Sales Targets Choose the Numbers You Want to See — Anatomy of Motivated Reasoning

Quarterly quotas, share targets, early post-launch data. When people search for evidence with those numbers already in mind, they unconsciously reach for evidence that confirms the conclusion they want. This cognitive distortion — motivated reasoning — operates without malice. The author does not feel they lied. Yet the hand selecting the data was moved by pressure aimed at a target. Real incidents reflect that structure precisely.

So What / So Why — The Core of This Installment

Motivated reasoning is a concept formalized in 1990 by social psychologist Ziva Kunda. It describes a cognitive process in which the desire to reach a preferred conclusion unconsciously distorts which evidence is gathered and how it is evaluated. The critical point is that the person reasoning has no sense of being biased. Motivation functions like a search engine: the instruction to "find valid data" is silently converted into "find data showing validity." Once the direction of search is set, data from different conditions gets accepted because "the result itself is real," while inconvenient data recedes as "supplementary reference material."

Two reasons make this phenomenon particularly serious in pharmaceutical marketing. First, physicians do not independently verify the assumptions behind the evidence they receive. Reviewing approval reports or package inserts at the point of prescribing almost never happens. Even when the information reaching a physician has already been filtered, the marks of that filtering are invisible. Second, because the manipulation is "not a lie," correction after the fact rarely occurs. The numbers come from real publications. The values in the charts are accurate. The problem is what was shown and what was withheld — and that only becomes apparent when the reader goes back to the original source.

The fact that the Ministry of Health, Labour and Welfare's monitoring project reports have recorded the same category of incidents every year from 2019 through 2025 demonstrates that this is not an individual ethics problem. It is what happens when a pressure structure and a cognitive mechanism work together. The answer to "why does this keep happening?" lives inside the person who acts.

The Structure of Pressure — Who, By When, Against What Numbers

At the end of a quarter, a regional manager opens a share dashboard. Post-launch six-month share targets are set for many products as a specific percentage threshold, and falling below that number triggers a PDCA cycle. The materials team member hears that number directly sometimes; other times it arrives indirectly as a request to "add more detail sessions with physicians" or to "sharpen the key message for doctors."

Pressure comes from multiple directions simultaneously. ① Regional managers and division heads checking progress against share targets and prescription counts. ② Requests from Medical Affairs or the medical division to "reflect the views of key opinion leaders." ③ Time pressure from competitors — a rival's new indication creates urgency: "They just got Indication A approved. Make our advantage clear." ④ Incentive structures that tie bonuses and promotion evaluations to share. ⑤ Fear of having materials rejected at an internal review meeting with comments like "this won't resonate with physicians."

Pressure concentrates most heavily in the six to twelve months after launch. During that window, evidence is thin — only approval-trial data exists — while expectations on the sales side peak. The question "how do we present limited data compellingly?" is itself the entry point for motivated reasoning. The same structure appears right after an indication expansion during conference season. Data on the new indication is sparse, yet differentiation from competitors is demanded. So subgroups from trials conducted in a different context, or surveys from different diseases or countries, get pulled in as "supporting material."

Internal Reconstruction — Belief, Feeling, Deeper Psychology

Belief (what the person believed was correct)
Conviction that "this drug works" often rests on genuine clinical grounds. The drug received approval. A pivotal trial showed a statistically significant difference on the primary endpoint. The materials author starts from the position that "the efficacy of this product has been established." That belief itself is not wrong. The problem lies in the next step. The question "how do we communicate established efficacy?" slides smoothly into "which data shows efficacy most persuasively?" At that moment of conversion, the direction of the search is set.

Feeling (the emotional state at the time)
The day after being told in a quarterly meeting that "the message isn't compelling enough for physicians," a materials author reviewing their work carries both pressure and a sense of mission simultaneously. "Prescriptions aren't growing because I'm not communicating effectively" blends with "this is a genuinely effective drug and I should be showing it better." In that state, narrowing the data feels like "organizing information." Omitting unfavorable data feels like "focusing on what the audience actually needs." Emotional pressure locks the cognitive filter in one direction.

Deeper psychology (the four drivers operating)
① Motivated reasoning moves first. The conclusion — "this product should be superior" — comes first, and the act of "searching" for subgroups, endpoints, and publications that support it follows. Data from different conditions is accepted because "the numbers are real." The differences in excluded conditions recede as "technical footnotes." This is what the author likely felt: "This subgroup data is honestly very good. Physicians will surely see it the same way."

② Local rationalization follows. "If I just skip the background conditions on this one graph, I can fill in verbally." "It's just one slide — it's not a serious problem." Each individual judgment looks small. But when every team member makes the same rationalization, the materials as a whole become systematically skewed. What looked "rational" locally becomes a deviation at the aggregate level.

③ The sin of omission kicks in. "I wasn't asked, so I didn't say." The underlying assumptions, excluded patients, differences in trial design, unfavorable side effects — none of these were actively falsified. The internal narrative is "I answered the questions I was asked accurately." The failure to disclose is processed not as a choice to conceal but as "not having been asked to disclose." Omission carries less guilt than commission.

④ Responsibility externalization completes the picture. "Physicians asked for Japanese data, which is why we used only the Japanese subgroup" (FY2019 ②-13, discussed below). "The KOL reviewed this slide in advance." "The materials passed the supervisor's review." When the rationale for a choice feels external, critical self-examination of one's own judgment stops. When responsibility is diffused, no one intervenes.

The Real Incidents Underneath

Each of the three cases below is documented in actual Ministry of Health, Labour and Welfare monitoring project reports. The words of those involved and the structure of their actions show that the internal reconstruction above is not a theoretical exercise. More detailed accounts of each incident are available on the individual pages of the analysis series.

Case 1 — "Because Physicians Ask for Japanese Data" (FY2019, Bronchial Asthma Treatment)

For the primary endpoint (annual asthma exacerbation rate) of a Phase III international joint clinical trial, the full-analysis dataset (approximately 250 patients per arm) was not presented. Only the results from the Japanese subgroup (approximately 15 patients per arm) were included in the slides and pamphlets for in-hospital product presentations. When a healthcare professional noted the inconsistency — other secondary endpoints used full-analysis data, yet the primary endpoint alone appeared as a Japanese subgroup — the responsible person's response was: "Because physicians ask for Japanese data." The Japanese subgroup showed a better result than the full-analysis set.

This statement is a textbook display of all four drivers. Motivated reasoning (searched for the subgroup with the favorable result), local rationalization (changed only one graph, the primary endpoint), the sin of omission (merely did not show the full-analysis data), and responsibility externalization (the physician's preference dictated it). There was almost certainly no internal sense of "I lied." Yet the monitoring report explicitly notes that the selection of the primary endpoint presentation was arbitrary.
→ Analysis vol-01 "Use and Manipulation of Data That May Cause Factual Misunderstanding", vol-02 "Arbitrary Extraction, Manipulation, and Presentation of Data and Graphs"

Case 2 — Showing Only the "Favorable Window" of a KM Curve for a Drug Whose Superiority Was Not Confirmed (FY2019, Oncology)

Despite the appropriate use guide explicitly stating that "superiority of this product over the control arm has not been confirmed," an oral presentation at a new drug briefing highlighted only the portion of the Kaplan-Meier curve during which the investigational arm exceeded the control arm, and claimed superiority. Additionally, information unfavorable to the product — namely that a shorter overall survival was observed in hematopoietic stem cell transplant patients compared to the control — was withheld.

The internal logic of this case is transparent. The conclusion (this product is superior) comes first. A period of the KM curve that supports the conclusion is selected. Information that does not support it (OS shortening) goes unmentioned. Looking at the full KM curve, the fact of "unconfirmed superiority" would have been obvious. That attention still went to the "period of outperformance" is consistent with motivated reasoning narrowing the search to "show what you want to see."
→ Analysis vol-01 "Use and Manipulation of Data That May Cause Factual Misunderstanding"

Case 3 — Extracting Only Two Favorable Items from a Publication Covering a Different Disease, Country, and Healthcare System (March 2024 Report, Dyslipidemia Treatment)

This product is administered subcutaneously at a medical institution every six months; the comparator allows self-injection at home every two to four weeks. At an online product presentation, patient and physician survey data was shown with an explanation to the effect that patients preferred injection every eight weeks over every two or four weeks, and preferred healthcare professional administration over self-injection. However, the survey population consisted of severe asthma patients (not an indication for this product) in overseas settings with substantially different healthcare insurance systems. The original publication compared five items across a graph, but the presentation materials extracted only the two items favorable to this product. The monitoring report records: "The company provided information based on materials created by extracting the parts of a graph from an overseas publication — covering a different healthcare system and patient population — that were convenient for their own product."

This case illustrates how far motivated reasoning can extend the search perimeter. The conclusion — "show this product favorably on administration convenience" — came first, and the search for supporting evidence turned up a publication with a different disease, different country, and different system. The authors almost certainly recognized the differences in conditions (disease, country, system). They used the data anyway, which is consistent with local rationalization operating on the basis that "the numbers themselves are real."
→ Analysis vol-01 "Use and Manipulation of Data That May Cause Factual Misunderstanding"