TBDC Cohort — Investor Matching System

PART 1 · POC v2 · AHMED KORAYEM · PARTNERSHIPS MANAGER APPLICATION
10 portfolio companies24 investors profiled7-dimension weighted scoringHard exclusion gatesWIDMO exception flagged
Hard gates \u2014 run before scoring begins

If any gate fires, the investor-startup pair is eliminated from the match output entirely, regardless of what score they would have received. There are three gates.

GateTriggerRationale & Implementation Note
01 \u2014 Founder Opt-OutCompany has explicitly declined investor introductions (WIDMO flag)Non-negotiable. Remove from all VC matching and reroute to customer facilitation. Violating a founder’s stated preference destroys trust with the founder and TBDC’s relationship capital with the investor simultaneously. This gate is set by the company, not the IR function.
02 \u2014 Geographic JurisdictionInvestor does not deploy in Canada at allEliminates investors whose geographic mandate categorically excludes Canada and all Canadian expansion markets. Note: regional preference within Canada (Ontario-focused, BC-focused, national) is NOT a hard gate — it carries into the scoring layer as a ranked dimension (Geographic Alignment, rank 05). Only country-level exclusions are handled here.
03 \u2014 Fund ActivityInvestor is not currently deploying capitalEliminates investors who are inactive, between funds, or fully deployed. This is treated as a gate rather than a scored dimension because it is binary — an investor is either writing cheques or they are not. There is no partial credit for fund cycle timing. A perfectly thesis-matched investor who is not deploying produces zero outcomes regardless of every other dimension.
Implementation note: the scoring algorithm must not run until all three gates have cleared. A company or investor that triggers any gate should never appear in the match output for that pair.
Scoring dimensions \u2014 weighted, not equal

Seven dimensions are scored after the hard gates clear. Dimensions are ranked by importance \u2014 higher-ranked dimensions carry greater weight. Maximum possible score: 14 points.

RankDimensionMax ptsScoring Logic & Rationale
01Stage Fit3 ptsExact stage match = 3 / Adjacent (one stage away) = 1 / Mismatch = 0. Stage determines whether a VC can legally and structurally deploy from their current fund into this company. One degree of adjacency is permitted because some investors operate across two adjacent stages.
02Sector & Thesis Fit3 ptsThree-layer analysis — all three layers inform the final score: (1) Stated thesis: what the investor publicly says they focus on — baseline signal only. (2) Revealed preference: what their portfolio shows they have actually backed. Weighted higher than stated thesis because portfolios are evidence; stated thesis is marketing. (3) Portfolio gap: does this startup fill an identified gap in the investor’s existing portfolio, or does it create unwanted concentration? Full match across all three layers = 3 / Adjacency or partial = 1 / Outside thesis or creates portfolio duplication = 0.
03Cheque Size Fit2 ptsCheque range covers the ask = 2 / Partial coverage = 1 / No coverage = 0. A VC who writes $500K cheques cannot lead a $5M round. A fund with a $5M minimum cheque floor should not receive a $500K raise introduction.
04Lead / Follow Fit2 ptsScored against the startup’s current round structure — not treated as a hard filter or binary disqualifier. Context-dependent logic: if the startup already has a lead investor, a follow-only investor scores 2 (strong match). If there is no lead yet, a follow-only investor scores 0 (wrong intro). Flexible or lead-capable investor with no lead present = 2. Partial or undetermined = 1.
05Geographic Alignment2 ptsRegional fit within Canada — scored, not filtered (country-level exclusion is handled at Gate 02). Primary region match = 2 / Will consider or has cross-regional history = 1 / Outside stated region with no cross-regional precedent = 0.
06Founder–Investor Fit1 ptFounder background maps to the investor’s demonstrated pattern of conviction = 1 / No signal = 0. Operator pedigree, academic credentials, domain expertise, or repeat-founder status that directly correlates with this investor’s historical attention and follow-on behaviour.
07Strategic Value1 ptInvestor can open doors beyond capital — network overlap, domain expertise, relevant portfolio connections, or operational relevance to the startup’s specific growth stage = 1 / Capital only = 0.
ModifierWarm PathActivation modifierDoes not affect score — changes execution protocol only. Warm path available = curated one-paragraph brief + permission ask to the connector. Cold path = gap-framed outreach referencing recent portfolio moves and specific thesis alignment.
Score tiers and action protocol

Score tiers are calibrated against a maximum of 14 points. Each tier maps to a specific action protocol for the IR function. The tier determines not just whether to make the introduction, but how.

11–14
Tier 1 — Priority intro
High-conviction match. Make the warm intro immediately, or craft targeted outreach with a specific portfolio gap angle if no warm path is available. Do not delay.
7–10
Tier 2 — Qualified outreach
Logical match with identified gaps. Worth an introduction if framed correctly — typically positioned as co-investor or follow-on, not lead. Review which dimensions pulled the score down before crafting the intro.
3–6
Tier 3 — Monitor
Premature or partial alignment. Log for reactivation at a future milestone. Do not make the intro now — a weak intro is worse than no intro. Note the specific milestone that would move this to Tier 2.
0–2 / Gate
Do not match
Structural mismatch or hard gate triggered. Making this introduction damages TBDC’s credibility with the investor. Log the reason and move on.
A weak introduction is worse than no introduction. Tier 3 and Do Not Match outcomes should be logged with the reason and the milestone that would change the outcome. This data feeds the learning loop across cohorts.
Design rationale \u2014 why these choices

Why weighted scoring outperforms binary matching

A binary model awards one point per signal — sector, stage, geography, and so on — treating all signals as equal. But a stage mismatch is a near-structural kill; geographic mandate exclusion is a hard structural kill; and sector adjacency is merely a weaker intro. Equating these produces false positives and wastes relationship capital on introductions that cannot convert. The weights encode what actually matters. Stage and Sector are weighted at 3 because they are the two dimensions most predictive of whether a VC has conviction and can deploy. Cheque size and Lead/Follow are at 2 because they are structural constraints that are slightly more flexible. Founder fit and Strategic Value are at 1 because they differentiate between otherwise equal matches but do not determine fundamental eligibility.

Why Fund Activity is a gate, not a scored dimension

Fund activity was absent from the previous version of this methodology. It has been elevated to a hard gate — not a scored dimension — because the question is binary: the investor is either currently deploying capital or they are not. There is no meaningful partial score for being 60% through a fund cycle or exploring a new fund. A match that cannot result in a cheque is not a match.

Why geography does two different jobs

Geographic mandate at the jurisdiction level (an investor who does not deploy in Canada) is a structural kill — it belongs at the gate layer. Regional preference within Canada is softer: an investor who primarily backs Ontario companies may still look at a BC deal if the opportunity is strong. Treating intra-Canada regional preference as a hard filter would over-exclude. Treating it as a scored dimension preserves the signal without making it eliminative.

Why Lead/Follow Fit is scored, not filtered

A follow-only investor is either a strong match or the wrong introduction entirely — determined entirely by whether the startup already has a lead in the current round. Because the correct answer depends on the startup’s situation at the time of matching, not on the investor’s preference in isolation, it cannot be a hard filter. It is scored against the startup’s actual round structure.

Why Revenue / Traction Threshold was removed

Revenue threshold is structurally redundant with Stage Fit. If stage fit is correctly defined — matching investors to the startup’s current raise stage, not future stage — a pre-revenue company will never reach an investor with a hard revenue floor, because that investor operates at a stage that has already been filtered out. Redundant dimensions in a weighted scoring model do not add signal; they double-count existing signal and distort output scores.

Why Portfolio Gap is inside Sector & Thesis, not standalone

Portfolio gap analysis asks: does this startup fill an identified gap in the investor’s existing portfolio, or does it create unwanted concentration? This is not an independent variable — it is the third layer of thesis analysis, because an investor’s receptiveness to a new deal is directly shaped by what they already own. Separating it as a standalone 1-point dimension makes it visible but structurally disconnected from the analysis it belongs to. Folding it into Sector & Thesis Fit as the third layer of a three-layer analysis is more coherent.

Warm path as activation modifier

Warm path availability is logged alongside the match score but does not contribute to it. It changes how the IR function executes a valid match — not whether the match is valid.

Warm path available
Curated one-paragraph brief to the connector. Permission ask before the introduction. Framed around the specific thesis angle the system identified.
No warm path
Gap-framed outreach referencing the investor’s recent portfolio moves and the specific thesis alignment. Direct, specific, short.

A warm path to a Tier 2 match is not preferable to a cold path to a Tier 1 match. The score determines priority. The warm path determines execution.

Scoring summary