When to run this analysis
This guide is for capital-stakes decisions — multi-year, multi-million-dollar commitments where being wrong is expensive. It is not for "should we let one person work from Toronto?" That's a remote-work policy question, not a market mapping exercise. The five most common triggers:
- Standing up a new design center, R&D site, or fab adjacency. Leadership is debating where to plant the flag. You need to compare 2–4 candidate markets with rigor.
- Considering a GCC or global capability center expansion. Especially in India, Eastern Europe, Latin America, or Southeast Asia. The default cities are often saturated; the real question is which second-tier city is the right one.
- Evaluating a satellite team in a new geo for a niche skill. Often AI/ML or specialized hardware. The "micro-hub" thesis — Toronto for visa-friendly AI, Pittsburgh for CMU outflow, Eindhoven for photonics — is now mainstream strategy.
- Build-vs-buy decision. Open a new office, hire 50 remote, or acquire a small company to inherit a market. Each has very different talent dynamics that need to be on the table.
- Annual location strategy refresh. The 2026 location landscape is shifting fast — visa rules, AI sovereignty, cluster saturation. Strategies older than 12–18 months are often quietly out of date.
What you need before starting
This guide takes 3–6 weeks for a serious comparison of 2–3 markets. Start without these inputs and the analysis stalls in week two.
One sentence on what's being decided — new design center, GCC, satellite team, acquisition target evaluation. Owned by the business leader, not the TI team.
2–4 candidate markets, not just one. Single-market analyses underdeliver because leadership inevitably asks "compared to what?" Pre-empt it.
3–8 specific roles you'd hire into this market. Not a function, not a job family — specific role definitions tied to the work that needs to happen there.
Operational by when? 6 months, 18 months, 3 years? Determines whether the answer is greenfield site, acqui-hire, or remote-first.
TalentNeuron, Lightcast, Draup, LinkedIn Talent Insights — plus government data: BLS OES, Eurostat, OECD Stats, national stat offices.
Comp/total rewards, mobility/immigration, real estate, finance, legal/compliance. Pre-engage them — you'll need their input by Step 4.
The seven steps
The order is deliberate. Steps 1–2 set the question. Step 3 quantifies the pool. Step 4 (the Five-Lens read) is the analytical heart. Steps 5–6 add depth. Step 7 compresses everything into a decision document. The discipline to follow the order — instead of jumping to data pulls — is what separates a credible market brief from a long-form report no one acts on.
Frame the decision and the success criteria
A market mapping exercise without a decision frame becomes a survey. Before you pull a single number, get explicit alignment on what's being decided and what "good" looks like.
- Entry mode under evaluation — greenfield site, satellite team, GCC build, acqui-hire, fully distributed remote, or "decide which of these."
- Comparison set — single-market evaluation or comparison of 2–4 candidates. Leadership almost always wants comparison; default to it.
- Roles in scope and target headcount per role over 3 years.
- Success criteria — what does a "yes" market look like? Pool size? Cost target? Time-to-stand-up? Risk threshold?
- Decision-maker and decision date. If both are unclear, the analysis lacks an audience and will drift.
Semi/AI example "Should we add an AI inference engineering team?" is not a decision frame. "Should we stand up a 40-person AI inference engineering team by Q4 2027, comparing Toronto, Krakow, and Bangalore, optimizing for visa-resilience and 24-month time-to-productivity?" is.
"Toronto" is not a market boundary. Greater Toronto Area, Toronto-Waterloo Corridor, and Ontario are three different markets that produce three different talent pool numbers. The boundary call has to come up front.
- The geographic boundary — metro, region, country, or cross-border cluster (e.g., the Toronto-Waterloo corridor spans two metros but functions as one talent market).
- The commute or relocation logic — what's the realistic commute radius for an on-site role? What's a reasonable national relo capture rate?
- The data source's matching boundary — BLS MSA, LinkedIn TI region, Eurostat NUTS-2 — pick one anchor and footnote the others.
Semi/AI example Toronto looks like one market. For AI engineering it's functionally three: downtown Toronto (Vector Institute, Cohere, Layer 6), Waterloo (Waterloo CS, UWaterloo coop pipeline, Tesla / Apple satellite teams), and Mississauga / 905 belt (Nvidia AI office, biotech AI). Treating it as one number overstates accessibility and understates concentration.
Now apply the pool sizing module from the playbook for each role in each candidate market. Output is a comparable matrix — markets across the top, roles down the side, with both total and accessible pool numbers in each cell.
- Total pool — raw count from chosen data source.
- Accessible pool — after employer concentration and stickiness adjustments.
- Movable pool (12-month) — applying realistic attrition and counter-offer logic.
- Annual graduate inflow — universities feeding the role, recent-graduate volume.
- Per-source data confidence — note where the data is thin or estimated.
Semi/AI example A 40-person ML inference team needs senior CUDA, mid-level CUDA, MLIR compiler, model deployment, and inference SRE roles. Sizing each across Toronto / Krakow / Bangalore gives you a 5x3 matrix that immediately shows which markets clear the bar for which roles — and which roles will need international relo capture regardless of market choice.
Run the Five-Lens evaluation for each candidate market
This is the analytical heart of the guide. The Five-Lens framework moves beyond cost arbitrage into the dimensions that actually determine whether a market is viable in 2026 — policy friction, economic gravity, capability density, sovereignty constraints, and execution reality. Score each market against each lens, ideally on a comparable 1–5 scale.
Policy Friction
Regulatory and administrative barriers to talent mobility, work authorization, and cross-border operations.
- Visa processing times and denial rates
- Work authorization volatility (recent rule changes)
- Non-compete enforceability
- Tax and employer-of-record complexity
- Student visa flows (for graduate pipeline)
Economic Gravity
Investment flows, sector growth trajectory, and wage pressure dynamics that determine medium-term cost and stability.
- Inbound investment and sector growth
- Wage growth trend (12–36 month)
- Currency volatility (for cross-border)
- Single-market dependency exposure
- R&D spend trajectory
Capability Density
Depth, breadth, and maturity of the talent ecosystem for the specific roles and skill families you need.
- Talent availability for target roles
- Ecosystem maturity (universities, startups, vendor density)
- Leadership bench depth (senior IC and management)
- Skills adjacency and reskilling potential
- Retention and attrition patterns
Sovereignty Constraints
Data residency, AI model localization, and export-control regimes that determine whether your work can legally run from this market.
- Data residency requirements
- AI sovereignty / model localization rules
- Export control exposure (EAR / ITAR / EU dual-use)
- Government access laws
- Critical-tech investment screening
Execution Reality
Time-to-stand-up, vendor and infrastructure ecosystem, and operational risk profile — the difference between a market that looks good on paper and one you can actually operate in.
- Time-to-stand-up (entity, payroll, real estate)
- Local vendor ecosystem (recruiters, PEOs, IT, legal)
- Business continuity risk (climate, political, infra)
- Local management talent availability
- Cultural fit and time-zone overlap
Semi/AI example Toronto scores high on Policy Friction (Canada study-work-permit pathway), strong on Capability Density (UofT, Waterloo, Vector Institute), moderate on Sovereignty (some US export-control complications for advanced ML work), and high on Execution Reality. Krakow scores high on Capability Density and Execution but moderate on Sovereignty Constraints (EU AI Act compliance overhead). Bangalore scores extreme on Capability Density and Execution but lower on Sovereignty for sensitive AI workloads. None of these markets is universally "better" — the scores expose the trade-off the decision actually needs to make.
Map the cluster ecosystem in depth
The Five-Lens read tells you the macro shape. The ecosystem map tells you who actually lives and works there. This is what determines whether you can hire fast or slowly, and from where.
- Anchor employers — top 10 employers of your target roles, with rough headcount share. Identifies both where talent comes from and competitive density.
- University feeders — top 5–8 universities producing recent graduates in the relevant fields. Annual graduating class size where available.
- R&D centers and labs — corporate R&D, research institutes (Vector, MILA, Max Planck, IIIT, Skoltech, Weizmann, etc.), national labs.
- Alumni networks — where do people go after leaving the anchors? Reveals exit patterns and adjacent pools.
- Government incentives — tax credits, R&D rebates, employment subsidies, accelerator programs.
- Infrastructure — international airport hub status, English-language depth, time-zone overlap with HQ, internet/data infrastructure for AI workloads.
Semi/AI example Mapping the Eindhoven photonics ecosystem reveals ASML as the dominant anchor (~25K employees), TU/e as the primary feeder, the Brainport Eindhoven cluster as an incentive layer, and Philips alumni as a deep adjacent pool. That ecosystem can support a satellite team — but you're not "competing in Eindhoven" so much as "operating downstream of ASML's gravitational pull." That's a very different strategic posture than competing in Bay Area where no one company dominates.
A market with a deep pool and weak ecosystem can still be miserable to hire in. This step asks: what's the realistic friction profile of entering, and how does that compare across candidates?
- Competitor saturation — how many other employers are actively hiring the same roles? How aggressive are they?
- Local comp pressure — wage trajectory over 12–24 months. Is the market overheating?
- Attrition norms — what's "normal" tenure in this market? India GCCs run shorter tenures than Eastern European hubs.
- Employer brand bar to entry — do you need to spend on brand-building before recruiting? How long does that take?
- Movability windows — are there cyclical or event-driven openings (post-bonus, equity cliffs, recent layoffs nearby)?
- Sourcing channel maturity — does the recruiting infrastructure exist (good agencies, strong referral norms, active dev communities) or are you building from scratch?
Semi/AI example Bangalore's AI talent pool is enormous but attrition runs 20–25% annually in hot GCCs and comp pressure is intense. The "pool" exists — the question is whether you can hold the people you hire. A market brief that lists Bangalore's headcount without flagging this is operationally misleading.
Build the go / no-go / conditional recommendation
The deliverable is a one-page market brief per candidate market, plus a comparison view if you've evaluated multiple. It ends with a named recommendation, not a "here are the trade-offs" summary that punts the decision back to the asker.
- Go — proceed with entry. Specify entry mode (greenfield site, satellite, acqui-hire), recommended initial headcount, and 18-month milestones.
- No-go — do not enter. Specify the single most important reason and what would have to change for you to revisit.
- Conditional — proceed only if specific conditions are met. List the conditions concretely (e.g., "if non-compete reform passes in MA," "if visa processing time drops below 4 months," "if AI export-control rule X is clarified").
A market brief that lands with "Toronto: Conditional — proceed with a 25-person Phase 1 in Q3 if (a) we secure a Vector Institute affiliate partnership by month 6, and (b) US export-control guidance for our specific model architecture is confirmed compliant" is a brief leadership can act on. A brief that ends with "Toronto has many advantages and some risks" is a brief that gets passed to a consultant.
"Location strategy is no longer an optimization problem — it is a portfolio design problem. The highest-performing organizations over the next five years will deliberately balance a small number of scale hubs for depth and efficiency, a wider layer of resilience hubs to hedge geopolitical, regulatory, and mobility risk, and a growing network of specialist micro-hubs aligned to scarce skills, universities, and regulated workloads."
The New Geography of Work, Draup, January 2026What good looks like
A finished market mapping is not a 60-slide deck — it is a one-page market brief per candidate, backed by a structured working file. Six markers separate one that gets acted on from one that gets shelved.
Comparison view, not single-market
Even if leadership only asked about one market, deliver a comparison with 2–3 named alternatives. They will ask. Have the answer ready.
Five-Lens scoring is visible
Each candidate market is scored on each lens with a one-line rationale. The strategic trade-off is obvious at a glance.
Pool sizing is per-role, not aggregate
Total "engineering talent" is not a useful number. The brief shows pool depth for each of the 3–8 roles you'd actually hire.
Named recommendation
Go / no-go / conditional. Not a list of trade-offs. The recommendation can be wrong; the absence of one is worse.
Risk register with mitigations
For each top risk, name the mitigation and the trigger condition that activates it. Risks without mitigations are anxieties, not analysis.
Dated, owned, scheduled for refresh
Location data shifts on a 12-month cycle in 2026. The brief commits to a refresh date and an owner who'll do it.
Common mistakes
- Single-market evaluation when leadership wants comparison. A brief that says "Toronto looks great" without showing how it compares to Krakow and Bangalore is a brief that gets sent back. Build comparison from the start, even when only one market was named.
- Skipping the Sovereignty lens because "we're domestic." US-only is not exempt from sovereignty considerations — state-level non-compete enforcement, ITAR/EAR for semi work, and federal-contractor compliance all live in this lens. Don't skip it.
- Quantifying the pool, skipping movability. The pool exists ≠ you can hire from it. A market brief without the movability read (links to Step 06 deep dive) is half a brief.
- Locking in on a familiar city too early. Leadership's "let's just go to Austin again" deserves real comparison. If Austin still wins after rigorous comparison, that's a credible answer. If you skip the comparison, "we went to Austin again" stops being a strategy.
- Treating cost as a primary lens. Cost shows up inside Economic Gravity, but it's not a lens on its own — and treating it as one produces the kind of brief that leads to under-resourced sites in expensive markets.
- Building a deck instead of a brief. A 60-slide deck is a documentation exercise. A 1-page brief per market is a decision tool. Discipline the output.
- Ignoring the time-to-stand-up reality. A market that scores 5/5 on Capability Density but takes 12 months to legally operate in is not the same market as one that takes 3 months. Execution Reality is a real lens, not a footnote.
- No refresh commitment. 2026 location data shifts fast — visa rules, AI sovereignty, cluster saturation. A brief without a refresh date is misleading by year-end.
Starter market brief template
Copy this structure into a Google Doc, Notion page, or Confluence template. One market per page, with a comparison page if you've evaluated multiple. Discipline the compression — if it spills, ask what gets cut.