The targets you can’t see

Human behaviors are infinitely subtle and complex in ways that metrics could never adequately capture.

⚠️ Disclaimer: This post is my own opinion, and not the opinions of Tavus!

When Steve Jobs released the Magic Mouse, it came with what users considered to be a critical design flaw: In order to charge it, the mouse had to be turned upside down, rendering it temporarily unusable.

“Our customers won’t be able to use this mouse for hours at a time while charging, this is a terrible idea,” everyone warned.

The decision for Apple computers to not feature touch screens went the same way; it was something Apple could’ve easily done, yet Jobs was opposed to it.

Again, he faced a bunch of criticism: Windows laptops all had touchscreens.

What Steve Jobs had was a vision that was completely removed from what we would consider to be “metrics.” His focus was on subtleties. Feelings, even.

And because of this vision, the Magic Mouse was truly “magic” at the time of its release, in the sense that no one uses it like a regular, wired mouse. (They couldn’t.)

Because of their lack of touch screens, Apple computers rarely have big fingerprints and smudges. They’re clean, clear, and pristine, just like Apple retail stores, Apple’s website, Apple’s product launches, Apple’s product design, and so on and so on.

The products, and the way they’re used, all “feel” the way Steve Jobs intended. And thus, much of Apple’s success to date has been from making decisions based on feelings, assumptions, and strong opinions about brand and UX.

Because, at the end of the day, “does the Magic Mouse feel magic?” is not a trackable metric.

What are metrics?

As if it needs additional explaining.

Metrics are quantifiable or countable measurements of characteristics, like user count, trial signups, and app downloads. They all boil down to a number with some directionality and an outcome. Something like:

• Number go up = good
• Number go down = bad

In software, finance, marketing, sales, and virtually every other industry, business decisions are made based on metrics, in one way or another.

The problem with metrics

The problem with metrics is fundamentally tied to their purpose: Metrics are meant to be proxies for user behaviors and other important signals. They’re analogues.

☝️ People forget this.

Metrics often represent the thing you’re trying to measure, but they’re not the actual thing itself.

Some of your business’s most important objectives are probably things that metrics can’t track:

• Is my app easy to use?
  
  
• Do users find our content to be authoritative and trustworthy?
  
  
• Is our brand perceived as luxurious?
  
  
• Is our website easy to navigate?
  
  
• Is my product actually useful?
  
  
• Did Beyonce really have the greatest music video of all time?

And so, in trying to answer these types of questions—which are critically tied to business outcomes (for the most part)—metrics serve as a substitute for actually knowing the answers.

This sounds acceptable, assuming the metrics do a good job of aligning with the sentiments you’re trying to influence. But if they don’t, your business can find itself in trouble, ranging from wasted time on initiatives to deeply-rooted and systemic failures to generate impact. More on this soon.

Don’t believe me? Here’s one of the richest people of all time saying the same thing:

“The metric isn’t the real underlying thing. Maybe the metric is customer contacts per units sold…You have to remind yourself, I don’t care about the metric. I care about customer happiness.”

KPIs getting in the way

I recently interviewed a staff engineer who just left a very well-known B2B SaaS company that had just fired most of its marketing team. He and the company will remain anonymous.

The reason I asked him for an interview is that, as a marketer myself, I noticed the company’s web metrics all looked very good. I actually wanted he inside scoop, and had no idea the marketing team had been effectively scrapped.

My friend, the staff engineer, told me two things that stood out:

1. The marketing team’s growth had no impact on business metrics
2. The company’s goals, KPIs, and metrics, were fundamentally flawed

In his mind, the marketing team wasn’t necessarily excused from having no impact; their metrics were likely tied to leads and revenue. However, for him, generating his own impact had its own challenges due to the way the company set and adhered to goals.

The company’s product was a data visualization platform and one of its big eng-wide KPIs was views created by [a specific persona] that were viewed by 2+ teammates.

Sounds great if you’re an idealist. The goal is clearly meant to be a proxy for how useful [specific persona] finds the product. But if you’re an engineering manager that wants to build the best product possible, metrics like this might just get in the way.

How to get promoted, given the company goals:

Step 1: Create a button that shares a view

Step 2: Make the button bigger

Step 3: Get rid of other buttons

Step 4: Get promoted

Or what if you’re the marketing team, and you’re trying to appeal to new, viable markets? Does this just mean you get let go for not hitting company goals?

In the case of these marketers, yes, that’s exactly what it means.

What went wrong? (The telephone game)

It’s like the telephone game. The sentiment is that the company wants its specific persona to love it. The problem is that that sentiment distills into individual goals, and loses some meaning each step along the way.

The end result is that, if employees want to get promoted, they’re going to optimize for this distilled goal, rather than what’s best for the company.

This subtracts almost all of the freedom and empowerment that leads to innovation, scrappiness, and new ideas. If you’re a mature company and your goals are set in stone, this might be fine, but it can just as easily kill innovation (and culture).

The downfalls of overindexing on metrics

The “telephone game” of metrics influencing KPIs and goals creates a disparity between internal metric lift and actual impact.

Most commonly, it’s possible to be doing everything perfectly within your role, hitting your KPIs and lifting the metrics you’re responsible for, and still see no tangible business impact.

If this isn’t managed correctly, it completely messes up incentives for individual employees.

Metric data versus actual impact

The following real world scenarios are examples of how two drastically different assumptions can be formed based on various metric interactions.

Scenario 1: The super reader

“This user was clicking around on the site and spent a lot of time reading this specific blog post.”

• Metric impact: Very high time on page
• Assumption 1: This user really loves this content and is reading every single word
• Assumption 2: The user is playing League of Legends on his other monitor

Scenario 2: The new website

“Ever since we redid our website, we’ve been seeing an overall increase in scroll depth with 15% of users reaching the bottom of the homepage.”

• Metric impact: Scroll depth
• Assumption 1: People are enjoying the new website
• Assumption 2: Users can’t find what they’re looking for

Scenario 3: Refer a friend

“We see lots of users sign up for accounts and then invite a teammate within the first week.”

• Metric impact: Accounts created, users, invites
• Assumption 1: They love the product so much that they’re telling their coworkers
• Assumption 2: Users are getting so confused that they’re inviting others to help them

Scenario 4: Power users

“We have a handful of power users within our app who click way more things and are exposed to far more surfaces than a typical user.”

• Metric impact: Clicks, time in app, surfaces touched
• Assumption 1: They’re a power user
• Assumption 2: These users can’t find the feature they’re looking for

Scenario 5: Our best customer

“This customer has purchased five products from our eCommerce site this year. They must love our products.”

• Metric impact: Sales, revenue
• Assumption 1: They’re a super fan
• Assumption 2: They’ve returned every product

Reclaiming the signal

Metrics were never meant to be the truth—they were meant to point us toward it. But somewhere along the way, we started worshipping the pointer instead of what it pointed to.

The truth is, metrics are lossy compressions of human behavior—simplified, flattened, and distorted versions of something infinitely richer. They’re like trying to capture a symphony in a single decibel reading. You can optimize the loudness, sure. But what about the harmony? The feeling? The reason people showed up to listen in the first place?

When a company starts chasing numbers instead of meaning, the entire system begins to decay:

• Designers stop designing for people and start designing for graphs.
• Marketers stop telling stories and start chasing conversions.
• Engineers start gaming KPIs instead of solving problems.
• Leaders start mistaking activity for impact.

It’s a slow erosion, a loss of information and meaning that happens one dashboard at a time.

The companies that transcend this trap don’t ignore metrics; they contextualize them. They remember that the most important signals are still qualitative: how something feels, how humans respond, what intuition whispers before data shouts.

And while, yes, the best companies all use metrics and data in amazing ways, they understand their place and the limited power they possess.

Metrics can tell you if something is working, but only humans can tell you why.

Tavus’s big bet

Tavus’s big bet is that the next great interface won’t be a screen or chat box. It’ll be a person—one that listens, responds, understands, and takes action.

Similar to the accessibility shift that the GUI provided, machines that perceive and communicate in accordance with the subtleties of humanity will become the next “interface unlock.”

And the metrics support this prediction too.

Ready to converse?

Get started with a free Tavus account and begin exploring the endless possibilities of CVI.

GET STARTED