How do I hire a Predictive Analytics expert?

January 31, 2014 7:39 AM

I'd like to hire a full time person who can use data mining and predictive analytics to steer our fundraising strategy. Only problem is I'm clueless.

I'm in non-profit fundraising, specifically I manage an annual fund. For those not familiar with the terminology, annual funds are typically an organization's broad based fundraising efforts. Think email, mail and calling programs; lots of small gifts.

I'd like to hire a full time person who can use data mining and predictive analytics to steer our fundraising strategy. My organization is in central Long Island and I could probably afford a salary range of $55k-$60k.

A few questions.

1. Is a full time job justified to design strategy and do analytics for a constituent base of 200,000?

2. Am I correct that my salary range is going to be able to only secure an entry level person? If so, what kind of college experience am I looking for in an applicant?

3. I am convinced I need better data analysis and predictive analytics but I've barely got a surface knowledge of the area. I have read "The Signal and the Noise" by Nate Silver, "Predictive Analytics" by Eric Siegel and "Fundraising Analytics" by Josh Birkholz but that is the full extent of my knowledge. I have zero statistics background in my education. When I'm interviewing, what am I looking for to know I got someone with the right skill set?

THANKS!

posted by meta x zen to Human Relations (8 answers total)

This could be contract work. If it is just to review data and steer your strategy, depending on the volume of data and types of questions you want answered (who to ask for funds, which fundraisers to run, which ones are most popular but don't make money etc etc), I would guess a month of work and then have them write you a report.

If it will be ongoing work, then you could hire them contract to start and morph it into full time if you need.
posted by St. Peepsburg at 7:49 AM on January 31, 2014 [2 favorites]

I would also suggest contract, maybe a couple of phases that coincide with your annual schedule (analyze, suggest, structure things, then monitor, then post-campaign refinement, etc.)
posted by aramaic at 7:54 AM on January 31, 2014 [1 favorite]

Also for the interview you could ask them to come prepared with examples of their work ("I analyzed such-and-such dataset to answer blah blah blah questions and this is the report I wrote"; bonus points if they can say "so the company took XYZ actions as a result of my data and saved/made $$$amount") would help get a sense of how attentive to detail they are and how well they communicate.

I don't think you could be expected to quiz them on statistics (that's why you're hiring them, duh), so at that point check their credentials (masters in statistics or comparable) and their references.
posted by St. Peepsburg at 8:34 AM on January 31, 2014 [1 favorite]

As you have guessed, that budget will likely only net you someone entry-level, and probably not very great entry-level candidates to boot. Unfortunately, it sounds like you need someone who is definitely not entry-level to help you figure out what it is you need them to do in the first place. I agree with the comment above that suggests looking for a freelancer to contract, with the possibility of bringing them on full time if together you discover there is a need for on-going substantial work.

Peepsburg has some good advice on asking for work samples to judge their competency and communication skills. In my experience, this is sometimes hard since so many companies/clients throw up legal barriers to sharing work produced on their dime. Another solution I've seen work: If your org is comfortable releasing some of their data (with whatever NDA your legal dept would like), you could give a strong candidate or two a copy of some of your data and give them a few days or a week to take a look at it, and provide feedback about the data characteristics, and suggestions on how it should be treated, analyzed, and transformed into actionable knowledge going forward. This can give you a good sense of a candidate's initiative, technical creativity, and most important of all, ability to communicate their findings and recommendations. This should only be considered after you have interviewed and identified a strong candidate, and you should pay them a fair hourly price for it under a simple short-term contract that places a cap on the number of hours they are permitted to expend. I've typically seen max hours in the range of 4-24, and giving the candidate 3-7 days to complete and deliver.
posted by stoffer at 10:33 AM on January 31, 2014 [2 favorites]

I'm going to say that this is a pretty big job for an entry-level person at that salary. For comparison, a friend of mine makes about $100k to be the "junior level" staff person on a 2-person team that does this kind of thing. Also it would be hard for you to judge if this entry-level person was doing a good job, nor would you know how to best manage them.

But not to fear! Being a nonprofit in this space gives you an edge in the hiring market that most companies don't have. You can reach out to groups like DataKind or Pro Bono Economics to see if they'll do your project for free. And more importantly, you can reach out to their members and advertise a contracting position. If you can get an experienced contractor, also see if they can work with an entry-level person with an undergrad stats background to do the grunt work. I'd expect to pay the contractor at least $100/hr, and the entry-level stats person about $20/hr.

Another option is to reach out to your local business schools and see if they have a class that would be willing to take this on as a project. Or, if you're willing for this person to work remotely, and part-time, I bet a number of grad students would take it on for the salary that your average RA-ship would pay, about $30/hour. If that route appeals to you, then I'd reach out to the statistics, economics and industrial engineering departments at any schools you already have connections to.
posted by tinymegalo at 10:34 AM on January 31, 2014 [2 favorites]

The most important question is: how is your staff at cold calls? Be honest. I've seen so many cases where the fundraisers are provided names from a model and do absolutely nothing with them.

I think some of that money might be better spent by having a vendor do a test annual giving model for you and see what the fundraisers do with it (if anything).
posted by mogget at 10:51 AM on January 31, 2014 [1 favorite]

Response by poster: ahhh mogget - you know the business ;)

Thanks for the feedback guys. A very clear conscientious coming through here.
posted by meta x zen at 12:00 PM on January 31, 2014

Best answer: I do this type of work on a similar salary at a relatively inexperienced level (i.e. no stats background). For what you are looking for, I would recommend contracting out and getting one or two models built. Bentz Whaley Flessner (where Josh Birkholz works) is fabulous, but other groups can do this type of work too. Try it with a contractor first, and if you see an ongoing need, then think about hiring a full-time staffer. You may not need a dedicated analytics staffer right now.
posted by JannaK at 6:38 PM on January 31, 2014 [2 favorites]