Visual Storyteller
Related Posts
Conversation Starter
More Posts
Visual Storyteller
Anonymous User
Conversation Starter
Additional Posts in Data & Analytics Consultants
New to Fishbowl?
Download the Fishbowl app to
unlock all discussions on Fishbowl.
unlock all discussions on Fishbowl.
I’ve done this a few times now. The largest hurdle is hiring good data scientists imo. You are competing with FANG/SV. Just be prepared to lose your cross offers and some of your best people.
Establishing good practices up front is helpful. Coding style, version control, etc. Take time to build out packages/libraries/methodologies for analyses you will use consistently. Even things like standard color palettes are useful. My team had UX/UI and we had standardized navigation/controls for our apps.
Knowledge sharing across the team is helpful. I had my team to present on an analytical technique or project they found interesting each month (e.g. time series clustering, mixed integer optimization, density/geospatial based clustering).
Try to protect the team from nonsense admin so they can code.
There are a few things:
1. Ability to execute projects independently. Everyone says this but in my time at fang I noticed that people there need significantly less management time to get to the same result as in consulting.
2. Focus on insights and results versus some cool methodology. We mostly worked in sql and excel. Most problems are so big that you don’t need an advanced method to identify and resolve the issue.
3. Counter to the above point people could go deep technically when needed to which wasn’t often. Usually it was figuring out how to get some massive query to run smoothly. Sometimes it was more traditional ML or forecasting work.
4. Ability to communicate and problem solve. Much of my work was with the c-suite and or product owners. Can you create the insight, deliver it clearly and influence actions?
5. Are you curious and do you love to learn? If I leave you alone with a problem you have never seen will you formulate a solution? This is huge imo. And a big differentiator.
Probably some stuff I’m forgetting but these are the main points.
Every consulting firm will say the same but talent is a bell curve and the mean of FANG’s bell curve is shifted to the right of consultings.
Establish core principles, methodologies and ways of working. Look for people that always keep an eye on what's going on in the data analytics landscape and never just go for the fancy models, but keep in mind that the goal is always to help people take decisions. Differentiate noise from signal etc. Good verbal and written communication are more valuable than technical skills or knowledge of a particular platform or technology. Drawing a good graph, a good diagram and an effective presentation is more valuable than a portfolio of dashboards. And everyone should have or be prepared to acquire a strong knowledge of Data Governance and Data Management principles, because without these two all things go south
Most of your hires will need to know how to code. That's just data. If you live in Excel land, you can only do parts of the job.
Hire people who'll write code by having them debug something. It tests coding ability, creativity, ability to navigate something new (huge!), and can easily be time-boxed to an interview round.
The alternatives kinda suck:
- Leetcode tests esoteric grind and you'll cut out a large portion of viable hires who understandably haven't looked at "manually" rotating a 2d array since undergrad (or ever, if biz side or stats hires). Google can afford to miss these candidates, but your boutique can't.
- Single-chance SQL questions are equally grindy, and cap your pool of hires to who memorizes best. That's not what you're hiring for. Again, Google can afford this inefficiency, but your boutique can't.
- Take-homes yield inconsistent results and annoy many candidates to the degree of dropping out of your interviews themselves. Yikes.
- Assuming coding ability instead of checking can lead to very subpar hires that'll torpedo your execution ability. Most people can cobble together something somehow working given time, but the amount of time and resulting reliability will differ by an order of magnitude. Experience and title don't always guarantee that learning's happened: lots of people coast in their jobs or never get mentorship. So, you *have* to check.