I’m a sales/marketing analyst, working in B2B distribution. At the end of last year I found myself a late casualty of the GFC and since then it’s been an unsuccessful slog getting back into work. As a 50y0 single Dad of 3 school age children, the outlook is getting dire.
The latest prospect is a data analyst role with a on-line job site. Which is just a touch ironic!
The technical elements of the position are fine, I’ve got ample experience in the functional elements.
The weakness is industry experience with on-line business.
One of the element they are looking for is “ability to work with imperfect data”.
Now every analyst works with a varying degree of confidence in their underlying data.
Could any of the SD help me with what, in context, they might be looking for?
I don’t have a clue? I was a tester and inputted my collected data and some passed the tests and some failed. The data had to be real. Is it for failure analysis?
“Is it for failure analysis?”
No, sales & marketing and sounds pretty routine apart from that.
“Our DB is particularly crappy”
May be the case, but they are the established market leader.
When it comes to the crunch the whole company is a data base on a website.
"Our last data analyst, when asked for information on ‘Acmee Co’ would come back to us with ‘can’t find that in the DB’. We’d kind of like someone with the common sense to also search for ‘Acme’ or ‘Ac%’ "
The company I work for is also a market leader whose entire business is basically a db with a website. And based on (at least) one former employee who would not necessarily have had the common sense to work around the above example, I wouldn’t be totally surprised to see a line about imperfect data coming out of our department.
Caveat - I’m an IT chick, I don’t really know what sales and marketing ppl do with the db, apart from ask my department to provide reports on XYZ. But in the past a reasonable amount of my job has been “think of creative ways to get around the fact that we didn’t keep a record of Important Fact X before Y-date”
I don’t think we’re currently hiring in Sydney, otherwise I’d wonder if you were applying to us!
(Note that this is not my field, so I may be completely wrong.)
I would guess that you answered your own question. They don’t want someone who will make decisions by blindly assuming all the data are perfect. If you are submitting an application in writing I would guess that they want you to talk about how you assess your level of confidence in the data and how this then affects the way you perform your analysis and present your conclusions.
In other words, they are checking that you have real world analysis experience.
I was once interviewed for a Quality Manager’s job (ISO 9001 based) and was asked how I would deal with resistance to the implementation of a Quality System. This wasn’t because the organisation had a particular problem with their staff resisting change, it’s just that there is always resistance to implementing Quality Systems. They just wanted to know how I handled it.
I have a Six Sigma background. One of my coworkers has a Six Sigma background. We are used differently.
I’ll work with imperfect data. I’ll take it - anecdotal data, data we’ve mined using our “expert judgment,” data where the sample isn’t great - in business you get a lot of that. I throw a zillion disclaimers around it and say what I think what we’ve got indicates. Moreover, I’m pretty good at it - I have a good feel for when an anecdote is a one off thing - and when its the start of a pattern. I’m not bad of eyeballing populations and saying “this is like this, but probably around 20% off” - and if we measure it, I’ll be close. (I like the book Blink, and honestly, often a lot more work goes into it than eyeballing it - if you understand the population, estimating the variance is a reasonable thing to do - depending on what sort of confidence you need .)
She works with perfect data. She’s great at it. But the data has got to be good or she throws up her hands, delays the project for months in an effort to measure something that can’t really be measured - or where it would cost too much to fix the measurement system. She is just NOT COMFORTABLE saying “this is the assumption I’m making.”
Upshot, if the project requires the “ability to work with imperfect data” - I’ll get called in. We often work as a team, me making the strategic, gut, risk analysis decisions - her working through the details (which I can do, but not nearly as well as she does).
More reading “How to Measure Anything” by Douglas Hubbard
It seems the “imperfect data” is rather less conceptual than due to the various business units, though all using SAP, having defined their data tables inconsistently. Not uncommon when you have a business trying to consolidate rapidly expanding autonomous units. Constructing a translation table would apply a bandaid to the system until the definitions/procedures are standardised.