# Referral network health metrics dashboard | SourceX

Canonical: https://refer.sourcex.si/guides/the-network-health-dashboard-beyond-registered-partner-count
Last updated: 2026-11-05

Registered partner count measures sign-ups, not health. Put four signals on the dashboard instead, adapted from Google's site reliability practice: traffic (unique companies proposed), errors (duplicates, fit failures and introductions the CEO did not choose), latency (days from proposal to a screen decision, and from the CEO's yes to a verified application) and saturation (screening hours used against hours available). Headline one number: partners whose last 90 days produced a unique, fit-passing, non-duplicate proposal.

The dashboard is for heads of channel partnerships who report on a bench of introducers. Program facts were read on 26 October 2026 in SourceX's terms (Version 2026-10-22.2, since replaced by version 2026-11-04) and on the who-qualifies and rewards pages. The tiles, warning lines and month shown below are this page's own design and Illustrative.

## Why a registered count looks healthy while the network stalls

A registered count can only rise. Every conference, webinar and bulk invitation adds to it, and nothing a partner fails to do ever takes anything away. PartnerStack's glossary, in paraphrase, calls a partner inactive when they have joined but contributed nothing of value, and by that test a long registration list can be mostly inactive while its headline climbs every month.

The count also hides the cost side. Each proposal takes screening time, each duplicate wastes some, and each introduction a CEO did not choose spends trust that the next partner needs. A health dashboard has to show those costs next to the activity, or it rewards whatever adds names fastest.

## Activation and referral quality, as dashboard terms

| Term | Definition | Counted from |
|---|---|---|
| Registered | Created an account | Sign-up records |
| Enabled | Signed the Channel Partner Agreement and holds a partner link | Signature dates |
| Active introducer | At least one proposal reached your screen in the last 90 days | Proposal log |
| Quality-adjusted partner | At least one proposal in the last 90 days that was unique, passed the fit facts and was not a duplicate | Proposal log with screen results |
| Referral quality | Fit-passing unique companies divided by unique companies proposed | Screen results, each company counted once |

The fit facts behind every screen are the published ones. Headcount must have reached 50 or more full-time employees at its peak, counting no contractors; the records must be ones the company produced itself over years of operation, with the rights to license them; and the CEO must know about the introduction and want it. 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. Answers come from conversations at metadata level, never from files or logins, and SourceX's own review still decides relevance to its current programs.

## Four signals from site reliability, translated for introductions

Google's site reliability book names four golden signals for monitoring a service: latency, traffic, errors and saturation, and says that if a team can measure only four things about a user-facing system, these are the four. Two details from that chapter transfer especially well. Latency for failed requests should be tracked apart from successful ones, because a slow error is worse than a fast one. And errors come in three kinds: explicit failures, implicit ones where a request appears to succeed but returns the wrong content, and failures defined by policy.

| Signal | Translation for a referral network | Numerator and denominator | Warning line (this page's) | Who acts |
|---|---|---|---|---|
| Traffic | Unique companies proposed this month | Proposals, de-duplicated across partners and your claim log | Three months of decline | Partner manager |
| Explicit errors | Duplicates: companies SourceX already knew, or already claimed by another partner | Duplicates divided by unique companies proposed | Above 1 in 5 | Claim-log owner |
| Implicit errors | Proposals that look complete but fail the fit facts at screen | Fit failures divided by unique companies proposed | Above half | Whoever trains partners |
| Policy errors | Introductions the CEO did not know about, or sent by a partner without a signed agreement | A count | Any at all | Head of channel |
| Latency | Median days from proposal to screen decision, passes and fails measured separately; days from the CEO's yes to a verified application | Medians, with the slowest case shown | Fails slower than passes; any application past day 60 of the 90-day window | Screener |
| Saturation | Screening hours needed against hours available | Proposals times hours per screen, divided by available hours | Above 80% | Head of channel |

Policy errors carry the hardest line because SourceX's screening guidance lists introductions made without the decision maker's knowledge among the usual poor fits, and the terms require a signed agreement before anyone submits. Latency's second clock is the 90-day attribution window: credit belongs to the first valid introduction of a company SourceX did not already know, made inside that window, and it is credited only when the company submits a verified application.

## The quality-adjusted network dashboard

One screen, read top to bottom:

1. **Headline:** quality-adjusted partners, with last quarter's figure beside it.
2. **Four signal tiles:** traffic, errors split into explicit, implicit and policy, latency for passes and fails, saturation.
3. **Two context rows:** concentration (the top partner's share of fit-passing companies) and overlap (the share of proposals also claimed by another partner or network).
4. **Footer:** registered, enabled and active counts, small.

Each tile is paired with the tile that would expose gaming it. Traffic sits beside implicit errors, so a partner who floods proposals shows up as a rising failure rate. Fast latency sits beside policy errors, so speed bought by skipping the CEO's knowledge cannot hide. Saturation sits beside traffic, so a growth month that overloads screening reads as a warning rather than a win.

## One month on the dashboard (Illustrative)

Illustrative: an invented network in an invented month. No figure describes a real partner, company or outcome.

| Tile | Value | Reading |
|---|---|---|
| Registered / enabled / active | 180 / 72 / 21 | Most of the list is inactive |
| Traffic | 34 proposals, 27 unique companies | Seven proposals named a company someone else had already named |
| Explicit errors | 5 duplicates of 27 | 4 had earlier email correspondence with SourceX; 1 was claimed by two partners |
| Implicit errors | 9 fit failures of the remaining 22 | 6 had fewer than 50 full-time employees at peak; 3 kept records owned by clients |
| Policy errors | 0 | Hold the line |
| Errors combined | 14 of 27 | More than half of unique proposals failed a check; brief partners before recruiting |
| Latency | 6 days for passes, 15 for fails; 19 days median from yes to application, 2 past day 60 | Fails are slow: a no reaches partners late |
| Saturation | 51 hours needed, 40 available: 128% | Over capacity; a backlog is forming |
| Headline | 8 quality-adjusted partners behind 13 fit-passing companies | The working network is 8 people |
| Concentration | One partner, 5 of 13 fit-passing companies | 38% from one person |
| Overlap | 3 of 27 also proposed through another network | Settle by dated records |

Three problems surface that the registered count of 180 hides. The underperforming cohort: 60 partners enrolled after a spring webinar produced one proposal between them, and none passed. The duplicate risk: four of five duplicates were companies SourceX already knew, which a single question to the partner, whether the company had ever emailed or applied to SourceX, would have caught. The capacity limit: at 1.5 hours a screen, 34 proposals need 51 hours, and the team has 40, so fail decisions slip first and partners hear "no" two weeks late.

## Reading saturation beside support, concentration and overlap

Saturation is the tile that changes decisions fastest. Above 80%, pause recruiting before adding anything else; the [guide to prioritizing introductions when review capacity is limited](/guides/how-to-prioritize-introductions-when-review-capacity-is-limited) covers which proposals take the scarce slots. Support load belongs next to it: count partner questions per week, and the share answered by pointing to the terms.

Concentration needs its own line because one prolific partner can make the whole dashboard look healthy; [when one large introduction distorts the dashboard](/guides/when-one-large-introduction-distorts-the-entire-dashboard) separates a single row from the rest. Overlap is the inverse problem: partners in two networks inflate both networks' traffic, and only de-duplicated company counts remove it.

## Evidence gates and correction paths

A tile is trusted only when its records are. Before reading any tile, check three gates:

- **Completeness:** at least nine in ten proposals in the period carry a date, a partner and a screen outcome.
- **Criteria version:** every screen result notes the fit-criteria version used, because SourceX applies the criteria of a specific program and a company that fails one program may fit another later.
- **Definition version:** any change to a tile's definition restarts its trend line, with the date of the change printed on the tile.

Partners should see their own rows and correct them, with the reason recorded beside the change. Disagreements about attribution or duplicates go to SourceX with the referral details and supporting evidence, as the terms' dispute section sets out; the network's dashboard never overrides that decision. For the denominator tests each ratio should pass, see [the referral metrics that survive a denominator check](/guides/the-referral-metrics-that-survive-a-denominator-check).

## Where the dashboard misleads

- **Small numbers swing.** A network with 30 proposals a quarter should read medians and ratios quarterly, not monthly.
- **The headline lags.** A partner's 90-day window means a newly excellent partner takes weeks to count, and a lapsed one keeps counting for weeks.
- **Outcomes are absent by design.** SourceX's reward statuses run from estimated to paid only after acceptance, handoff, cleared buyer payment and the supplier's own payment, and a lead, meeting or signed agreement alone earns no payable reward, so licensing outcomes belong in a quarterly review, not on a weekly screen.
- **Never add projected rewards.** SourceX's rewards page says it does not provide income projections, and a dashboard that invents them turns a health check into a forecast nobody can stand behind.
- **Healthy is not the same as valuable.** A network with every tile green can still see no completed deals, because demand and rights are assessed in SourceX's review.

The month above is Illustrative, and SourceX has published no network health benchmarks; set warning lines from two quarters of your own records. For counting quality growth over quarters, see [measuring qualified introductions instead of logos](/guides/measure-growth-in-qualified-introductions-not-logo-counts), and for the partner ladder underneath the headline, [the registered-enabled-active ladder](/questions/how-to-qualify-referral-partners). The [business health check template](/resources/business-health-check-template) applies a similar records-first idea to the companies themselves.

## 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 the headline shows who your working introducers are, make sure each has [registered for a partner link](/register), screens against [who qualifies](/who-qualifies) and has read the [program terms](/terms) before their next proposal.

## Sources

- https://sre.google/sre-book/monitoring-distributed-systems/
- https://partnerstack.com/glossary/inactive-partner
- https://refer.sourcex.si/terms
- https://refer.sourcex.si/who-qualifies
- https://refer.sourcex.si/rewards
