---
title: "Goodhart’s Law does not disprove OKR—Goodhart’s Law calls for OKR."
id: "16681"
type: "page"
slug: "when-ai-learns-faster-than-your-okr-cycle-2"
published_at: "2026-07-22T17:00:54+00:00"
modified_at: "2026-11-01T18:15:45+00:00"
url: "https://okrexperts.com/okr-method/when-ai-learns-faster-than-your-okr-cycle-2/"
markdown_url: "https://okrexperts.com/okr-method/when-ai-learns-faster-than-your-okr-cycle-2.md"
excerpt: "[vc_row css_animation=”” row_type=”row” use_row_as_full_screen_section=”no” type=”grid” angled_section=”no” text_align=”left” background_image_as_pattern=”without_pattern” z_index=””][vc_column][vc_empty_space height=”64px”][/vc_column][/vc_row][vc_row css_animation=”” row_type=”row” use_row_as_full_screen_section=”no” type=”grid” angled_section=”no” text_align=”left” background_image_as_pattern=”without_pattern” z_index=””][vc_column width=”1/2″][vc_column_text css=””] Goodhart’s Law does not disprove OKR—Goodhart’s Law calls for OKR. [/vc_column_text][vc_column_text]By Erno Marius Obogeanu-Hempel[/vc_column_text][vc_column_text]The most well-known saying goes: “When a..."
---

[#](#)
# Goodhart’s Law does not disprove OKR—Goodhart’s Law calls for OKR.

By Erno Marius Obogeanu-Hempel

The most well-known saying goes: “When a measure becomes a target, it ceases to be a good measure.” In German: “As soon as a metric becomes a goal in itself, it loses its significance.”

The author of this law is Charles A. E. Goodhart (born 1936), a British economist, former advisor to the Bank of England, and professor at the London School of Economics.

The idea behind this is simple: Key performance indicators (KPIs) are originally intended to measure success. However, when they become an end in themselves, people’s behavior often changes—and with it, the KPI’s relevance.

A key performance indicator (KPI) is, first and foremost, an indicator of something valuable. However, as soon as people are rewarded or punished for optimizing that metric, they begin to focus on optimizing the metric itself rather than the actual goal. As a result, the metric loses its meaning. Our article ” [OKR vs. KPI](https://okrexperts.com/okr-vs-kpi-the-difference-and-how-they-work-together/)
” explains how KPIs and OKRs differ.

**Examples**

- **Sales:** Goal = as many leads as possible. → Low-quality leads are generated.
- **Customer Support:** Goal = keep calls as short as possible. → Customers are rushed through the process, and problems remain unresolved.

## Why This Is So Relevant to OKR

Goodhart’s Law explains why Objectives and Key Results serve different purposes:

- **Objective:** The desired target state and the intended benefit (outcome). Question: What target state do we want to achieve by the end of the cycle—and why?
- **Key Results:** Several measurable indicators of progress and goal achievement (success drivers). Question: How can we tell, both during the cycle and at the end, that we have reached the target state?

[https://www.ki-plattform-vergleich.de/?pid=1&prid=okrn0716](https://www.ki-plattform-vergleich.de/?pid=1&prid=okrn0716)

## Objectives define the outcome—key results make it measurable

The Objective describes the desired end state and the intended benefit. The key results are measurable indicators of whether we are moving closer to the target state or have achieved it by the end of the cycle. They are proxies for the desired outcome—not the outcome itself. That’s why you should regularly ask yourself:

**Are we really optimizing the desired outcome—or just the metrics?**

## Multiple key results instead of a single metric

Goodhart’s Law doesn’t call for less measurement—but for smarter measurement. That’s exactly what OKR is.

Traditional goal-setting often involves stating, “Increase the number of leads.” Then, that specific metric is maximized.

In the case of OKR, this could be, for example:

**Objective**

We have consistently attracted more qualified new customers to ensure our profitable growth over the long term.

**Key Results**

- Increased marketing-qualified leads from 320 to 500.
- Increased the conversion rate of Marketing Qualified Leads (MQL) to Sales Qualified Leads (SQL) from 25% to at least 35%.
- Improved the average lead score from 68 to over 80 points.
- Customer Acquisition Cost (CAC) reduced from €620 to less than €450.
- The Net Promoter Score (NPS) for the initial consultation increased from +42 to over +60.

This creates a tension:

- Having a lot of leads alone isn’t enough.
- High-quality leads alone aren’t enough.
- Low costs alone aren’t enough.
- Satisfied prospective customers alone are not enough.

It is only the interplay of several key results that defines the desired success. They represent different facets of the outcome and prevent a single metric from becoming the sole target for optimization.

**Customer Support**

**Objective:** We have enabled our customers to experience the best support in the industry.

A single key result such as “Reduced average call duration from 6 to less than 4 minutes” would obviously be problematic.

Because the question immediately arises: How does this contribute to the objective?

A well-balanced set of key results might look like this instead:

- First-contact resolution increased from 82% to over 90%.
- Customer satisfaction (CSAT) increased from 4.3 to over 4.8 out of 5 points.
- Average response time reduced from 90 to less than 30 seconds.
- The re-contact rate was reduced from 18% to less than 10%.

It is only the interplay of these key results that defines excellent support. No single metric could achieve this on its own. The objective serves as a plausibility filter: Each key result must clearly contribute to the desired target state.

[https://okrexperts.com/okr-training-seminars-further-education-and-training/okr-seminar-offers/1-day-okr-seminar-2/](https://okrexperts.com/okr-training-seminars-further-education-and-training/okr-seminar-offers/1-day-okr-seminar-2/)

## Outcome Instead of Output

Perhaps the most important difference, however, lies in its consistent focus on outcomes.

Many KPIs measure activities. Examples:

- Number of calls
- Number of Meetings
- Number of Features
- Number of Campaigns

Instead, OKR asks: What impact do we want to achieve?

For example, don’t: ❌ Develop 20 features

but ✅:

- User activation increased from 35% to 50%.
- Churn rate reduced from 8% to 5%.
- Average processing time reduced by 40%.

So you’re not optimizing the work itself, but rather its impact.

Success isn’t defined by a single metric—but rather by the interplay of several key results.

That is precisely why a good OKR does not consist of as many metrics as possible, but rather of a few Key Results that complement each other effectively.

[https://okrexperts.com/okr-training-seminars-further-education-and-training/okr-seminar-offers/okr-coach-master-champion-training/](https://okrexperts.com/okr-training-seminars-further-education-and-training/okr-seminar-offers/okr-coach-master-champion-training/)

## Conclusion

In OKR as well, Key Results remain metrics. That is why Goodhart’s Law generally applies within an OKR system as well.

The key difference, however, is that OKR systematically reduces these negative effects:

- Objectives define the desired end state and benefits.
- Multiple key results reflect different dimensions of progress and success.
- Outcome is more important than output.
- Conflicting goals become apparent.
- Key Results are reviewed in each cycle and adjusted as needed.

Goodhart’s Law therefore does not refute OKR. On the contrary, it provides a compelling rationale for why modern goal-setting systems like OKR require multiple outcome-oriented key results rather than a single metric.

[https://www.ki-strategie-berater.io?pid=9&prid=okrn0716](https://www.ki-strategie-berater.io?pid=9&prid=okrn0716)

- [Deutsch](https://okrexperten.de/okr-methode/okr-goodharts-law/)
- [English](https://okrexperts.com/okr-method/when-ai-learns-faster-than-your-okr-cycle-2/)
