a purchase, and how much revenue followed.
  2. Behavior: who saw the offer, who shared it, and through which channels.

Your analytics dashboard shows both. Reading it in this order makes the weak step easy to spot.

Use the table to match what you see with a likely cause and a change to try:

What you see Check this metric Likely cause What to change
Many visitors saw the offer, few shared it Share rate in the Funnel report The reward does not feel worth sharing Reward type or value
Many shares, few new customers New customers per sharer and Claim-to-purchase rate in the Funnel report The friend offer is weak, or the friend discount does not work at checkout Friend discount, minimum spend, or audience
Strong first orders, few repeat orders Repeat orders per customer in the Revenue and customer quality report Customers are not coming back Reward type such as store credit, or follow-up after purchase

The Funnel report with Share rate (2) and New customers per sharer (5) marked

Total revenue is a good headline, but two metrics show what to change: New customers per sharer (new customers divided by sharers) shows whether shares turn into customers, and Repeat orders per customer (repeat orders from referred customers divided by new customers) shows whether friends keep buying.

The Revenue and customer quality report with Repeat orders per customer (6) marked

Change one lever at a time

Section titled “Change one lever at a time”

When the data points to one step, change only the variable you expect to matter most. In most programs that is the reward value or the friend discount, because they affect conversion more than wording or layout.

Changing one thing at a time makes the cause of any improvement clear.

Test before you keep a change

Section titled “Test before you keep a change”

A reward that is too small or too generous can look the same from the outside, so test before you commit. Bloop’s A/B testing splits new participants between two versions:

  • Control A: your current rewards.
  • Variation B: a copy where you change one thing.

Bloop shows results from real conversions side by side. When you end the test, you choose Control A or Variation B.

The End test dialog with Control A and Variation B marked

Follow these rules:

  • Test the change with the biggest effect first, such as the reward value, then the friend discount.
  • Change one thing per test.
  • Wait for real volume. Early results swing back and forth. If the test shows no difference, keep the cheaper reward.
  • If traffic is low, grow referral traffic first. A test needs enough orders on both sides to mean anything.

Start this week

Section titled “Start this week”
  1. Open the Funnel report, find the weakest rate, and match it to a row in the table above.
  2. Change one thing that affects that step: the reward value, the friend discount, or the sharing channels.
  3. Set up an A/B test for that change, and leave it running until it has enough orders.
  4. Review the results and keep the winner before you start the next test.
Section titled “Related”