This is the fifth essay in Reckoning with Reality. The previous essay asked what kind of action can produce evidence. This one examines the speed at which evidence arrives—and how speed quietly changes our goals.
A friend told me that content and retail have one enormous advantage: fast feedback.
Change the opening today and the view count responds within hours. Adjust the price or the page and clicks or sales may change at once. Compared with spending months building a product in private, only to receive a faint response at launch, this rhythm does make it easier to preserve morale.
I agree. But I later realized that fast feedback and fast results are not the same thing.
I have published more than four hundred posts on Xiaohongshu and over a hundred essays on Zhihu. That record is enough to prove I can produce consistently. It does not automatically prove that my content can generate an attributable transaction that remains profitable after final settlement. The body of work is a real achievement; it answers “Have I written over time?” rather than “Have I closed the commercial loop?” Put the same number against the wrong question and it begins to lie.
We Chase the Number That Rewards Us First
Suppose the full chain looks like this:
Seen → Clicked → Paid → Fulfilled → Refund window closed → Settled → Repeat purchase
The data on the left appears within minutes. The data on the right may take weeks or months.
Human attention is not distributed evenly. The number that appears first, changes every day, and can be attributed immediately to our own action naturally moves to the center of the decision. Rising views excite us; the refund rate enters the picture much later.
The reward system begins to tilt. A team spends more time improving clicks than narrowing the gap between promise and delivery. An individual keeps producing content that travels well, even if it attracts people who never truly needed the product.
No one formally decides to abandon the long-term goal. It is simply replaced, little by little, by a proxy with faster feedback.
Once a Metric Becomes a Target, It Begins to Distort
A plain version of Goodhart’s law says: when a measure becomes a target, it ceases to be a good measure.
Click-through rate begins as a way to observe whether a message attracts attention. Once bonuses depend only on the rate, sensational headlines become rational. Call duration begins as a measure of customer-service efficiency. Once every call must end within a few minutes, difficult problems are closed carelessly.
The metric is not wrong. It captures only one slice of the outcome. Incentives then encourage people to optimize that slice until it separates from the whole.
The same happens in life. Pages read can help us begin reading, then train us to turn pages quickly. A streak can establish a habit, then tempt us to lower the quality of practice merely to keep the streak alive. The number of tasks closed makes work visible, then rewards us for placing easy chores ahead of important problems.
Anything easy to count can overpower something important but difficult to count.
Delayed Costs Can Make a Bad Business Look Excellent
Some models produce beautiful early numbers because their costs arrive later than their revenue.
An order is placed today; the return arrives next week. A customer’s advance payment enters the account immediately; the delivery burden accumulates later. Cheap acquisition looks like rapid growth; the cost of support, reputation, and dependence on the channel appears only after scale.
The time lag creates an accounting illusion. Current income has been recognized, while future costs have not yet entered the same picture.
Worse, early success encourages expansion. By the time the slow variables deteriorate, the organization has already expanded around the fast ones. Inventory, headcount, and promises have become difficult to reverse.
This cannot be solved merely by “looking at more data.” The key is to return data from different time scales to the same causal chain. A cohort of orders from the first week of July should be compared only after it has passed through the same fulfillment, refund, and settlement windows as another cohort. Today’s newly paid orders cannot be compared with last month’s revenue after all returns have cleared. More data does not mean consistent measurement.
Fast Feedback Is Still Precious
None of this means that slowness is more profound.
Feedback that arrives too slowly is dangerous too. In a long silence, people cannot correct themselves; small errors travel a long way, and morale is difficult to sustain. If a product is kept from users for six months, one false assumption may already have grown into an entire system.
Fast feedback is best suited to local questions: Was this message noticed? Did anyone click this button? Did this proposal make someone willing to continue the conversation? It lowers the cost of iteration and helps us find obvious mistakes early.
The mistake is using a local answer to judge the whole direction.
Views can improve the next piece of content; they cannot prove a business. A first sale proves that someone was willing to try; it cannot prove repeatable demand. Contribution profit from a mature cohort can describe current unit economics, but it cannot by itself tell us how reputation and repeat purchase will change.
Every metric has a boundary of authority. Beyond that boundary, it stops being a tool and becomes a story.
Give Feedback Two Clocks
I now prefer to keep two clocks.
The fast clock is for learning. It records signals available often enough to keep daily action from going blind. The slow clock is for judgment. It tracks final settlement, repeat purchase, cash tied up, the user’s long-term outcome, and whether I still want to continue.
The clocks need not update at the same pace, and neither can substitute for the other.
Watch only the slow clock and action becomes sluggish. Watch only the fast clock and direction drifts. A good system allows rapid correction in daily work, then stops at intervals to let slow outcomes decide whether the game itself is worth continuing.
I add three notes to every number: Which part of the chain does it observe? What costs does it omit? When will this cohort be mature? These three questions restrain impulse better than another real-time dashboard.
First Ask What Authority the Number Has
When I see a metric now, I try not to ask immediately whether it is “good.” I ask first what it has the right to decide.
Ten thousand views can influence the opening of the next piece; they cannot decide whether to invest several years. One customer’s praise can clarify a use case; it cannot prove a market. One failure deserves an examination of the action; it may not be enough to reject the direction.
Keeping each metric within its authority prevents the newest and brightest number from leading us around.
This brings an uncomfortable conclusion: the closer a metric is to the final outcome, the less suitable it is as a source of daily emotional reward. If we must receive daily confirmation of “ultimate success,” we will inevitably let proxy metrics exceed their authority. A mature feedback system arranges not only the data, but also what sustains a person while waiting.
We need fast feedback because a life cannot afford years of blind flight. We also need slow outcomes because many important costs do not warn us in time.
Part of judgment is the willingness to wait for the bill after the applause has begun—and, before the final result arrives, to make only the limited adjustments justified by the signals already within our reach.
Responses