By the time a site selector calls your office about a prospect's workforce needs, your labor market has already been run through at least one commercial analytics platform.
JobsEQ, built by Chmura Economics & Analytics, reports occupation employment, wages, commuting patterns, education pipelines, and job-posting activity at ZIP, drive-time, and custom labor-shed geographies. Its postings layer normalizes employer names and shows which companies are actively recruiting for a searched occupation. Lightcast Analyst provides occupation employment, earnings, projected growth, staffing patterns, job postings, compensation, and graduates by institution and program down to MSA, county, ZIP, and drive-time boundaries.
Matt Samler of Vista Site Selection presented his preselection workflow at the 2023 Area Development Consultants Forum. It combines multiple databases, targets specific occupations, quantifies supply and potential growth, and compares existing employers. All of that runs before any refinement or site visit.
That is the baseline your labor market presentation has to survive. If your package lists 4,200 production workers and 380 CNC machinists without mentioning that three incumbent manufacturers are actively posting for the same occupation codes, the selector can check that claim against platform data in minutes.
The credibility discount
No public source from the past three years documents a site selector formally scoring an EDO's omission of competing employers as dishonesty, or stating that it automatically eliminates a community. The evidence supports a narrower claim: selectors hold independent occupation, employer, wage, and posting data, and they treat local assertions as inputs to validate against it.
Chris Schwinden, managing director at Site Selection Group, advises prospects to make communities substantiate expansive workforce claims with actual employer evidence. His prescribed site-visit questions include "Who's the employer of choice in my industry?" and which incumbent employers have unattractive shifts, management practices, or benefits. Those questions are diagnostic of the community, not just the labor market.
Alexandra Segers, general manager of Tochi Advisors, writes in Area Development:
"Regional employment figures and occupational concentration statistics describe what exists — they do not describe what is available."
A region with an established industrial base can have more intense talent competition than a greenfield market, because a new plant must recruit against incumbent employers with deeper workforce ties.
The cost of omission is a discount applied to everything else in your package. If the workforce section is incomplete in ways the selector can verify independently, the infrastructure section and the incentive section inherit the same suspicion.
Incumbent competition as a quality signal
Segers calls an existing foundation of engineering, technical, and skilled-trades talent essential for advanced manufacturing. Ceci Grover, an associate with Site Selection Group, writes that some markets offer large specialized labor pools while others offer less competition and a stronger opportunity for a new employer to establish itself, and that "neither is inherently better." Her firm evaluates posting intensity, the number of companies posting, turnover, pay, and whether the prospect values stability, scalability, immediate experience, or trainability.
A market where three manufacturers already employ CNC machinists is a market where CNC machinists exist, where training programs produce them, and where employers have found the workforce reliable enough to stay and expand. That is a different presentation than "we have 380 machinists available," and it is the one that holds up against the selector's own data.
Incumbent manufacturers in the target sector have already answered questions the selector needs answered: whether the local workforce can perform these operations, at what wage workers can be recruited and retained, and whether the training pipeline produces hires that employers keep. Retention patterns, wage levels, and continued hiring from local programs are testable. The four documentation categories below present that evidence in a form a selector can verify.
Wage position against competing markets
This element has the cleanest public source. The BLS Occupational Employment and Wage Statistics program publishes employment, wages, and location quotients for roughly 530 metropolitan and nonmetropolitan areas. The most recent release at this writing covers May 2025, published May 15, 2026.
Build a comparison table: one row per geography — your market and three to five named competitor markets — with one block per project-critical occupation code. Columns: estimated employment, jobs per 1,000, location quotient, median hourly wage, 25th and 75th percentile wages, and estimate vintage. Hold the release year and the occupation code constant across all geographies.
Two constraints matter. OEWS areas are MSAs or BLS-defined nonmetropolitan areas, not custom drive-time labor sheds. If your labor shed crosses MSA boundaries, or covers a nonmetropolitan area that BLS aggregates differently than your functional market, label the difference rather than letting the selector find it. And location quotient establishes that an occupation is concentrated in your market relative to the national average. It says nothing about whether those workers are available to a new employer. Segers' distinction between what exists and what is available applies directly.
The wage table tells the selector where the prospect's proposed offer sits against the local market and against competing geographies. If the prospect plans to offer $24 an hour for industrial maintenance technicians and your market's median is $26.50, that is information the selector will find independently. Present it first.
Training completions by program and institution
IPEDS Completions data reports awards by institution, six-digit CIP code, and award level. There is no built-in drive-time labor-shed filter. The reproducible method: define your geography, identify institutions whose physical locations fall inside it, retain their IPEDS UnitIDs, and pull the relevant program and award-level records.
The resulting number is awards conferred. It is not a program completion rate and it is not available workers. IPEDS does not show whether award earners remain in the labor shed, enter the corresponding occupation, meet a particular employer's qualification standard, or will be available by the project's ramp date. Label the measure for what it is.
The table should show institution name, campus location, six-digit CIP, program title, award level, academic year, and annual awards. Current enrollment, scheduled cohort seats, employer-linked placements, and first possible completion dates belong in separate columns, sourced from the institution itself. As I covered in Issue #2, the college controls that claim, not you. The pre-authorized data-sharing arrangement needs to exist before the information request arrives, not after the selector calls.
Incumbent turnover and retention
Public data fails here. JOLTS does not collect occupation information, and its regional estimates cover only total nonfarm employment. The CPS Employee Tenure supplement gives national median tenure by broad occupation group and nothing at the MSA or employer level. Census Quarterly Workforce Indicators offer hires, separations, and turnover by industry at county and workforce-area geographies, but not by detailed occupation or named employer.
Occupation-level incumbent turnover requires a local employer survey, confidential interviews with incumbent manufacturers, or staffing-firm data.
The Savannah Harbor-Interstate 16 Joint Development Authority's 2023 workforce study shows what this looks like when it is done well. Interviews with 19 selected employers produced average annual turnover of 35 percent, with turnover of 60 to 94 percent among employers paying below $17 an hour and 5 to 18 percent among those paying above $18. The study attributed the variation to wage level, culture, advancement opportunity, benefits, flexibility, and supervision quality.
Turnover segmented by wage band, with respondent count and reporting period identified, converts a retention claim from assertion into evidence. A forwardable table should identify the occupation or job family, respondent count, reporting period, the separations or turnover definition in use, median tenure, shift, wage band, and whether the statistic is employer-reported or calculated from supplied data.
The Savannah study also documented that projected industrial demand would exceed supply by 2025 under a no-intervention scenario. The region published the constraint alongside a workforce development plan to address it, which is why the rest of the document reads as analysis rather than promotion.
Employer satisfaction with training programs
There is no federal indicator for employer satisfaction with a local industrial training program. Perkins V performance data report student measures — technical-skill attainment, credentials, placement — organized by state and education level, not by named local program or employer rating. WIOA's employer effectiveness measure tracks participant retention with the same employer after exit, which answers a different question.
Individual colleges sometimes run their own employer surveys, and quality varies enormously. Hennepin Technical College's FY2024 employer satisfaction survey drew five responses from 87 contacts, a 5.7 percent response rate, and the college stated plainly that the results could not be extrapolated to its graduate population.
If your community college has conducted a survey with a defensible sample, present it with full transparency: program or CIP evaluated, employers invited, responses received, response rate, question wording, skill ratings, willingness to rehire. If the survey does not exist, or the sample is too thin to carry weight, say so and substitute advisory-committee minutes, accreditation records, or direct testimonials from named employers who will take a phone call. Three named employers willing to talk beats a five-response survey presented as a satisfaction finding.
When the honest answer is no
The public record does not supply a disqualification threshold. Area Development's 2024 workforce analysis, drawing on practitioners Sam Moses, Mark Simmons, and Azad Khan, states that there is not necessarily a "magic labor number" for a manufacturing location. The SelectUSA Investor Guide directs investors to evaluate competition, cost, and availability of required skill sets without publishing an acceptable ratio of project hires to existing occupation employment.
I will not invent a threshold that practitioners have not published. What I can specify are the variables that determine whether your market absorbs a given project, and the comparison structure that makes your assessment credible when you deliver it.
Start with the project's staffing specification: number of hires by occupation, ramp schedule, shift structure, experience requirements, proposed wage range. Then assemble the local side.
Existing occupation base. OEWS employment and location quotient for each project-critical occupation in your labor shed.
Incumbent and announced demand. Named employers with relevant operations, documented expansions, companies currently posting for each occupation, posting intensity, advertised wages. Source dates on all of it.
Annual pipeline output. IPEDS awards by relevant program code, current enrollment, and the first possible completion date for new cohorts.
Replacement demand. QWI industry separations as context, supplemented by locally surveyed occupation-level turnover where you have it.
Plausible additional entrants. Unemployed and nonparticipating workers with relevant experience, in-commuters from adjacent labor sheds, cross-industry transfers, apprenticeship and retraining candidates.
Wage feasibility. Where the prospect's proposed offer sits against OEWS percentiles and against incumbent advertised wages.
Project hires as a share of existing occupation employment is worth calculating and presenting, but it is not a pass/fail number. A project needing 120 industrial electricians in a market with 400 is asking a different question than one needing 120 in a market with 2,000. Either answer depends on incumbent turnover, pipeline output, wage premium, ramp timeline, and whether the existing workers are recruitable at all or locked into employers with strong retention.
When the comparison shows that occupation demand, at the proposed wage and timeline, would require recruiting a share of the incumbent base that exceeds plausible voluntary turnover plus pipeline output plus entrants, deliver that answer. A documented, specific no-bid builds the relationship for the next search.
Directors discount this. A selector running 15 searches a year keeps a short list of markets whose ED offices produce analysis that survives diligence. A specific explanation of why a project does not fit — the arithmetic visible, the sources dated — is more useful to that selector than an optimistic submission that collapses in month three. The next search, with a different occupation mix or a different wage structure, starts with your market already credible. As I wrote in Issue #5, the documented no-bid also builds the internal record that reaches your capital budget and your workforce board, which is a second return on the same work.
Where the evidence shows constraint — high posting intensity, wage pressure, pipeline gaps — that goes in the package too, at the same level of specificity as the favorable numbers, so the selector evaluates the constraint instead of discovering it.
The selector's job is to compare your market against four or five others on the same variables. Give them that comparison on your terms, sourced and dated. The alternative is that they build it from a commercial platform without your context, your employer relationships, or your pipeline commitments.
- Lightcast renamed its measure: The platform changed "Employers Competing" to "Companies Posting" in February 2026 because the field counts companies listing postings, not verified employer headcounts — a distinction that matters when citing the number in a workforce exhibit, as documented in the Analyst Platform Changelog.
- FERC's large-load proceedings affect manufacturing: FERC's June 2026 orders directed all six jurisdictional RTOs and ISOs to justify or reform rules governing the connection of large loads including manufacturing facilities, covering study procedures, cost shifting, and capacity reservation obligations that could change how utilities respond to industrial workforce-adjacent power questions.
- Apprenticeship data now reaches mid-2026: The DOL's Apprentices by State Dashboard covers fiscal years through June 2026 with monthly refreshes, offering county-level apprenticeship counts that can supplement IPEDS completions for occupation-specific pipeline evidence.
- Savannah's study preceded major commitments: The Savannah JDA's 2023 workforce study documented that projected industrial demand would exceed supply under a no-intervention scenario, but no public record establishes that the study's candor about constraints caused a specific site-selection win — a gap worth watching as the region's megasite projects advance.