The Field-Switch Assumption Test, Nine Calls to a Dated Go or No-Go

This guide is for anyone thinking about a field switch who wants to know whether the field is what they imagine before they pay for retraining or rebuild a resume around it. It follows one switch, a high school science teacher testing UX research, from a written list of assumptions through LinkedIn outreach and nine informational calls to a dated go, pivot, or no-go record. Every fork is shown, including the wrong turns.

The worked case is a composite. The teacher and the call outcomes are illustrative, there to show how scoring works. The profile counts come from Refolk's index, and the reply rates and saturation figures come from published sources, attributed where they are used.

What does an informational interview for a career change actually test?

A useful informational interview tests one written assumption your switch depends on. It does that by asking about the other person's recent past, so that the answer can confirm the assumption or kill it. A call that only collects impressions of "a typical day" cannot move a decision.

The pages that rank for this topic are mostly question lists. CareerOneStop, UMBC and Career Contessa (40+ questions) ask about opinions or the future, and none of them shows a specific assumption being tested and then confirmed or killed. CareerOneStop's list even includes "Are my skills and interests a good fit". That is a leading question, and the answer will almost always be a polite yes.

Three terms do most of the work here:

• Assumption ledger: a short written list of the claims your switch depends on, each paired with a kill condition.
• Kill condition: the specific thing you could see or hear that would prove a claim false.
• Saturation: the point in a set of interviews where new conversations stop producing new information.

Not every assumption deserves a call. Some are answered by published data in an afternoon, and spending a stranger's 20 minutes on those wastes the scarcest thing you have. The matrix below sorts each claim before any outreach starts.

The assumption test, step by step

The procedure runs from a written ledger to a dated decision in about three to four weeks. Most of the time goes to waiting for replies, not to the calls.

Rough time budget: the ledger takes 2 to 3 hours, the desk check one day, and the contact list one day. Outreach runs over weeks 1 to 3, with calls booked as replies arrive.

Sources disagree on the order. Amherst and Interview Guys say to start with your own network. The Mom Test (Rob Fitzpatrick) puts desk research before any conversation. My view: do the desk check first, because it removes questions you would otherwise waste a contact on. Send warm-intro requests on the same day, though, because a reply can take a week to arrive.

The worked case: a teacher's ledger and desk check

The ledger is the whole method in miniature. A claim with no kill condition is a hope, and you cannot schedule a call to test a hope. Here is the teacher's ledger as first written, before any desk research:

1. A1, credential: a bootcamp certificate, not a master's degree, is enough to get a career changer hired. Killed if recent switchers or hiring managers say something else got them the first interview.
2. A2, junior roles: entry-level openings exist in numbers someone outside the field can reach. Killed if hiring managers describe junior openings as rare or filled internally.
3. A3, pay: entry pay is at or above the teacher's current salary. Killed if recent switchers' first offers came in below what they earned before.
4. A4, setting: the work happens on in-house product teams. Killed if most people reached work at one large platform or at a consultancy.
5. A5, the work itself: most of the week is spent with users, not building decks. Killed if walkthroughs of a recent week show user contact as a minority of the time.
6. A6, staying power: UX research is a destination, not a stop on the way to product management. Killed if leavers describe moving to product management as the normal path.
7. A7, demand: demand for the occupation is not shrinking. Killed if the published outlook says it is.

The desk check

Two public sources cover most of what a desk check needs. Each Occupational Outlook Handbook (OOH) profile covers typical duties, work environment, the education and training typically needed to enter, median pay, and the job outlook over the coming decade. Its pay figures come from the BLS Occupational Employment and Wage Statistics program. The OOH has more than 300 occupational profiles covering about 4 out of 5 jobs, and each profile is split into nine tabs. O*NET reports cover tasks, knowledge, skills, abilities, work activities, work context, job zone, interests, work values and work needs. The Job Zone data includes fields for education, related experience, job training and SVP range.

Check one thing before trusting either source: does your target title have its own profile, or is it folded into a broader occupation? I have not verified how UX research is classified, so look it up for your own target. If the closest profile covers a broader group, its median pay describes that group, not your entry offer.

For the teacher, the desk check did two things:

• Closed A7. The outlook question is answered by the OOH outlook tab. Whatever that number says, it is not a question for a call.
• Narrowed A3. The desk data gives a median, and a median is not an entry offer. A3 stayed open, but as a narrower question: what did recent switchers' first offers look like?

A1 stayed open too. A typical education level from O*NET is a starting point, but the teacher's real question is what hiring teams actually accept from someone switching in.

The index check, and the first wrong turn

Next, the teacher sized the pool of insiders. In Refolk's index of professional profiles, a search for US profiles titled UX Researcher or User Researcher returned 1,393.

The wrong turn came next. To answer A2 from data, the teacher added an Entry Level seniority filter. The count stayed at 1,393. A filter that narrows nothing has been silently ignored, so the entry-level share is not established. A2 had to go to calls.

The 25-profile sample had two things to say. Meta held 7 of the 25 (28%), which matters for A4: a quarter of the insiders you reach may describe one company's culture. And 18 of the 25 (72%) were titled "UX Researcher". The UK picture is different, and it shows why search vocabulary decides who you can even reach:

All columns are from Refolk's index, and the percentages are derived. The index measures coverage, not the size of the labor market. In the UK sample, 36% work at two consultancies, so a UK switcher would need to test the "in-house product team" claim much harder than the US one.

How many informational interviews before changing careers?

Nine calls is enough only if you count evidence for each assumption separately and keep each contact tier internally similar. Research on interview saturation puts the floor for common themes at about 12 interviews, and higher for mixed groups.

Guest, Bunce and Johnson found that 12 interviews surfaced 88% of their codes, with basic elements present by six. Hennink and colleagues separate two kinds of saturation. Code saturation means you have heard every topic, which typically took 9 interviews. Meaning saturation means you understand each topic, which typically took 16 to 24.

Nobody has published a stopping rule for career-switch calls. These studies were about research interviews, not about job seekers, so treat them as a guide to scale, not a rule.

The stopping rule I use is derived, not published. Stop when two calls in a row add nothing new on any assumption, and every open assumption has been tested by at least two calls from different tiers. Without the second condition, two quiet calls from the same kind of person will look like saturation when they are only an echo.

Who to contact, and how many requests nine calls take

Plan your list from reply rates by tier, because those rates can differ by a factor of ten. Warm introductions are cheap to get but tilt toward happy insiders. Leavers and hiring managers are expensive but carry the evidence most likely to kill an assumption.

The five tiers:

• Warm: people you know, or people one mutual contact will introduce you to.
• Alumni or shared employer: a shared school or past employer, here former teachers now working in UX research.
• Cold insiders: current practitioners you have no link to.
• Recent switchers: people who entered the field from another career within the last few years.
• Leavers: people who left the field, who are the only ones likely to tell you why people go.

No one has published reply rates with a stated methodology, and every figure below is a practitioner estimate. One finance-recruiting source gives 5 to 10% for cold outreach, 20 to 30% for college alumni, 50 to 85% for high school alumni or hometown connections, and 70 to 90% for warm introductions. Another source claims 30 to 40% for well-written cold messages, which conflicts with the first.

The counts are derived (9 divided by the rate) and assume every reply becomes a booked call, which is optimistic.

The teacher's plan, and the second wrong turn

The first contact list had 40 names, all of them warm contacts or former teachers now in UX research. That meant zero leavers and zero hiring managers. Every one of them had stayed in the field, so nobody on the list could test A6.

The rebuilt plan, by call:

• Calls 1 to 2: warm insiders. About 4 to 5 requests at 70 to 90%.
• Calls 3 to 4: former teachers now in UX research. About 7 to 10 requests at 20 to 30%.
• Calls 5 to 6: recent switchers who came through a bootcamp. Cold, so 20 to 40 requests at 5 to 10%.
• Call 7: a hiring manager. Cold, so 10 to 20 requests.
• Calls 8 to 9: people who left UX research for product management. Cold, so 20 to 40 requests.

At the pessimistic cold rate, the five cold calls alone need 50 to 100 requests. At the optimistic rate they need about 13 to 18. A list of 40 to 60 names covers the space between those two numbers. If week-one replies come in at the low end, add cold names right away rather than waiting. The weekly invitation ceiling is reported as roughly 100 a week, but that is community consensus, not an official number, and this plan stays below it.

How to ask for an informational interview on LinkedIn

On a free LinkedIn account, plan for 3 personalized connection notes a month, each up to 200 characters. Invitations without a note are still allowed after the notes run out. That makes the notes the scarcest resource in the whole test, so spend them where a stranger has no other reason to accept.

LinkedIn's own help pages have contradicted each other. The newer page says three notes a month at 200 characters. An older page says five, and many public guides still repeat five. Premium accounts get unlimited notes at 300 characters. LinkedIn does not say whether "per month" means a calendar month or a rolling 30 days. If your test crosses a calendar-month boundary you might get up to six notes, but that is unverified. Plan for three, and re-check the Help Center page on connection request limits before you start, because this is the kind of number that changes.

The teacher's third wrong turn was spending note one on a former-teacher alumna who would have accepted a plain invitation. That left two notes for three high-value strangers, and the second leaver had to wait for the monthly reset.

Follow-up timing comes from Duke's informational interviewing guide. If there is no response after seven business days, send a nudge. After another seven business days, try once more, then move on. Duke, Dartmouth and WMU all say to send a thank-you within 24 hours, and WMU says plainly that informational interviews get thank-you notes too.

NOTE (free account, 200 characters max)
Hi [Name], I teach high school science and I'm testing whether UX research fits me before retraining. Could I ask you 3 questions about your move out of research? 20 min, your schedule.

NUDGE (7 business days after they accept with no reply)
Hi [Name], a quick nudge on my note. I have three specific questions about [one assumption, e.g. how junior hires get past the screen] and can work around your calendar. If now is a bad time, no problem at all.

THANK-YOU AND REFERRAL ASK (within 24 hours of the call)
Thank you for the time today. What you said about [specific thing they said] changed how I'm thinking about [assumption]. Could you recommend a couple more people I should talk to, especially anyone who [tier you are short on, e.g. hires junior researchers or left the field]?

How often an informational call leads to a referral or introduction has not been established publicly. Ask anyway, and log the answer, so you learn your own rate.

Questions to ask in an informational interview as a career changer

Write one question per open assumption about a specific past event, phrased so the honest answer could kill the assumption. Opinion and prediction questions get polite answers, and polite answers cannot be scored.

The best-documented method is The Mom Test. It has three rules: talk about the other person's life, not your idea. Ask about specifics in the past, not generalities or opinions about the future. Talk less and listen more. Its model question is "Talk me through the last time that happened." It also advises asking at least one question per conversation that could destroy the plan you currently have in mind. Watch for the phrasing it flags as fluff: "Would you ever... Do you think you... Could you see yourself...".

A call that cannot kill an assumption is a pleasant conversation, not evidence.

The teacher's first two calls went wrong because the script asked "Do you think a teacher could do well in research?" Both insiders said yes, warmly, and that told the teacher nothing. Those answers were scored as zero evidence, and the script was rewritten:

I found no career-center page that publishes tested wording for questions about entry pay or credential requirements. The A3 and A2 rows are my own suggestions, so check them against your first two calls and revise. Prepare your top three questions for each type of contact. A 20 to 30 minute call has room for three good questions plus the follow-ups that "the last time" questions produce.

How the assumption test goes wrong

Most failed tests do not fail for lack of calls. They fail because the calls could not have produced a no. These eight failure modes cover most of it, and each comes with a check.

1. Leading questions. "Is this a good field for career changers?" invites a polite yes. Check: rewrite every question until it asks about something that already happened, such as where the team's last hire came from.
2. Talking only to happy insiders. If all nine calls go to people who stayed in the field, nobody tells you why people leave, and A6-type claims get confirmed by default. Check: hold at least one call with a leaver and one with a hiring manager before you score anything as confirmed.
3. Miscounting saturation. Nine calls across three tiers is three per subgroup. Two quiet calls in a row can simply mean two people with the same view. The teacher almost stopped after call 4, because calls 3 and 4 echoed calls 1 and 2, but A6 had not been tested by a single call. Check: count saturation per assumption.
4. Using up the note cap too early. Spending notes on people who would have accepted a plain invitation leaves you nothing for leavers. Check: send plain invitations to anyone with a shared school or employer.
5. Using one market's job title in another. Searching only "UX Researcher" in the UK misses about half the sample, which is titled "User Researcher". Check: sample 25 profiles in your market and list the titles before you build the contact list.
6. A search filter that is silently ignored. The Entry Level filter returned the same 1,393 as the unfiltered search. Check: compare totals before and after every filter, and do not trust a segmented number until the total actually moves.
7. One big employer dominating the sample. One employer holds 28% of the US sample, so one company's culture can drown out the rest of the field. Check: cap any single employer at two calls, and tag each piece of evidence with where it came from.
8. Expecting too many replies. If you plan for 30 to 40% cold replies and get 5 to 10%, you need three to four times the list. Check: compare week-one reply counts with the plan and top up the cold tiers right away.

Scoring the calls and writing the decision record

After every call, mark each assumption confirmed, killed, or open, and record which tier the evidence came from. Once the stopping rule is met or the call budget runs out, write a dated decision: go, pivot, or no-go.

Here is the teacher's ledger after nine calls (illustrative):

The stopping rule did not trigger. Calls 8 and 9 both added new information on A6, so the test ended because the call budget ran out, not because of saturation. That is why the decision is a pivot rather than a go.

The decision was a pivot. Do not buy the bootcamp, because A1 was killed. Build two portfolio studies from classroom research instead. Hold three more calls aimed at the open items: a hiring manager outside the big platform for A2 and A4, and a third leaver for A6. Decide again in six weeks. The bootcamp fee stays unspent until A1 comes back with a different answer.

DECISION RECORD
Field tested:
Date opened:            Date decided:
Calls held: [n]  Warm [ ]  Alumni/shared [ ]  Cold insiders [ ]  Switchers [ ]  Hiring managers [ ]  Leavers [ ]

A1 [claim] | confirmed / killed / open | calls that tested it | strongest evidence
A2 [claim] | confirmed / killed / open | calls that tested it | strongest evidence
(one line per assumption)

Decision: GO / PIVOT / NO-GO
Because: [the confirmed or killed assumptions that drove it]
Spending unlocked: [retraining, resume rebuild, applications]
Spending blocked: [what you will not pay for yet, and why]
Open items and the call or source that will close each:
Re-decide by: [date]

What to do after a go, pivot, or no-go

Each outcome triggers different work. A go unlocks the resume rebuild and the first applications. A pivot sends you back to a shorter ledger with a new date. A no-go is a finished result, not a failure, and it should cost you nothing more than the calls.

After a go, turn the confirmed assumptions into the resume. The killed ones tell you what not to lean on. The teacher, for example, would lead with portfolio work rather than a certificate. The Field-Switch Resume Rebuild guide covers the task crosswalk. Refolk writes the resume from your own work history and tailors it to each posting, so a rebuild based on confirmed assumptions does not have to be redone by hand for every application.

After a pivot, there is a cheap check you can run before booking the extra calls. Score your current resume against a handful of real junior postings. Refolk's fit score shows which gaps are real on paper, which can confirm or contradict what the calls suggested about credential requirements.

To keep the test current, re-check three things that change:

• LinkedIn's note limit. The help pages have already shown two different numbers.
• The OOH profile you used. Re-read its outlook and pay tabs when the Handbook is revised.
• O*NET's work activities data. Releases add data in batches, and one recent release added Work Activities and Work Context data for 177 occupations.