SB StartupBasket
All ideas
72 /100 GO Low complexity

PrevailGuard — LMIA wage-drift watch for TFW employers

Catches the January 1 wage review Canadian TFW employers forget, and keeps the six-year inspection file ready.

Compliance SaaS Global Compliance-drivenSMBAI-agentSolo-builderCanada
— views
Evaluation Scores
72/100

GO

Overall Score

15
Problem
10
Demand
13
Build
13
Distrib.
11
Revenue
7
Time
3
Defense

PrevailGuard

1. One-liner

Catches the January 1 wage review Canadian TFW employers forget, and keeps the six-year inspection file ready.

2. Trend signal — why now?

Three things moved in the last fourteen months, and they compound.

Enforcement roughly doubled in one year. ESDC completed 1,488 compliance inspections in the year ending 31 March 2026, found 12% of employers non-compliant, banned 30 outright, and issued $10.2 million in penalties — up from $4.5 million the prior year. The government put out a press release specifically to advertise that fact in July 2026. When a regulator brags about doubling its fines, it is telling you the inspection rate is going up, not down.

The inspection is a paperwork exercise, not a site visit. Per an Investigative Journalism Foundation analysis, 77% of the 12,000+ inspections conducted since 2020 were strictly paper-based — documents requested remotely, no inspector on site. That matters enormously for product shape: the thing being tested is whether you can produce a coherent document set on demand, not whether your workplace looks good.

There is a silent, dated duty that nobody is reminded about. Since 1 January 2024, every employer holding an LMIA-based work permit must conduct an annual wage review and raise the worker’s pay to match the updated Job Bank median for that NOC and region. Job Bank refreshes wages each fall; employers have until January 1 to comply. No source I could find indicates ESDC notifies employers when their occupation’s median moves. The wage was legal in December and illegal in January, and nothing announced the change.

Layer on the 17 July 2026 increase to provincial high-wage/low-wage thresholds (Ontario $36.92/hr, BC $38.40/hr, Alberta $37.50/hr) and you have a population that is drifting out of compliance while sitting still.

Provenance:

3. The opportunity

The market has solved getting the LMIA. LMIA.io productized it at $499 a pop and explicitly markets itself against “expensive consultants.” CaseEasy, Immicase and a dozen RCIC practices sell application workflow. Envoy Global and MOVEPLUS sell visa-expiry tracking to multinationals.

Nobody sells surviving the six years afterward.

That is the whole gap. The LMIA is approved on day one. The compliance obligations then run for six years: keep every payroll record, timesheet, contract, recruitment artefact and housing document; keep paying at or above a median wage that changes annually without warning; stay under the 10% low-wage cap measured on hours worked per worksite — a ratio that shifts every pay period as Canadian staff turn over; and produce all of it inside 30 days when a Notice of Preliminary Finding lands.

The incumbent for defence is an immigration lawyer at $145–$290/hour who advertises a “mock audit” with no published price, quoted only on consultation. That opacity is the tell: this is bespoke retainer work because nobody has systematised it.

The 10× is not clever AI. It is that a $99/month background process can watch a wage table you will otherwise never check, and keep a folder assembled that you will otherwise assemble in a panic. A lawyer cannot profitably watch your Job Bank median every fall for $99. That is the structural refusal.

4. Target market

Primary customer: Owner or office manager at a Canadian SMB with 5–150 employees holding at least one positive LMIA — concentrated in long-haul trucking, restaurants and food service, food processing, and agriculture. These are firms with a bookkeeper, not an HR department, and certainly not an immigration counsel on retainer.

Why they buy: Not because they love compliance. Because AMPs run to $100,000 per violation and $1M per year, because failure to produce documents now draws up to $45,000 plus a five-year ban, and because a ban means losing the workers the business currently runs on. CFIB’s own survey work found that under TFWP restrictions about 1 in 5 businesses using the program said they would very likely close, 24% would cut hours, and 52% could not fill orders. The workers are load-bearing; losing program access is existential.

Rough TAM reasoning: Canada approved over 228,000 LMIA positions in 2025. Positions cluster heavily in agriculture and food service, and employers appear repeatedly across quarters. A conservative read is 15,000–30,000 distinct employers holding live LMIA obligations at any time, skewed small. I only need a few hundred.

Why now for them: The next wage review deadline is 1 January 2027, and Job Bank’s updated medians land this fall. There is a four-month selling window opening right now, with a hard date at the end of it. Meanwhile penalties doubled and the government publicised it.

5. Product sketch (MVP)

  • Wage drift watch. Enter each TFW’s NOC, work location and LMIA-stated wage once. Every fall when Job Bank refreshes, the system diffs the new median against what you actually pay and tells you the exact dollar increase required by January 1 — or confirms in writing that no change is needed.
  • Compliance-condition file. A structured folder per worker holding the LMIA decision letter, offer of employment, recruitment/advertising proof, contracts and housing documentation, with a completeness meter showing which of the required conditions has no evidence behind it.
  • Payroll evidence capture. Monthly pull or upload of pay stubs and timesheets, timestamped and retained on the six-year clock so the record survives a payroll-provider switch — a documented failure point where employers lose historical records mid-window.
  • Low-wage cap tracker. Running calculation of low-wage TFW share of total hours worked per worksite against the applicable 10% / 20% ceiling, flagging when Canadian-staff turnover pushes the ratio toward breach.
  • Inspection response pack. When a document request or Notice of Preliminary Finding arrives, generate a single indexed PDF bundle mapped condition-by-condition to what ESDC asked for, with gaps flagged and a draft justification narrative for any variance.
  • Variance log. Timestamped record of every wage change, duty change, and location change with the reason attached — because a justified variance defended contemporaneously beats one reconstructed two years later.
  • Retention calendar. Per-worker six-year expiry dates so nothing is deleted early and nothing is kept forever.

6. AI angle — what’s load-bearing

Two jobs, both real.

Document classification and extraction. Employers arrive with a shoebox: PDF pay stubs from three different payroll providers, photographed timesheets, emailed offer letters, screenshots of Job Bank postings. The system reads them, identifies what each one is, extracts the worker, pay period, hours and gross pay, and files it against the right compliance condition. That is the tedious middle of every mock audit, and it is exactly what vision-capable models now do cheaply.

Variance justification drafting. When the record shows the paid wage diverged from the LMIA wage, ESDC’s own framework allows an employer to escape penalty entirely with an acceptable justification. Drafting that narrative — citing the specific regulation, the good-faith-error provision, and the evidence in the file — is the highest-value document in the whole process, and it is currently a lawyer’s billable hour.

Strip the AI out and this becomes a spreadsheet with reminders. It would still have some value — the wage diff is arithmetic once the data is clean — but the data is never clean, and getting it clean is the actual work.

7. Localization angle (if any)

Canada-specific by construction, and that is the point rather than a limitation. The product encodes Job Bank NOC-2021 median wage tables, provincial high/low-wage thresholds, the IRPR 209.3 condition set, and Quebec’s separate MIFI wage guide — Quebec being a genuine fork, since its wage floors come from a different authority entirely.

The natural second market is the US H-2A/H-2B and H-1B Labor Condition Application world, which has a structurally identical prevailing-wage-plus-public-access-file duty. Same product shape, different tables. But that is a year-two conversation, and the US market has established incumbents; Canada does not.

8. Business model — path to $1M–$5M ARR

  • Pricing: $89/month base for up to 3 TFWs, then $25/month per additional worker. Annual plan at 10 months’ price. One-time $1,500 Inspection Response Pack when a document request or NPF actually arrives — priced well under a lawyer’s retainer and sold at the moment of maximum fear.
  • ACV: A typical trucking or restaurant customer with 6–10 TFWs lands at roughly $215–$265/month, so ~$2,700 ACV. Larger food processors with 30+ workers exceed $700/month.
  • Rough math to $1M ARR: 370 customers at $2,700 ACV. Against a plausible 15,000–30,000 employer base, that is 1.2–2.5% penetration.
  • Rough math to $5M ARR: ~1,600 customers at a slightly higher blended ACV, plus meaningful revenue from response packs, plus the agency channel — RCIC and immigration-law practices reselling monitoring to their own employer books under their brand. Realistically this needs the US LCA expansion to be comfortable.
  • Expansion path: ACV grows with headcount automatically. Response packs are episodic but high-margin. The genuine expansion is the agency tier: a consultancy managing 40 employer clients is a single sale worth 40 logos.

9. Go-to-market wedge — first 100 customers

The customer list is published by the government, free, quarterly.

  • Mine the open-data LMIA list. ESDC’s Positive LMIA Employers List gives employer name, NOC and business location, updated quarterly through 2026Q1. Filter to trucking, food service and food processing employers outside primary agriculture, enrich with a business-data provider for contact details, and you have several thousand named prospects who provably hold LMIA obligations right now.
  • Send the wage diff, not a pitch. For each prospect, compute the actual Job Bank median for their NOC and region against the low-wage threshold, and email one line: “Your NOC 73300 drivers in Peel Region: the median moved to $X. If you’re paying below that on January 1, that’s an IRPR 209.3 wage violation.” This is a free, specific, verifiable diagnostic — the strongest cold email available, and it costs nothing to compute at scale. Target the 1 January 2027 deadline with a September–December sequence.
  • Sell through RCICs and immigration boutiques. Every firm advertising “mock audits” with no published price has a book of employer clients and no software. Offer white-label monitoring at 30% revenue share. Ten firms with 20 employer clients each is 200 potential logos, and the firm does the trust-building.
  • Own the non-compliant employer list as content. ESDC and IRCC publish the names of penalised employers. A quarterly teardown — what they were fined for, which condition failed, what the record should have shown — is the highest-intent SEO asset in this niche, and it is factual public record rather than fear-mongering.
  • Provincial trucking and restaurant associations. Canadian Trucking Alliance provincial members and restaurant associations run compliance webinars for exactly this membership. A free “wage review clinic” in October and November lands directly in front of the deadline.

The first 100 customers come from bullets one and two alone. The list is named, the diagnostic is computable, and the deadline is fixed.

10. Build complexity — justification

Low. Job Bank wage data and the LMIA employer list are public. Document ingestion is off-the-shelf vision-model extraction. The compliance rules are a modest decision table — a few dozen conditions, not a rules engine. There is no government API to integrate with, which is a build simplification even though it is a product limitation: the system is a system of record the employer maintains, not a filing pipeline.

The genuine work is domain encoding: getting the NOC-2021 mapping, provincial thresholds, Quebec’s MIFI fork and the cap arithmetic exactly right. That is research time, not engineering risk. A technical founder with an RCIC advisor ships v1 in 8–10 weeks.

11. Gating checklist

GatePass?Note
Legal in target market✅Record-keeping and monitoring software. Must avoid giving immigration advice, which is regulated — CICC licensing governs who may advise. Product outputs evidence and arithmetic; an RCIC partner handles advice.
Ethical — no harm / dark patterns✅Pushes employers toward paying workers the wage they legally owe. Aligned with worker interest, not against it.
Market exists (evidence above)✅228K+ positions approved 2025; LMIA.io already sells software to this exact buyer at $499–$4,999.
1–5 person team can build this✅Solo technical founder plus a part-time RCIC advisor.
Launchable with <$50K / ₹40L✅Well under. Data is free; main cost is the advisor and inference.

All five pass.

12. Feasibility score

AxisWeightScoreNotes
Problem intensity2015/20Severe consequence — $100K/violation, bans that can close the business, and 1 in 5 CFIB respondents said restrictions would likely close them. Docked because the pain is episodic and invisible: 88% of inspected employers pass, and most employers will never be inspected in a given year. It is not a daily bleed, and per my own pattern notes that shape converts worse than a weekly wound.
Demand evidence1510/15Hard economic and regulatory signals are strong and government-sourced. But a dedicated search found zero verbatim employer complaints about this duty across Reddit, trade forums and CFIB. The buyers are not publicly asking for this — plausibly because complaining about TFWP compliance in 2026 Canada is politically radioactive. Willingness-to-pay for adjacent software is proven by LMIA.io; willingness to pay for this is inferred, not observed. That gap is why this is a 10, not a 13.
Build feasibility1513/15Public data, off-the-shelf extraction, small rules table, no integrations required. 8–10 weeks.
Distribution clarity1513/15Government publishes the prospect list quarterly with occupation and location. The cold email contains a free computed diagnostic specific to the recipient. Hard to do better than that. Docked slightly because email deliverability to small trucking and restaurant firms is genuinely mediocre.
Revenue mechanics1511/15$2,700 ACV and 370 customers to $1M is credible against the base. The $1,500 response pack is high-margin and well-timed. Uncertainty is retention: an employer who passes an inspection may decide they no longer need the watch.
Time to first revenue107/10The January 1 deadline creates urgency, but this is an annual duty, so a prospect who has already done their review has no reason to buy until next fall. Expect 6–8 weeks to first paying customer if launched into the autumn window.
Defensibility103/10Honestly weak. The wage tables are public, the rules are public, and LMIA.io could add this as a module in a quarter — they already have the employer relationships and the median-wage data at application time. The only real moat is the accumulating six-year evidence archive, which gets stickier every year but is worth nothing in month three.
Total10072/100

13. Qualitative modifiers

Founder-fit tags

technical-heavy · domain-expertise-required

This needs someone who will read IRPR 209.3 and the ESDC program requirements carefully and get the wage tables exactly right. A wrong wage number is worse than no product. Pair with an RCIC — both for domain accuracy and because immigration advice is a licensed activity in Canada.

Key assumptions to validate (3–5)

  1. Assumption: SMB employers do not reliably perform the annual wage review, and a meaningful share are currently underpaying relative to the updated median. How to test: Take 50 employers from the open-data LMIA list, compute their NOC/region median against the low-wage threshold, and cold-call 20 asking a single question — “when did you last check your Job Bank median?” If most answer confidently and correctly, the core wedge is gone.
  2. Assumption: Employers will pay a recurring subscription for a risk that is episodic rather than continuous. How to test: Offer the January 1 wage review as a paid one-time $199 service to 30 prospects before building anything. If they will not pay $199 for the specific deliverable, they will not pay $89/month for the watch.
  3. Assumption: The absence of public employer complaints reflects political silence, not absence of pain. How to test: Twenty confidential phone interviews with LMIA-holding employers, promising anonymity explicitly. If they are relaxed rather than quietly anxious, the demand thesis is wrong.
  4. Assumption: RCIC firms will resell rather than view this as competitive with their audit-prep billings. How to test: Pitch five firms on the 30% white-label share and count how many engage versus how many say it cannibalises them.

Risk flags

  1. Incumbent adjacency: LMIA.io already owns this buyer, already holds median-wage logic, and sells at $499–$4,999. Post-approval monitoring is a natural product extension for them. This is the single biggest risk and the reason defensibility scores 3.
  2. Political/program risk: TFWP is under sustained public attack and active restriction — low-wage approvals dropped 40% year-over-year. If Ottawa contracts the program further, the customer base shrinks underneath the business. Conversely, restriction increases per-employer compliance intensity, so the effect is genuinely ambiguous.
  3. Regulated-advice boundary: Immigration advice is a licensed activity under CICC. The product must stay on the evidence-and-arithmetic side of the line, which constrains how prescriptive the justification drafting can be without an RCIC in the loop.
  4. Episodic-pain conversion: This has the shape my own pattern notes flag as weak-converting — a dated regulatory duty with no accompanying platform penalty already costing money. The wage-increase obligation partly mitigates this (it costs real dollars on a known date), but the inspection risk itself is a lottery.

14. Structured verdict

Score:                  72/100
Verdict:                GO
Confidence:             Medium
Best-fit builder:       Technical solo founder paired with a licensed RCIC advisor;
                        comfortable with cold outbound to non-technical SMB owners
Time to revenue:        6–8 weeks if launched into the Sept–Dec 2026 window
Capital to launch:      CAD $8–15K (advisor retainer, data enrichment, inference)
Top 3 assumptions to validate first:
  1. Employers are actually missing the annual wage review — cold-call 20 from the
     open-data list and ask when they last checked their Job Bank median
  2. Willingness to pay is real — pre-sell the January 1 review as a $199 one-time
     deliverable to 30 prospects before writing code
  3. Employer anxiety exists but is politically silent — 20 confidential interviews
     under explicit anonymity
Kill criteria:
  - Abandon if fewer than 3 of 30 prospects will pay $199 for the one-time wage review
  - Abandon if LMIA.io ships post-approval compliance monitoring before your v1
  - Abandon if >60% of cold-called employers correctly state their current Job Bank
    median unprompted — it means the duty is already being handled

15. Next step — 1-week validation sprint

  • Day 1–2: Download the 2026Q1 Positive LMIA Employers List. Filter to trucking, food service and food processing employers with 3+ approved positions. Build the Job Bank median lookup for their NOC/region combinations and compute, for each, what the January 1 2027 wage floor will be. This is the asset and the sales weapon simultaneously.
  • Day 3–4: Send 100 personalised emails containing only that employer’s computed number and the deadline. No product, no demo, no landing page — just the diagnostic and a question: “Want us to run this properly for your full roster for $199?” Call 20 of them.
  • Day 5: Count paid commitments and measure the interview signal.

Falsifiable outcome: ≥3 prepaid $199 commitments out of 100 emails and 20 calls, and ≥8 of 20 interviewed employers unable to state their current median wage. Below either threshold, the duty is already being handled or the fear is not purchasable — and I stop.

Interested in a detailed proposal?

Get a deep-dive with market research, competitive analysis, and implementation roadmap.

Contact us

info@startupbasket.ai