What was submitted
The application form asks four things: what keeps happening, one recent example, what it is costing, and what you think is going on and what you have tried. The submission, condensed:
What keeps happening. “I’ve been deciding for seven weeks whether to leave my job. I’ve done more research than I’ve done for any decision in my life. I’m no closer. Every time I think I’ve found the thing that settles it, it doesn’t.”
The example. She (call her R.) is 34, a senior product manager at a mid-sized logistics software company. A former colleague, now founder of a nine-month-old startup, has offered her the role of first product hire: a 30% pay cut, meaningful equity, and a seed round that she was told is “signed, closing in about three weeks.” At her current company, her manager has said a director promotion is “likely in the next two quarters,” not confirmed. The founder wants an answer by the end of the month. The offer has been informally open for three weeks. She has already asked once for more time. On her situation: “We have about four months of expenses saved. My partner earns roughly what I do. We have a mortgage. We’ve talked about kids in a couple of years, but nothing is planned.”
Her research so far: three spreadsheets (compensation scenarios, equity outcomes, a weighted criteria matrix), eleven conversations with people who have joined or left startups, reading on early-stage failure rates, two long calls with the founder. Her leaning has flipped four times, each time after the most recent conversation.
The cost. She estimates two to three hours a day thinking about it, mostly at night. Her partner has said they are “done talking about it in circles.” She is not sleeping well, which she attributes to the decision. Work at her current job has slipped; she describes “half-attending” meetings.
What she thinks is going on, and what she has tried. “I think I’m afraid of making the wrong choice. I’ve tried pros and cons lists, I’ve tried imagining myself in five years in each, I’ve tried a weighted matrix. The matrix gave a 52/48 result. I didn’t trust it and reweighted it. Then it gave the opposite result.”
Sorting what we have
Reported, taken at face value: the seven weeks, the flipping, the volume of research, the partner’s fatigue, the sleep and attention costs.
Established, as concrete terms rather than interpretations: the offer’s pay and equity; the founder’s deadline; a promotion that is possible but unconfirmed; four months of savings; a partner with similar income; a mortgage; children discussed but not scheduled.
Not known, and material:
- Whether the round closes on schedule. “Signed, closing in three weeks” comes from the party who wants a yes.
- Whether the promotion happens. “Likely” is not a commitment.
- What she would do if the startup failed at month eighteen. She has researched failure rates, not her own recovery plan.
- What she wants most from the next three years. Nine criteria in a matrix; no ranking.
- How her partner weighs it, as distinct from being tired of the topic.
- How she has made large decisions before. This tells us whether this is a pattern or a single hard case.
Seven weeks on a decision that changes income, equity and daily life is not by itself a sign of anything wrong. The signal is narrower: the research has stopped changing anything. Four flips, each tracking the last input, means new information is producing motion without direction. That is what needs explaining.
Plausible explanations
There are at least five, and they call for different responses. The review’s job is not to pick one from the armchair but to say what would tell them apart.
1. The options are close on her real criteria, and more information cannot separate them. If both paths score within a few points on what she actually cares about, every new fact tips the balance and the next fact tips it back. That is not a failure of judgment; it is what a near-tie looks like from inside. The mistake would be reading the flipping as proof she has not researched enough.
2. She has not ranked what she wants, so every fact re-weights everything. Nine criteria, and she reweighted the matrix when she disliked the result. The weights were never settled. If security, ownership, status, learning speed and her partner’s stability are all “important,” a conversation about dilution makes equity feel decisive and a conversation about layoffs makes security feel decisive. The research is substituting for a smaller, harder task: deciding what matters most.
3. Continued research is protecting her from owning an outcome. While the decision is open she has not chosen, and if she has not chosen she cannot be wrong. On this reading, “one more thing to check” keeps the exposure away. Relief when a new question appears would fit, though relief is also what someone feels when a real gap is finally named. The extension request fits too but proves nothing; people also ask for time when facts are genuinely pending.
4. A real unresolved dependency exists, and the decision should be conditional rather than delayed. Two facts she lacks (whether the round closes; whether the promotion is confirmed) bear directly on the choice. If those are what she is waiting for, the problem is not psychological. She is treating a conditional decision as if it had to be unconditional.
5. The decision is partly not hers alone, and the research is building a case. She may already have a preference and expect her partner to resist it, so the research assembles justification rather than seeking information. Nothing in the submission establishes this. It is listed because it is common, and because if true, no spreadsheet will resolve it.
None of these requires a hidden motive. 1, 2 and 4 are ordinary decision problems; 3 is a pattern; 5 is a relationship question.
What would distinguish them
On explanation 4 (a real dependency): If the round closed tomorrow and the promotion were confirmed in writing, would she know what to do? If yes, immediately, the difficulty is not internal and most of this review is moot. If she would still be stuck, the missing facts are not what is holding her.
On explanation 2 (unranked criteria): Can she name, in order, the three things she most wants from the next three years, without reference to either option? If she can, and the options still tie, we are in explanation 1. If she cannot, or keeps adding a fourth, we are in explanation 2 and research should stop until the ranking is done.
On explanation 1 (a genuine near-tie): Do her flips track the most recent input every time? The submission suggests yes. Second check: of the last five things she researched, could any result have changed her decision? If none had that power, the research was not deciding anything. That is consistent with 3, but equally with 2: without ranked criteria she cannot know what would flip her. So this observation narrows the field to 2 or 3 rather than settling it.
On explanation 3 (research as protection): A test she can run: draft both messages to the founder, one accepting, one declining. Send neither. Notice which draft she stalls on, and which produces “but what if” the moment it is written. Neither reaction proves anything on its own; stalling on one draft is also consistent with a plain preference she has not admitted, which would point to 5 as much as 3. But it is an observation about a decision being made rather than about an imagined future, and it costs an evening.
On explanation 5 (the partner): Has she told her partner her current leaning in one sentence, and asked for theirs in one? If not, that is the next conversation, before any further research.
On whether this is a pattern: How did she make her last two large decisions? If she chose within a reasonable window, this is probably specific to this decision. If she describes the same stretch of research, flipping and exhaustion, the pattern matters more than the decision, because it will return at the next fork whatever she chooses now.
How the answers change the recommendation
If it is a near-tie (1): stop researching. Choose a tie-breaker in advance and apply it. The one that fits her situation is recoverability: which choice, if it turns out badly, is easier to undo in two years? Her stated facts bear on this but do not settle it. Four months of savings and a partner with similar income say something about how long a gap could be covered; they say nothing about the household’s fixed costs, what the mortgage requires each month, how a 30% cut changes the monthly balance, or how quickly a comparable role could be found in her market. Before recoverability can serve as the rule, three things need checking: the monthly shortfall on the lower salary, how many months the savings actually cover at that shortfall, and a realistic time-to-rehire, which people who have recently made the return move can tell her better than I can. Fixing the rule in advance means the rule decides the tie, instead of the most recent conversation. (A note on evidence is at the end.)
If the criteria are unranked (2): the ranking is the whole task. Three criteria, in order, written down. Then re-read both options organized around those three, while keeping in view any hard constraint (the mortgage, the runway) and anything outside the three that could decide the matter on its own. The ranking sets the order of the comparison; it does not exclude a decisive consideration because it came fourth. The decision may turn out lopsided across three where it felt balanced across nine. If it is still balanced, that is itself an answer, and she is in 1.
If research is protecting her from committing (3): the decision matters less than how it gets closed. A fixed date, and an explicit acceptance that the choice will be made on incomplete information because that is the only kind available. If “but what if” arrives the evening after she decides, it should be treated as a predictable after-effect, not new evidence. This is also the reading under which the pattern recurs at the next decision, so it is where ongoing work would have the most to do.
If there is a real dependency (4): make the decision conditional and say so. “If the round closes by the 30th, I accept. If it slips, I stay.” Honest with the founder, ends the research, and hands the remaining uncertainty to the person who can resolve it.
If the partner conversation has not happened (5): have it. Two sentences each. Then return to the rest.
A proportionate next step
Given the deadline, this is a one-week plan, not a program.
- Day 1: write the three-criteria ranking. Fifteen minutes, no reference to either option.
- Day 1–2: ask the founder one factual question in writing: what is the closing date for the round, and what happens to the offer if it slips by a month? Ask her manager one factual question: what would have to be true for the promotion to be confirmed, and when would she know?
- Day 2: the two-drafts test. Note the reactions.
- Day 3: the two-sentence exchange with her partner.
- Day 5: decide, using the ranking and, if needed, the recoverability tie-breaker. If a real dependency remains, decide conditionally and tell the founder the condition.
Stopping rule for research: no further research unless she first writes down the question and can say in advance what result would change her decision under her ranked criteria. At most two such questions before the decision date. Two, because the deadline is under two weeks away and each question has been costing her a day or more; the number is provisional and specific to that deadline.
Conditions that would justify revisiting the decision once made:
- A material term changes: the round fails to close, the offer is altered, the promotion is confirmed or refused.
- She learns something that changes a reason. That includes information bearing on a ranked criterion, but also a factor the ranking left out that she now thinks belongs in it, an error in her original reasoning, a change in what she wants, or an option that did not exist when she decided. The test is not whether information is new but whether it changes a reason.
- A scheduled review point. Six months is suggested because it is long enough to know whether the round closed and the role is what was described, or whether the promotion happened. That is a figure for this case, not a rule.
A feeling on a Sunday, or a friend’s opinion, does not by itself reopen the decision. It can prompt her to write down one specific question and check it against the Day 1 record. If the check turns up something on the list above, the decision is reopened on purpose. If not, it stays closed.
Confidence and limits
This review is strongest on explanations 1, 2 and 4, which can be tested in days. It is weakest on 3 and 5, for which the submission holds no direct evidence; they are possibilities, not findings. A reviewer who wrote confidently about “fear of commitment” from this submission would be guessing. Several observations above are consistent with more than one explanation, and the review says so rather than forcing a verdict.
One note outside this review’s scope: she attributes her sleep to the decision, which is plausible. If the sleep problem outlasts the decision by more than a few weeks, that is a question for a physician.
Note on evidence. There is published research associating a disposition to keep searching for the best option with more regret and lower satisfaction: Schwartz, B., Ward, A., Monterosso, J., Lyubomirsky, S., White, K., & Lehman, D. R. (2002). Maximizing versus satisficing: Happiness is a matter of choice. Journal of Personality and Social Psychology, 83(5), 1178–1197. That evidence is correlational and concerns a personality measure. It does not show that a tie-breaker rule improves decisions, and it is not cited as support for the recommendation above. It is mentioned only because it bears on how much weight to give the impulse for one more check, and it says nothing about the hypothetical person in this case.