# Fieldwork kit — drafts only, unsent

No request has been sent and no interviewee recruited. Request existing aggregate records and data dictionaries; accept privacy-preserving bins. Do not request taxpayer identities or confidential case notes.

## Priority aggregate records requests

1. **City business denominator — Revenue / appropriate custodian.** For 2019–2025 and 2026 year to date if held, provide annual active operating businesses inside city limits by broad industry, employee-size bin, employer status and start year. Include definitions, address rule, duplicate treatment, exclusions, disclosure rule and revisions. Separate registrations from operating firms. Provide separate safe tables if a cross-tab would reveal identities.
2. **OSB / Prosper funnel.** Provide monthly eligible inquiries, distinct businesses, accepted cases, referrals, completed services, unresolved/abandoned cases, follow-up invitations/responses and outcomes, with safe district and industry aggregates. Include case time percentiles and censored open cases. Define “touchpoint,” “interaction,” “unique,” “served” and “resolved.” Reconcile 103+182+154+142 district businesses with 759 unique total and 1,600 touchpoints with 1,700-plus interactions. Identify OSB/IBRN/provider overlap without identities.
3. **Actual costs — Prosper / City.** For FY2019–20 through FY2025–26, provide actual expenditure, budget and FTE by service, provider, language access, administration and evaluation. Report grants, loan principal, guarantees, repayments, write-offs and pass-throughs separately. Supply contracts, scopes, performance definitions and fund-code changes; avoid double-counting transfers.
4. **Permit cohorts — City.** For commercial applications filed 2019–2026, provide monthly cohort counts filed, complete, issued, withdrawn, denied, expired and final-inspected by use, complexity and broad location. Supply median/p10/p90 elapsed days, separating city/other-agency and applicant waits; include dashboard definitions and safe business-size bins only if recorded.
5. **Procurement/payment — City / Port.** Provide annual distinct bidders, awards, contract value, invoices and paid suppliers by certification/size, industry and location, with first-time suppliers and days to payment. Distinguish prime/subcontractor and contract value/cash paid.
6. **Capital/premises — Prosper / Business Oregon.** Provide application, approval, denial, disbursement, default and closure cohorts with terms/reasons in safe bins. Request public premises/occupancy studies and vendor coverage. Approvals alone do not measure unmet credit need.

For each request, later track custodian, date, response, withheld fields, fees, alternatives, data dictionary and affected claims. Nothing above authorizes sending.

## Interviews and qualitative evidence

Recruit purposively across solo, 1–4, 5–19, 20–99 and growth-stage firms; storefront, home-based, healthcare/childcare, trades, manufacturing, technology and franchise models; neighborhoods and languages. Include closed/relocated firms, unsuccessful applicants and nonusers. Recruit workers independently of owner referrals. Log recruitment route and refusals. Interviews explain mechanisms and decisions, **not population prevalence**. Obtain recording consent and offer anonymity.

**Common opening:** What decision were you trying to make, when, and what alternatives did you consider? What specific cost, delay or uncertainty mattered? What evidence would change your interpretation? Record sector, approximate firm age, location, employment band and respondent role, with skip options.

**Owners and solo workers:** Walk through a recent month of revenue, expenses, own paid/unpaid hours, cash left after costs, benefits and volatility using ranges. What outside job or household support makes the business feasible? Who buys from you, including large firms? Describe the last hire, investment, expansion, relocation or choice to remain small. Trace a permit, capital or assistance case from first contact to resolution: contacts, hours, fees, handoffs, rejections and unresolved issues. Why did you use or avoid support? Which AI tasks would you trust, which need human review, and what data would you not share?

**Workers:** Ask actual pay, hours, tips/benefits, scheduling notice, paid leave, safety, advancement and stability. Compare differently sized employers *within the same occupation* where possible. Ask for a concrete change caused by employer growth, closure or public support, without leading the answer.

**Closed, sold and relocated owners:** Establish the last viable year and sequence of demand, rent, debt, owner health, succession, staffing and public-process factors. Distinguish planned exit, acquisition, retirement, failure and movement across a boundary. Ask where workers/customers went and what realistic intervention could have changed the decision.

**Unsuccessful applicants and nonusers:** What assistance did you know existed and how? Which eligibility, paperwork, language, timing or trust issue mattered? If denied or abandoned, ask for the written reason. What alternative was used, and what would have made participation worthwhile?

**Lenders, landlords, providers and anchor customers:** Lenders: applications versus approvals, pricing, withdrawals and viable firms declined. Landlords: improvements, deposits, vacancies, use rules and who bears delay. Providers: intake, capacity, referral completion, overlap, nonresponse and real cost. Large buyers: supplier requirements, invoice terms and whether new local purchases expand output or shift contracts.

For a later representative survey, first obtain a sampling frame, pre-register strata, sample nonusers, document weighting and response rates, pilot multiple languages, and report nonresponse. Do not turn purposive interview counts into city percentages.