From objective to outcome: Make your data driven projects more successful in 6 steps

nanochomp monster data chomp logo

We recently wrote a post on “When data driven goes wrong: How focusing on single metrics can have unintended consequences.” We’ve seen data projects go south in a lot of interesting ways; it’s often a combination of great intentions and a misaligned objective, with the wrong data giving you bad insights and just a dash of bad process. That recipe can send a data-driven project sideways in an epic fashion, even with an otherwise solid strategy. And that brand of failure? It looks bad on everyone (follow us for more fashion advice.)

OK, enough of the doom, gloom and cautionary warnings; today we’re discussing how to make your data driven projects more successful.

Why should you care? Because when things go well, you and your team can:

  • save time
  • make informed decisions instead of emotional ones
  • save money you didn’t know you were wasting
  • find hidden correlations by connecting dots faster
  • connect dots that mere mortals (humans) can’t see and didn’t even know were there
  • gain a huge competitive advantage

We think we just heard you say, “I’m sold, give me the blueprint.” Great, we’re happy to have you on the squad.

Here’s the process, nice and easy:

Objective -> Strategy-> Collect -> Analyze-> Outcome -> Iterate*

You’ll see this structure pop up over and over again in the content we develop. We are firm believers in hypothesis-based problem solving as the starting point for just about everything, from working with marketing teams on strategies to developing and deploying AI solutions for clients.

We use this workflow because it’s effective, it stops you from chasing cars, it makes your actions intentional and it forces you to map your effort to success criteria as well as overarching business goals.

1. Objective

“If you don’t know where you’re going, any road will take you there” — Cheshire Cat.

The Cheshire Cat isn’t usually thought of as a sage of business wisdom and strategy, but he absolutely nails this one. Have a clear objective so you know what road to take to get there. Start with a hypothesis and clearly state what you’re trying to achieve or accomplish. If we’re spending time and money on something, we need to know why we’re doing it. Your objective also drives what data you need to collect, where you need to collect it from and what metrics are proving (or disproving) your hypothesis.

How does this map to the higher level business goals?

The wrong objective can have disastrous results on your overall business performance. We’ve seen customer service teams set metrics that inadvertently decrease revenue. We’ve seen marketing teams succeed at targeting customers that are more likely to take a call, but less likely to convert to an actual sale. Having the right goals with the right skin in the game is very important.

Today, businesses are becoming more Agile, and that means your competitors too. The teams that communicate effectively and transparently set goals that all map to the larger business objectives will always outperform the siloed teams high-fiving themselves for successfully executing a project that perfectly fails to move the ball toward the high-level business goals.

What does success look like?

Now, you’ll need to define what a win looks like. Setting minimum success criteria will give you some much-needed guard rails that will tell you if your effort is tracking to your expectation of success. Drawing that line in the sand gives you a baseline to determine what is working and what needs to change, even if the only thing to change is your early expectations.

We often like to take our initial minimum success criteria and say, “What would we need to do to 10x this result? 100x that result?” or “How would we accomplish this in half the time, or with half the resources?” Sometimes you get really creative ideas that help you accomplish more than you initially expected.

2. Strategy

To frame your data-driven strategy we need to turn to the data, but first we consider the following:

What data do you need to collect?

To accomplish our objective and prove if our effort is exceeding our minimum success criteria and working (or not), we need to determine what data we will actually need. A poorly thought-out set of metrics can derail an otherwise excellent plan. It’s thinking through what helpful data you need to collect for analysis.

How it will be collected

What’s your plan for collecting your data? For marketing campaigns you’ll likely have multiple data sources across several platforms, and you’ll need to know how to get your hands on them. There are metrics that may require some cross-department collaboration to be granted that access.

Where it will be stored

Collecting and getting access to the data is one thing, but making sure you know where it will live, how it will be processed and cleaned is another. This task ends up being the real grunt work of data projects and a famous failure point for even seasoned teams. You might start off oldschool with spreadsheets, and that’s perfectly fine, but over time your data initiatives will be more sophisticated and your tools will need to grow as well.

For example, with any data-driven marketing initiative you’ll likely be leaning on Google Analytics, Facebook Business Manager and other tools. Over time you’ll want to consolidate those separate data sources using APIs to get the most out of your effort by analyzing cross-platform data collectively. When that time comes you might be relying less on spreadsheets and more on databases in AWS or Azure.

The important thing is to make sure you are storing the appropriate data in a place you have ready access to so you can rapidly analyze and iterate.

How it will be structured so it maps to your stated objectives

With the data sorted, you can really dig into your strategic planning and how the data will be an integral part of validating your efforts.

3. Collect

Strategy: Check.

At this point we are well into executing our initiative and collecting all that data we planned for in the previous step. This task can be as simple as looking at dashboards or as in-depth as pooling all of the data sources into a data lake that we can perform advanced analytics on. The important thing (especially for your early projects) is to start. Many teams stay on the bench because the process feels intimidating, but don’t be that person. As you progress through your data-driven journey you can keep optimizing and get better with each cycle.

4. Analyze

Depending on the objective or problem you’re trying to solve you’ll likely take different approaches here with different algorithm families to get your outcome. For example:

  • Working on an automation problem? Sounds like reinforcement learning.
  • Doing keyword detection in audio or object detection in video? This smells like a job for a neural network.
  • Processing huge amounts of customer data and linking it back to their persona types? K-means clustering to the rescue.

Few things are more worthless than unanalyzed data. A car without a motor, a staircase to nowhere, diet water, the G in lasagna. You did all the hard work; this is where you get your reward. Plus, you get to use buzzword-rich jargon at this part of the process. Does any discipline have better buzzwords right now than the AI/ML space? Artificial Neural Networks, Support Vector Machines, Random Forests, Recurrent Neural Networks. These are the words LinkedIn flexes are made of.

5. Outcome

All of that hard work is paying off and your cup is now spilling over with those sweet, sweet insights.

So, how did it turn out? Did you validate your hypothesis? Were there any surprises? Did you get a few ideas on what to test next? Did the results reinforce some of your assumptions? Are you enjoying those data visualizations from Seaborn and Plotly?

What adjustments do you need to make to get the most from your insights and run another test?

6. Iterate

This is just the start.

Use the data to help direct you to what tweaks to make, what hypothesis to test next and go back to the top of this list. Each time through you’ll get better, your data will give you better insights, and your process will get a little cleaner. The first few passes through you will likely be doing a lot of manual testing and analysis.

Over time you will be able to automate more and more. If something is measurable and continuous you can eventually let the robots do the lifting.

And if you feel intimidated or overwhelmed, just know that according to NewVantage Partners, 72% of companies have yet to forge a data culture and 52% of companies are not treating data as a business asset. So you’re already ahead of the pack.

And if you need help or advice making your next project data-driven, let us know. If you couldn’t tell, we love this stuff.

co founder of nanochomp. 🚀 Startup builder 🤖 Machine learning/AI nerd ⚙️ Gearhead 📍 RDU/DTW

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store