SOLVED

Do random samples have to be in 1 step or receive dupes?

Go to solution
Highlighted
Anonymous
Not applicable

Do random samples have to be in 1 step or receive dupes?

I set up a campaign and we are throttling our emails. We used to convert our smart list to many static lists and then put wait steps between each send.

At Summit, we talked to someone from Marketo who told us a more efficient way would be to create 1 static list and then do a random sample. So we have:

1. If random sample is 2%, send email.

2. Wait step

3. If random sample is 2%, send email...and it goes on and on.

We have our campaign set that each lead can only run through once.

We have people receiving multiple emails - is it because they are being randomly sampled over and over?

Thanks.

1 ACCEPTED SOLUTION

Accepted Solutions
Highlighted

Re: Do random samples have to be in 1 step or receive dupes?

Hey Eumee,

The reason you're getting people receiving duplicates is exactly what you mentioned. What you're doing is saying:

  1. Take 100% of the leads
  2. Randomly select 2% and send them the email
  3. Wait 1 day
  4. Take 100% of the leads
  5. Randomly select 2% and send them the email
  6. etc.

While the probability of the same person receiving both emails is only 0.04%, when you factor in that you're probably doing it 50 times the odds go up significantly.

If you're looking for an elegant solution that still requires some manual effort you could do something like this (but you would need to manually adjust the % of the random sample occasionally)

What you could do to make the process more elegant is to create a looping batch campaign that's set up with:

  • 1 smart list
  • 1 smart campaign

Your smart list would be defined as <your target audience> + Not was sent email is <your email>

Your smart campaign would be -

Smart List: The smart list you've defined above

Flow: If | Random Sample | 2% | Send email | <your email>

Else: Do Nothing

Note: When you're doing something like this the elegance factor comes at a cost since 2% of 100% is more than 2% of 98% is more than 2% of 96%.

If you're looking for an automated solution that gets you the results you're looking for what you might need to do is create a system with static lists and assign all of them at step 1. The remaining steps would then send the emails to each static list.

So you'd have a flow like:

If Random Sample 2% add to list group 1

If Random Sample 2% add to list group 2

If Random Sample 2% add to list group 3

...

Else add to list group n

If group 1 send email <your email>

else do nothing

wait 1 days

if group 2 send email <your email>

else do nothing

...

It sounds like a pain in the arse to set up, but to all intents and purposes once you've set it up in a program which you can then clone (never clone a live program with dumb lists it'll clone membership too) then it wouldn't be too much work to replicate the process.

Hope this helps

Guy

View solution in original post

8 REPLIES 8
Highlighted

Re: Do random samples have to be in 1 step or receive dupes?

Hey Eumee,

The reason you're getting people receiving duplicates is exactly what you mentioned. What you're doing is saying:

  1. Take 100% of the leads
  2. Randomly select 2% and send them the email
  3. Wait 1 day
  4. Take 100% of the leads
  5. Randomly select 2% and send them the email
  6. etc.

While the probability of the same person receiving both emails is only 0.04%, when you factor in that you're probably doing it 50 times the odds go up significantly.

If you're looking for an elegant solution that still requires some manual effort you could do something like this (but you would need to manually adjust the % of the random sample occasionally)

What you could do to make the process more elegant is to create a looping batch campaign that's set up with:

  • 1 smart list
  • 1 smart campaign

Your smart list would be defined as <your target audience> + Not was sent email is <your email>

Your smart campaign would be -

Smart List: The smart list you've defined above

Flow: If | Random Sample | 2% | Send email | <your email>

Else: Do Nothing

Note: When you're doing something like this the elegance factor comes at a cost since 2% of 100% is more than 2% of 98% is more than 2% of 96%.

If you're looking for an automated solution that gets you the results you're looking for what you might need to do is create a system with static lists and assign all of them at step 1. The remaining steps would then send the emails to each static list.

So you'd have a flow like:

If Random Sample 2% add to list group 1

If Random Sample 2% add to list group 2

If Random Sample 2% add to list group 3

...

Else add to list group n

If group 1 send email <your email>

else do nothing

wait 1 days

if group 2 send email <your email>

else do nothing

...

It sounds like a pain in the arse to set up, but to all intents and purposes once you've set it up in a program which you can then clone (never clone a live program with dumb lists it'll clone membership too) then it wouldn't be too much work to replicate the process.

Hope this helps

Guy

View solution in original post

Highlighted
Anonymous
Not applicable

Re: Do random samples have to be in 1 step or receive dupes?

Thanks Guy. We normally did the way of doing a bunch of static lists. When at Summit, we asked around for a more efficient solution and someone who works at Marketo told us the random sample way. Guess that was wrong!

Highlighted

Re: Do random samples have to be in 1 step or receive dupes?

Hey Eumee,

Theres a great quote by H.L. Mencken that I swear by in these sorts of situations

"For every complex problem there is a solution which is clear, simple, and wrong"

my personal approach is that sometimes we set lofty goals and want to achieve complex processes and that's totally awesome... I love pushing the boundaries, but when you do the challenge is to develop an efficiency that will allow you to do something complex but make it easy to replicate and enable others in your organization to replicate.

That said - if you always ask yourself:

why are we doing it? And is it worth it?

you'll know whether you're onto a winning concept.

Good luck and keep pushing the limits!

Highlighted

Re: Do random samples have to be in 1 step or receive dupes?

The post that Guy set up makes sense, but just to throw one more option out there (in case this is easier), I'd recommend setting it up this way:

Step 1: If random sample is 2%, send email

Step 2: wait

Step 3: Option 1: If delivered email is (email from first send), do nothing

               Option 2: If random sample is 2%, send email

This filters out the first group from the 2nd send, since Marketo always looks for the first applicable match in a multi option flow step, and saves you from creating a bunch of different emails. If you continue with this process you can cumulatively add all the previous emails so step 3 would be If delivered email is Email 1, 2 or 3, do nothing.

Highlighted
Anonymous
Not applicable

Re: Do random samples have to be in 1 step or receive dupes?

Hi Anna,

I am using your advice, but have people who need to qualify for this send at a later date. If I use was delivered logic, then they will not qualify for this email again. Any idea on logic that might work for this scenario that would allow people to not get dupes AND also qualify for a second time through. I am brain-dead and stuck.

Thanks,

nikki

Highlighted
Level 6

Re: Do random samples have to be in 1 step or receive dupes?

Hi Nikki

Could you clarify for me? When you say "people to not get dupes" but also "qualify for a second time through", isn't that the same thing? If you could qualify again, wouldn't that be a duplicate send?

Highlighted
Anonymous
Not applicable

Re: Do random samples have to be in 1 step or receive dupes?

Phillip Wild​ - you may find this thread helpful

Highlighted
Level 6

Re: Do random samples have to be in 1 step or receive dupes?

I like it. Thanks Christina Reynolds