Accuracy and Precision

tl;dr: The words are used interchangeably, but they don’t mean the same thing.

One of my close friends is a senior executive at a blue-chip biotech firm. He’s been in that business his entire career and has been the M&A lead on a number of important deals.

He was the first to raise my awareness that “accuracy” and “precision” are not the same thing, even though I had pretty much used them that way.

The difference, however, is really important and I’m still getting my head around it.

Per Wikipedia:

accuracy is closeness of the measurements to a specific value, while precision is the closeness of the measurements to each other.

Let’s start there.

Accuracy

Given that definition, I think of accuracy as a measure of “original intent.” (I sound like a Constitutionalist now.)

Anyway, here’s how I’m thinking of it.

As a sniper, my intention is to hit the bullseye. As LeBron James, my intention is to get the ball in the basket. As a politician, my intention is to get re-elected.

The overall accuracy of an activity or initiative, then, is a measure of how close I came to the original, intended target.

• Did the quarterback get the ball to the receiver?
• Did the bomb land on the head of the terrorist?
• Did we grow our Annual Recurring Revenue this quarter?

So, the specific value of “caught balls,” “dead terrorists” and “growth in ARR this quarter” can be assessed.

• 75% of the balls thrown by the quarterback were caught
• 59% of the shots by LeBron James went through the hoop
• 67% of terrorists targeted were killed
• ARR has grown in 3 of the last 5 quarters

…and so on.

Now, we can tell how accurate we are.

Precision

Precision, from what my friend tells me, offers the allure of accuracy but is deceptive.

Precision tells us things like:

• 97% of missiles fired by a Predator drone landed within 3 feet of the desired GPS coordinates
• 83% of the quarterback’s passes leave his hands within 3 seconds at speeds above 20 mph
• 98% of shots that LeBron James takes from within 1 foot of the basket go in
• 64% of customers who are offered a “free bonus month” sign up for an annual renewal

These are all interesting data points and they can tell us a lot about elements of the initiative, but not the whole story.

While it may be worthwhile to know that the missiles all and within a small radius, it doesn’t tell us if we were accurate in the use of the missiles.

The quarterback may have precise delivery and control of the ball, but that doesn’t mean it was received.

and so on.

Data Explosion, Dreams of Precision

The risk that I see is in the magnetic pull of precision. And I think that is what my biotech friend was saying to me.

I was catching up with a former colleague from my Sprinklr days and we were talking about the explosion of Marketing Technology tools (MarTech) over the last 10 years.

As he said, “there’s a MarTech SaaS product for everything now.”

But don’t take it from us, take it from Scott Brinker’s annual update of the MarTech 5000 (which is now 8000)

It’s looks like something that Georges Seurat would have painted. (For more Seurat: see the Obvious and Non-Obvious future of Crypto)

But I digress.

The point is that, with 8000 tools at a marketer’s disposal (and any other function), we are awash in data.

We can measure how effectively our display ads work in Germany, France, and India versus Canada and the US. We can fine tune everything. We can get really granular and really precise.

But that doesn’t mean we have increased accuracy.

More data and more information may make us feel better and give us a sense of control.

It’s an illusion.

If we’re going to lead the type of organizations that can realign quickly in the face of “high VUCA,” we have to start from a framework of accuracy. And end there as well.

Precision may help us choose the path, but accuracy is the destination.

Postscript

I asked my friend for his feedback on this post. Here’s what he sent.

I like the breadth of examples.

I always think of Shaq’s pitiful foul shot performance. Probably 100% of his foul shots touched the rim and/or hit that square on the backboard above the rim, but very few of them actually went in. He was so good at getting the ball to precisely the right spot, but it was never accurate.

My goal in a transaction is to reach an accurate assessment of an asset’s value. We build low, base and high NPV scenarios that reflect different possibilities.

We consider all sorts of inputs (revenue growth, sales and marketing costs, manufacturing costs, etc.) all of which contribute to the model’s perceived accuracy.

The inputs to the NPV model, which in most cases are reasonably informed projections of the future generated by our experts, are treated as highly precise numbers, often calculated out to several decimal places.

The net result is an NPV valuation that is seen as highly accurate, because everyone knows it’s built on a foundation of precision. But that foundation, while built on a series of precise assumptions, can be very wrong.

So we lull ourselves into a feeling of comfort (accuracy) based on a foundation of well-intentioned (precise) assumptions.