By the time most people finish researching a company, they have already decided. The research that follows is gathering support, and it feels exactly like objectivity from the inside.
Writing a full bull case and a full bear case is the cheapest correction for this. It is also easy to do badly, in a way that makes the bias worse rather than better.
The failure mode: two strawmen
The bad version writes a strong case for the conclusion already reached and a weak case for the alternative, then presents the exercise as balance.
This is worse than not doing it, because it launders a predetermined conclusion through a process that looks rigorous. A reader sees both sides considered and gives the output more credence than it earned.
The test is simple and uncomfortable: would someone who genuinely held the opposite view recognise their argument in what you wrote? If they would call it a caricature, you have not written the case.
What each case has to contain
Both sides need the same three components, or they are not comparable.
A mechanism. Not a sentiment — a causal story. “They will grow” is not a bull case. “Their per-seat model means expansion revenue arrives without new sales cost, and their published customer count against their sales headcount suggests they are already realising it” is a mechanism.
Evidence. Specific, sourced, dated observations supporting the mechanism.
A falsifier. What observation would show this case to be wrong? This is the component almost always missing, and it is what makes the analysis usable later — it tells you what to watch.
Building the bear case honestly
Harder than the bull case, because optimism is the default in company research — you are usually researching a company you are already inclined toward.
Three techniques help:
Start from the failure, work backwards. Assume it is two years from now and this went badly. What happened? This is a pre-mortem, and it surfaces risks that forward-looking analysis skips because they feel unlikely in prospect.
Read the negative evidence first. Before forming a view, read the three-star reviews, the complaint threads, the departed-employee commentary. Doing this early prevents the confirmation filter from engaging. See what customers say vs what they rate.
Ask what the model requires. Every business model has load-bearing assumptions — a retention rate, a conversion rate, a cost that stays flat as volume grows. Name them, then ask which is most fragile. See reading a business model from public signals.
Building the bull case honestly
The mirror problem exists too, particularly in due diligence contexts where finding problems feels like doing the job well.
A bear-only assessment is not rigorous, it is just negative — and it is easy to produce, because every company has visible flaws. The discipline is asking what has to go right, and whether there is evidence it is going right.
A useful prompt: what do this company's most enthusiastic customers know that the market has not priced in? Their five-star reviews are usually written by people who found something real, whatever else is true.
The synthesis
Two cases side by side is not a conclusion. The synthesis is the third piece of work, and it answers a specific question: which case does the current evidence favour, and what would change that?
Three components:
- Where the cases disagree. Usually they agree on most facts and differ on one or two interpretations. Naming those pivot points is most of the value.
- Which evidence is stronger. Not which case is more appealing — which rests on better-sourced, more recent, more independent evidence.
- The watch list. The specific observations that would move you from one case to the other. This converts a static assessment into something that stays useful.
Avoiding false balance
A real risk in the other direction: presenting the cases as equally weighted when the evidence is lopsided.
If the bear case rests on one speculative concern and the bull case on six documented observations, saying so is the honest output. Balance means giving each case a fair hearing, not giving each equal weight in the conclusion.
An assessment that always lands at “it could go either way” has avoided the work rather than done it.
Where this applies
Beyond investment. The same structure works for a vendor decision (what happens if they fail, and what happens if they succeed and raise prices), a partnership, a hire, and an acquisition.
In every case the value is the same: it forces the argument you were avoiding into writing, where it can be examined.
Our Company Deep-Dive closes with a bull and bear synthesis built this way — both cases argued in full, evidence weighted rather than balanced for appearance, and falsifiers named so the assessment stays useful after delivery. The full structure is in how to research a company.