Going deeper to 'optimize' a pull #

**Warning… long post**

Okay, Direct Mail folks, I want to pick your brain.

I've been in the role for a year now, and I've been enjoying the ability to stats the heck out of everything, use some logic to build queries etc... that really tickles the right part of my brain.

But, I've been asked to 'go deeper', and I think I'm up against a wall. See if you can point out the obvious to me as I'm stymied.

--Normally--

We pick our folks who'll get our DM appeal based on criteria like "Have they given to this appeal in the past few years? or have they given to this kind of fund in the past few years?" Then I also add in that maybe that gift should be more than $X... something along those lines.

We'd get lists of 9-15k recipients. Spray and pray.

I'd like to reduce that and be a little more narrowly focused. So I've been playing with the # of years (i.e. more recent donors), and/or how much is their average gift. So I can get that to reduce down to like a comfy 5k recipients. (We'll still complement our traditional media with digital counterparts so those not 'mailed' may still get email or social media hits)

--Boss wants deeper--

He wants me to go even further and sure, reduce the recipients, but by a lesser degree and I should use some math to back up where that optimal point is.

If this were some 1st year calculus or even Grade 10 quadratics, I'm sure I could calculate.

What he's thinking is that I can do one query pull of a 'high calibre' potential donor list.

Then do the same where the criteria is relaxed a little.

Then again...

...and keep going until I can calculate the ROI might be approaching to negligible.

I understand the logic, but I don't know how to 'implement' that.

If I say 'within 3 years and more than $100'... how can I reasonably say that those 3000 donors are likely to get a 6% return with an average gift of $150... [so I could 'expect $150 x .06 x 3000

Then those 2000 donors who are 3-5 years or more than $50 are likely to say 3% return with an average give of $100 [so expect 2000 x 0.03 x $100]...

The 1000 who are 5-7 years (all double checking that they're not also in 1st or 2nd group) are going to get 2% return with an average of $50 [so expectation of 1000 x .02 x $50]

and so forth...

those are some 'assumptions' that I feel we're guessing on... or how could I calculate it? I could pull the query of those that match and have the summary of their average gift and try that... but the response rate is just ...

I don't know if this is possible.