KPIs are not enough

We don’t have to convince many people that they need to set objectives, state KPIs and collect data; books like “Digital Transformation”, “The Lean Startup” and “Measure What Matters” did that heavy lifting for us. And if you’re not familiar with those books, you’ve at least heard the buzzwords they helped mint echo through your Zoom calls more often than “Sorry, I was on mute” and “Can you go back a slide?”

Let’s kick off this blog post by getting on the same page with a few terms. We love TLAs like KPI and OKR, but if no one knows what we’re talking about we all lose:

1. TLA — A three-letter acronym for “three-letter acronym”
2. KPI — Key performance indicator. KPIs are essential to effectively managing your business and knowing what’s happening. They are the diagnostic beacons that can tell us if things are going as expected.
3. OKR — Objectives and key results. OKRs take KPIs a bit further and force you to state your objective, followed by what key results would be indicators of success.

While teams are now trained to be data-driven and collect data, somewhere along the way we started substituting data collection for actual decision making. Instead of making decisions based on the data, teams are flying the “mission accomplished” banner for simply collecting it and taking a look once a week or so. We would argue that just stating your KPIs, and even collecting that data just isn’t enough, and in many cases not even necessary.

“Wait, did you just say collecting data isn’t necessary?”

What Imean is, collecting the wrong data can be worse than collecting no data at all. Creating tons of digital exhaust to export as a CSV and contort into a pivot table is a waste of effort if it doesn’t get you closer to your objective or isn’t action-oriented.

So, getting the right data is just as important as having a good objective. But before we go down the path of collecting data, cleaning it, making a beautiful dashboard and smugly smiling when we see it move up and to the right, we need to ask a fundamental question:

“What action will I take depending on what this data tells me?”

If the answer is “Nothing” or “I’m not sure,” it might be a bad metric, or you need to step back and think through your action. It’s not as daunting as it sounds. With a good objective in place the KPIs and action plans almost write themselves.

Data Driven in 5 Steps

1) Start with an Objective

Any good initiative must be rooted in a testable plan with clear objectives

What do you want to accomplish? Before we mindlessly jot down revenue, margin, reach and customer acquisition cost as the KPIs we’ll track, let’s think about what we’re trying to accomplish, what we want to test and what we want to optimize.

It helps to state your core assumptions that helped you land on that path. We find that teams think they’re aligned, but when you ask a few “why’s” you realize that teammates had different assumptions and assumed everyone else had them as well. Get these on the table for better results and more successful data-driven projects.

2) Define KPIs

Great, we have an objective in place, we thoughtfully built our assumptions and we know what we want to accomplish. Now we can start defining what KPIs and metrics matter. What data will give us the best chance of proving our success criteria, validating our effort and guiding our team’s activity?

It’s worth asking yourself “Is there a better metric?” Many times a ratio of metrics is a much better KPI to track than singular metrics. “Revenue” might be a good metric, “# of customers” might be a good metric, too, but “revenue per customer” is likely better to track if we’re trying to optimize customer lifetime value and acquisition costs.

You might also consider a hero metric. You might see this called the “One Metric That Matters” (OMTM). This is your keystone metric that drives most of your team’s activity. It’s worth noting that this will change over time depending on the problem you are trying to solve, so don’t hold on to a metric that is no longer useful for the objective at hand. As a cautionary tale, we have even seen teams inadvertently set KPIs that directly conflict with achieving their OMTM. To prevent that trainwreck, we lean into radical transparency to make sure everyone knows not only what the team is doing, but why they are doing it.

3) What Actions Will You Take

It may sound obvious, but repeatedly we see teams tracking metrics with no plan on what action they will take. Stating your action plan before you get started can help you make actual progress. This step is the up,up,down,down,left,right,left,right,B,A,select,start of successful data projects.

State what action you will take based on what the data tells you.

Since we all love formulas, this should get you started:

• If the KPI goes above X, we will do Y
• If the KPI goes below X, we will do W
• If the KPI is flat, we will do Z

4) Collect and Analyze

It’s extremely hard to be data driven without, you know, data. So, we’ll need to determine what data to collect to help us validate our assumption. Oh, and a non-minor consideration: where we’re going to get it from. With your objective in hand and the KPIs you want to track, you’re well on your way to knowing what you need.

Business data is often siloed, incoherent, underutilized and unoptimized, so getting your hands on it can be a little tricky. Data sources like your company’s sales data, inventory levels and cost structure are probably scattered throughout your organization, so it may take a little finesse to secure. Other data like web traffic, digital ad campaign results, customer cohort data will live in Google Analytics (GA), Facebook Ad Manager (FB), and Linkedin (LI) with various stakeholders holding the keys to the gates.

While we love nothing more than clean datasets imported into Numpy and processed with Matplotlib, Scikitlearn and Tensorflow then beautifully displayed with Plotly for data viz <insert drool emoji>, it’s OK to start with a spreadsheet and a dream. Treat this stage like a pilot/minimum viable product and trickle in the AI and automation as you go and prove value.

We know we say “when you’re ready to trickle in AI and automation” like it’s no big thing, and that’s because it doesn’t have to be. When you’re ready to build your approachable AI strategy, we know people who can help. If something is measurable, repeatable and rule-based you should let a robot do it. Life is too short to use pivot tables all day.

5) Iterate

After you analyze your data you can see what’s working, try new things and see how they are performing vs your baseline. You should constantly test, collect new data and make tweaks based on what you learn. Making incremental improvements with each pass through the process will prevent you from suffering the slow death of “Well, it worked before so it will keep working” fallacy.

TLDR (too long, didn’t read)

KPIs without a strategy or a plan don’t help; if you don’t have an action plan for the data you’re collecting, what’s the point? And not only could it result in a lot of wasted effort, it may actually steer you in the wrong direction.

If you’re listening to the data you’ll have your finger on the pulse of what’s working, and if you have an action plan for your KPIs (and you actually execute that plan) you’ll see results that’ll help inform your next move.

1) State your objective, 2) define your KPIs, 3) build your data-driven action plan, 4) collect and analyze the data, 5) iterate

Buzzwords — check

5-step process — check

Contra cheat code — check

Let me know how you approach initiatives and how your teams are navigating this digital transformation. We would love to hear about it or even have you as a podcast guest.

In the meantime, if you love AI buzzwords, check out this short Business Leaders Guide to Data and ML at the bottom of the page.

www.nanochomp.com