Platform teams usually inherit a list sorted by growth, because growth is the number everyone has. This page replaces that sort with one whose inputs, weights and exclusions are all written down, then runs eight Illustrative companies through it.
What a fastest-growing list measures, and what it leaves out
A growth ranking answers a different question. Deloitte's 2026 North America Technology Fast 500, for example, measures revenue growth from fiscal 2022 to fiscal 2025, needs at least $50K of base-year revenue and $5 million in the current year, requires growth of 50% or more and a founding date on or before December 31, 2021, and excludes consulting companies, professional service firms, resellers, subsidiaries and companies that have been acquired.
For a referral pipeline:
| Feature of the growth list | What it does to a referral ranking |
|---|---|
| Growth is a percentage from a small floor | A company going from $50K to $5 million shows 9,900% growth; a 20-year-old firm adding 8% a year never appears |
| No headcount, record span or rights test | The two facts SourceX screens first, size at peak and years of records, are absent |
| Whole categories are excluded by rule | Professional service firms, subsidiaries and acquired companies drop out, yet SourceX's investor overview lists professional services among its focus areas |
None of this is a criticism of growth lists; they measure what they say. The point is that "fastest-growing" and "most suitable to introduce" are different objects, and funding stage or growth rate never establishes that a company can license its records. Which fit-passing companies a leaderboard's entry rules can never list is mapped in the business history missing from startup leaderboards; this page builds the ranking that replaces it. Market-wide demand analysis sits on sourcex.si, and the AI data market growth page summarizes it.
The transparent ranking method, step by step
| Step | Rule | Why |
|---|---|---|
| 1. Universe | Every company with a named relationship owner, listed with the date the list was drawn | Stops the list from quietly becoming your favorites |
| 2. Gates | Pass, fail or unknown on: 50+ full-time employees at peak, contractors excluded; location; documented operating history; rights to license records the company created; an authorized CEO path | A failed gate cannot be outweighed by scores; failed rows move to a revisit list with the condition that would change them |
| 3. Criteria | Five, each scored 0 to 3 from metadata: historical depth, work detail and outcomes, volume, data structure, path readiness | Four map to fit signals SourceX publishes; path readiness concerns the conversation, not the data |
| 4. Weights | Fixed and dated before scoring: depth 30, detail and outcomes 25, volume 15, structure 10, path 20 | Weights chosen after seeing scores are a route to a preferred answer |
| 5. Evidence | Each cell tagged document, CEO statement or partner estimate; an estimate scores at most 1; unknown scores 0 | Rewards facts, not optimism |
| 6. Ties | A dated record event first, then all gates confirmed, then deeper history | Integer scores tie often |
| 7. Sensitivity | Rerun with three other weight sets; report each company's best and worst rank | A rank that moves three places is not a finding |
| 8. Exclusions | Growth rate, funding stage or amount, valuation, brand, any projected payout, closeness of the relationship | None is a published fit criterion |
| 9. Language | Outside the team, describe fit facts, never a rank | A rank is a planning artifact, not a claim about a company |
Location uses the program's own wording. Companies in the U.S., Canada and the EU qualify, as do companies in the UK, Australia, New Zealand, India, Singapore, South Africa, Latin America, the UAE and Saudi Arabia when at least 70% of the data is in English.
The criteria come from SourceX's ten published fit signals: volume, historical depth, uniqueness, how detailed the work is, domain expertise, whether records show how work ended, human feedback, data structure, clear ownership and Acquisition Partner demand. Five can be described from metadata, so the method scores depth, volume and structure, folds detail and outcomes into one criterion and treats ownership as a gate. The other four need records or SourceX's view of its current programs, so they stay with SourceX's review.
Step 7 follows the OECD and Joint Research Centre handbook on composite indicators, which recommends testing how weights, missing data and the aggregation method change the results, and attaching uncertainty bounds to ranks instead of presenting one ordering as fact.
Eight candidates through the gates and four weight sets (Illustrative)
Illustrative. Generic companies, invented facts, no outcome implied.
| Company | Peak full-time headcount | Records | Rights | Result |
|---|---|---|---|---|
| A. Software firm, headcount up 150% in two years | 140 | 3 years in the current CRM | Unknown: much product data sits in customer tenants | Ranked, flagged |
| B. Industrial distributor, down from 210 to 105 | 210 | 22 years in one ERP | Company-created | Ranked |
| C. Contact center outsourcer | 600 | 11 years | Fails: interactions belong to its clients | Revisit list |
| D. Design studio growing fast | 45 | 6 years | Company-created | Revisit list: below 50 at peak |
| E. Property manager | 260 | 15 years of work orders | Company-created | Ranked |
| F. Engineering consultancy | 180 | 9 years | Unknown: some project files are client-owned | Ranked, flagged |
| G. Logistics firm with a dated system switch-off | 320 | 8 years | Company-created | Ranked |
| H. Claims administrator | 90 | 14 years | Company-created | Ranked |
Scores for the six that pass or remain unknown, with the base weights:
| Company | Depth | Detail and outcomes | Volume | Structure | Path | Base score | Rank range |
|---|---|---|---|---|---|---|---|
| B | 3 | 2 | 3 | 3 | 2 | 85.0 | 1 to 1 |
| E | 3 | 3 | 2 | 2 | 1 | 78.3 | 2 to 5 |
| F | 2 | 3 | 1 | 1 | 3 | 73.3 | 2 to 6 |
| G | 2 | 2 | 3 | 2 | 2 | 71.7 | 2 to 4 |
| H | 3 | 3 | 1 (estimate) | 3 | 0 | 70.0 | 3 to 6 |
| A | 1 | 2 | 2 | 3 | 3 | 66.7 | 3 to 6 |
The score is each criterion's weight times its score out of 3, summed, so the maximum is 100. The ranges come from four weight sets:
| Weight set (depth, detail, volume, structure, path) | 1st | 2nd | 3rd | 4th | 5th | 6th |
|---|---|---|---|---|---|---|
| Base (30, 25, 15, 10, 20) | B | E | F | G | H | A |
| Equal (20 each) | B | G | E | A | H | F |
| Records-heavy (35, 30, 15, 15, 5) | B | E | H | G | F | A |
| Path-heavy (20, 20, 10, 10, 40) | B | F | A | G | E | H |
Read the two growth stories side by side. Company A, the fastest grower, never ranks above third, because three years of records and an unresolved rights question cap it however warm the path. Company B, which halved in size, ranks first under every weight set: its peak headcount still counts, and 22 years of records in one system score at the top. Shrinking did not remove the history; rapid growth did not create it.
Is the top of the list carried by records or by the relationship?
A rank can rest on the wrong thing. Split each company's base score into what comes from records (depth plus detail and outcomes) and what comes from path readiness.
| Company | From records | From path readiness | Reading |
|---|---|---|---|
| B | 46.7 of 85.0 | 13.3 | Records carry it |
| E | 55.0 of 78.3 | 6.7 | Records carry it; the path is the work to do |
| F | 45.0 of 73.3 | 20.0 | Warm-and-unproven; resolve rights first |
| A | 26.7 of 66.7 | 20.0 | Warm-and-unproven; carried by path and structure |
Our rule of thumb: when path readiness supplies more than a quarter of a company's score, label it warm-and-unproven and ask the record questions before the CEO question. F and A both cross that line, and F's swing from second under path-heavy weights to sixth under equal weights is what the label warns about. The OECD handbook calls the general problem a composite dominated by a few indicators.
From ranked list to asks, with owners and hours as limits
A rank orders whom to think about, not whom to contact. Three limits come first:
- The relationship owner decides. The person who can ask the CEO agrees first; if they decline, the row is marked and nobody goes around them. SourceX's screening page counts an introduction the decision maker did not know about among the cases that usually do not fit.
- One partner per company. The terms credit just one partner for each company, namely the partner whose valid introduction arrived first, and that credit only counts once the company files a verified application inside the 90-day window; they also treat any earlier lead, account or email correspondence with SourceX as an existing relationship. Agree internally who submits before two colleagues reach the same CEO.
- Hours. Cap the companies in active preparation at what your team can support; the capacity-limited queue sets out one way to do it.
Re-rank monthly or at a dated event, keeping each dated version. For comparing companies at different stages rather than ranking them, the stage-normalized scorecard is the companion tool, and the growth-versus-suitability plot shows the same tension as a picture.
Where a transparent ranking still misleads
- Precision it does not have. The gap between 78.3 and 73.3 is worth exactly one point on the volume criterion; the rank range is the honest output.
- Zero for unknown favors companies you know well, because closer relationships yield more facts. The sensitivity run and the evidence tags expose this bias without removing it; weights, likewise, become visible when written down, not correct.
- Demand is outside the model. SourceX checks relevance to its current data needs in review, and programs differ in size, history, language and category thresholds. A first-ranked company can still be declined today and fit another program later.
- Scores invite padding. If teammates know a high score earns attention, estimates drift up. The rule that an estimate scores at most 1 is there for that reason.
Methods note. The gates come from SourceX's program terms and screening page; the scored criteria from its published fit signals; the sensitivity design from the OECD and JRC handbook; the growth-list comparison from Deloitte's 2026 eligibility page. All were checked on 2026-10-26. The method does not measure demand, uniqueness, value, likelihood of acceptance or any payout, and no SourceX outcome data was used because none is published.
A measurable next step. Rank your current list once with the base weights and once with each alternative, and record every range. After the next ten CEO conversations, count how many CEOs in the top third agreed to talk compared with the rest. If the ranges were wide and the counts similar, simplify to the gates and a single depth score.
Next step
Refer a company. No cap on rewards. Default reward on a company's first qualifying transaction: the greater of $50,000 plus 5% and 15% of Actual Supplier Payout, with no cap; further qualifying transactions first agreed within 12 months of the first Closing pay 15%. Eligibility and payment conditions apply, including the USD 100,000 Minimum Transaction, and a signed Schedule A or addendum can differ. A lead, meeting or signature alone does not trigger payment.
When a company holds first place across every weight set and its relationship owner has agreed to ask, register for a partner link, confirm the gates against who qualifies or the company fit checker, and read the program terms before the CEO is contacted. The venture platform overview covers the rest of the program for funds.
Ranking candidates for introductions is referral work; deciding whether SourceX itself is worth investing in is a separate question, one the SourceX investor overview handles, with its materials sent to investors who request them.
This article is for information only and is not an offer to sell or a solicitation of an offer to buy securities. Any offering by SourceX is made only to verified accredited investors through its offering documents.