Topic: Research/Metrics

Calculating Roi On Purchased Data

Posted by HELEN.ESSEX on 125 Points
I am trying to calculate ROI on a data count I've been provided by a third party, but the numbers I'm getting seem wildly inaccurate. Here are my calculations:

Quote: $6000 for 304k contacts (rounded down to 200k to allow for quality)

I've estimated a 1.5% conversion to MQLs per month which comes to 3000?! Is that conversion rate really off?

CPL: 6000/3000 = $2

I've estimated a 15% lead to prospect rate:
15% of 3000 = 450

Our typical prospect to converted rate is 11.28%
11.28% of 450 = 50

Cost per conversion = $6000/(50x12) = $10

Does this look right? I've never had to purchase anything of this sort before so need to make sure it's a sensible approach.

To continue reading this question and the solution, sign up ... it's free!


  • Posted by Jay Hamilton-Roth on Member
    What is the quality of the purchased list? Have these leads opted in, have proven conversion from other clientele, and are a match for your target criteria? Many lists that you purchase are likely to have very low conversions for you: they don't know you, they didn't opt in for any marketing you may provide, unknown email validity, and unknown customer profile.
  • Posted by Peter (henna gaijin) on Member
    Hard to provide comments on the numbers, as we don't know the quality of the list (as Jay pointed out), nor do we know your product, what the offer is, what the sales cycle for the product is, etc.

    But in general, purchased lists are of much lower quality than those which you may bring together yourself by hand.

    If you are doing this exercise as a part of the decision of whether to buy the list or not, because there is so much fudge in all these estimates, you could easily make it look like a go or a no go. There is no real way to come up with accurate numbers short of doing it. Practically, you just need to have a gut feel for whether this may be worth trying or not, and follow that.
  • Posted by Jay Hamilton-Roth on Member
    One thing you could ask/pay for is a random 1K sample of the data, and then see for yourself what the quality is.

Post a Comment