Throttle mailing via wait step - last 25% did not send

[Jamie_Barclay]

I'm testing an approach to throttle out a large mailing.  I had a test group set up this way (deploying 25% over a 4 hour period) - but the last group did not send.

Can anyone tell me what I've missed in my set up?

[Michael_Florin · Accepted solution]

Well, that whole construction doesn't really add up, does it?

Let's assume your whole list is 100 leads.

In the first step you send to 25.

In the second you ignore 25, then send to 25%*75, which would be 18.

In the third step you ignore 43, then send to 25%*57, which would be 14

In the fourth step you ignore 57, then send to 25%*43, which would be 10.

At the end of the flow you have sent to 67 people. 33 didn't receive the email.

If I'm not mistaken we had that topic before and some folks proposed to split the list beforehand and create four equal lists. Another option would be to do the math a little differently:

Step 1: Send to Random Sample 25

Step 2: Send to Random Sample 33 (1/3 of 75)

Step 3: Send to Random Sample 50 (1/2 of 50)

Step 4: Send to all

[Josh_Hill13]

You have better control if you setup the lists ahead of time, otherwise what Michael describes happens. Random Sample should be done within ONE flow step for proper use.

[Jamie_Barclay]

Hi Josh,
Can you explain more?

[Josh_Hill13]

It's just what he said

The cohort of leads from your Smart List run through the ENTIRE flow.

Step 1: Random Sample IS 50% = 50/100

Step 2: do nothing -->> so now the other 50 still continue.

Step 1: Random Sample IS 25% of 100=25 (but this group overlaps with Step 1 sometimes.

etc.

[Josh_Hill13]

So within a single flow you say:

Send Email:

if Random Sample=25%=25people

if Random Sample = 25

if Random Sample=25

Default=25

so then everyone is accounted for and no one gets two emails or no emails.

[Jamie_Barclay]

Thanks Josh - I think I'm going with Michael's recommnedation and I've created individual lists

Step 1: Send to Random Sample 25

Step 2: Send to Random Sample 33 (1/3 of 75)

Step 3: Send to Random Sample 50 (1/2 of 50)

Step 4: Send to all

[SanfordWhiteman]

Those aren't individual lists, Jamie.

Individual lists means in a single Flow step you split the qualified leads into N static lists, then use ​Member Of List to determine who to send to.

[Michael_Florin]

To make it little more explicit:

Step I: Splitting the list using random sample in one "Add to List" flow step:

Step II: Sending the email using the "Member of List" condition:

It's not exactly elegant, but no workaround is. 🙂 But it's a solid setup that created 4 equal lists with no overlap.

[Grégoire_Miche2]

In other terms, the random will cover 100% of the cohort only of they are split in the same unique flow step using choices. The way the smart campaign is implemented here restart the selection at each send email flow step and does not guarantee a partition (in the mathematical sense) of the whole database.

-Greg