Companies recruit, appoint, license, certify, and integrate third-party sellers, then report those counts as distribution. Eleven cases across insurance, mortgage, travel, real estate, software, and solar show what actually separates a recruited network from a producing one, and how far apart the two numbers sit when a company publishes both.
The discipline of this study is refusing to let any of these three stand in for another. Every case below where the numbers are disclosed shows them diverging by large multiples in the same period.
Distributors who signed, licensed, appointed, integrated, or certified. Measures potential supply. Cheap, scalable, and almost always the number that gets reported.
Distributors who transacted at least once in a defined recent window. Measures whether potential supply became real.
Distributors who transact repeatedly without prompting, and the share of their own volume they give you. This is the only stage that is a channel.
Companies that publish only the first are not being evasive. They are reporting the one they control.
Independent distributor networks do not become productive because more distributors are recruited or trained. Sustained production emerges when three things are true together. The three are not equally weighted and they do not behave the same way.
The evidence for compensation moving independent distributor behavior is overwhelming in one direction only. Remove it and production stops quickly, reliably, and even on products customers are actively choosing. Raise it above the level that clears the distributor's opportunity cost and this cohort contains no case of a durable step change in output.
Model compensation as a gate with a hard floor rather than a dial with a slope. Below the floor, nothing else can be fixed. Above it, expect little from paying more. Detailed in section 05.
The sharper version of the workflow claim is not that being quotable inside the distributor's system produces volume. Availability is a precondition. What produced volume, at a magnitude that required twenty years of federal rulemaking to address, was ordering: which option appeared first on the screen the agent was already looking at, in the transaction they were already completing.
Where the distributor must define what a product is before selling what it does, every channel here that produced volume had a paid layer between manufacturer and seller whose job was education. Independent marketing organizations in annuities, Territory Partners in pet insurance, uplines and field management in supplemental health. Where that layer was absent, distributors defaulted to the products they could already explain.
The model is a conjunction, not a set of universal laws. The strongest claim the evidence supports is that these three conditions account for the observed variation better than any single-factor account, and that a channel weak on one tends to be weak in output regardless of strength on the other two. It does not support the claim that any single failure drives production to zero in all cases.
The Medicare Advantage evidence for the economics condition is close to experimental. The evidence for the other two conditions is directional and historical rather than experimental, and is weaker for that reason.
Training and enablement do not produce durable productivity change. Primerica has built training and tools for the largest independent life insurance sales force in North America for fifteen years, and its disclosed productivity metric has stayed inside roughly 0.16 to 0.24 policies per licensed representative per month across that entire period Fact. The number moves with market and mix. It does not move with enablement.
Customer demand is not sufficient by itself. Medicare Advantage plans that customers were selecting at growth rates of 54 percent and 112 percent went into sustained decline when the compensation attached to them was removed, with the plans unchanged and still available Reported. Demand routes through the distributor's incentive rather than around it.
Term life and third-party investments sold through a very large network of independent, largely part-time, life-licensed representatives.
Following convention-timed incentives in Q3 2024 the company recruited 142,655 people and added 14,349 licenses Fact. The FY2021 10-K describes the historical productivity range as 0.18 to 0.22; a 2025 release describes it as 0.20 to 0.24 Fact. At roughly two policies per licensed representative per year, the arithmetic strongly suggests a conclusion the company does not state: production concentrates in a small subset and most licensed representatives sell almost nothing Inference, from disclosed averages. The distribution itself is not disclosed.
Recruits and their families are themselves natural buyers, so mass recruitment may function partly as customer acquisition. In that model a low producing ratio is a design feature rather than a broken channel Inference. This complicates the cleanest number in the study and is left in deliberately.
Recruitment and licensing scale independently of production. Fifteen years of enablement investment did not move the productivity band.
Aflac is unusual in publishing both sides of the ledger. Its quarterly supplement reports Recruited Agents beside Average Weekly Producer Equivalents in the same table Fact.
| Year | Recruited agents | Average weekly producer equivalents |
|---|---|---|
| 2021 | 16,086 | 5,993 |
| 2022 | 11,050 | 6,186 |
The FY2022 annual report describes roughly 6,200 agents including brokers actively producing on a weekly basis, and states that the company treats this as a leading indicator of future production capacity Fact. On the Q1 2026 call, management described a new agent conversion rate of 16 percent, defining conversion as producing about $25,000 and adding three new accounts in the first three months Reported, earnings call transcript, May 2026.
The broker pivot is the activation lesson. In 2021 an Aflac executive explained the shift on the grounds that brokers arrive already licensed and already familiar with the products Reported, ThinkAdvisor, 30 Jul 2021. Recruited brokers rose from 1,861 in 2020 to 5,445 in 2021 Fact.
The producing base sat near 6,000 while recruitment swung between 11,000 and 16,000. Selection at intake appears more powerful than enablement after intake.
Plans cannot be withdrawn mid-year and their benefits and premiums cannot be changed mid-year. The one remaining lever to slow enrollment is agent compensation, and agents are the primary distribution channel.
UnitedHealthcare made approximately 220 plans non-commissionable effective 1 July 2025 Reported, Medicare Market Insights, 5 Mar 2026, citing carrier commission schedules matched to CMS monthly enrollment data. The plans stayed available with identical benefits.
The selection detail is what makes this decisive. The plans whose commissions were pulled had grown 54 percent during the preceding annual enrollment period, from 348,178 to 537,619 members, while plans that stayed commissionable shrank 1 percent over the same window Reported. Humana repeated the maneuver effective 1 April 2026 on plans carrying 112 percent growth Reported.
Corroboration is broad. Insurers cut commissions on 15 to 20 percent of plans ahead of the 2026 enrollment period Estimate, Chapter via CNBC, 29 Sep 2025. Aetna, Cigna, Elevance, and Medical Mutual announced non-commissionable plans during November 2024 Reported. A coalition including NABIP, Health Agents for America, the Big I, and NAIFA petitioned CMS over the practice Reported, 13 Oct 2025.
Removing compensation is fast, repeatable, and strong enough to override active customer preference. A plan that pays nothing does not leave the market. It leaves the conversation.
Through the 1980s and 1990s travel agents were the dominant airline channel and booked through terminals supplied by airline-owned reservation systems. Over 90 percent of agents were on a system by 1985 Estimate.
System-owning carriers ordered displays to place their own flights first. The Civil Aeronautics Board adopted rules against display bias in 1984, the Department of Transportation readopted and amended them in 1992, and the display-bias prohibition survived the 2004 deregulation of nearly everything else in the rules Fact, 49 FR 32540; 14 CFR Part 255; Federal Register, 7 Jan 2004. The literature names the residual advantage the halo effect: incremental bookings accruing to the system owner from the tendency of agents to book the carrier providing the terminal Reported, Journal of Air Transport Management, 2001.
Separately, carriers paid override commissions for booking a target share of an agency's volume. The DOT Inspector General reported in 1999 that overrides could induce agencies to recommend services that increased their own commissions, and noted that not all agencies had such arrangements Fact.
Competing carriers were broadly available within the systems. What was contested, litigated, and regulated for two decades was ordering. This is the cleanest available separation of presence from position.
Trupanion sells through referrals from veterinary practices, independent businesses with no obligation to mention it. The company fields Territory Partners, described in its 10-K as independent contractors paid based on activity in their regions whose role is to build long-term relationships with veterinarians and educate them about the product Fact, 10-K FY2023. It also embeds in practice workflow through patented software that pays the veterinarian directly at checkout Fact, 10-K FY2020.
The strategy section separates two objectives explicitly: increase the number of hospitals that refer, and separately increase the rate at which already-active hospitals refer Fact, 10-K FY2019 and FY2020. That is the activation and sustained-production split, written by a company into its own filings. The standing risk paragraph names the willingness of Territory Partners and veterinarians to recommend the product alongside the ability to increase active hospitals Fact.
Where the distributor must explain an unfamiliar category, the vendor pays for a human layer to carry it, and tracks referral rate per active distributor as a metric distinct from distributor count.
FIAs were sold overwhelmingly by independent agents, with independent marketing organizations accounting for roughly two thirds of sales in 2015 Estimate, LIMRA via InvestmentNews, Aug 2016. Fixed annuity sales hit a record $117.4 billion in 2016 and FIA sales reached a record $60.9 billion Fact.
The Department of Labor's fiduciary rule moved FIAs under the Best Interest Contract Exemption, which required a financial institution to contract with the investor. IMOs did not qualify Fact. LIMRA reported FIA sales declining quarter over quarter after the reclassification Reported, and first-quarter 2017 FIA sales fell 13 percent to $13.6 billion Reported. Carriers responded by launching fixed-rate products with income riders, which is product redesign to keep an existing channel able to sell Reported, May 2017.
The highest-commission product in its category captured the independent channel. When the structure supporting that channel was disturbed, volume fell with customer demand unchanged, and manufacturers changed the product rather than retraining the channel.
Ended 2025 with 83,060 agents and 440,163 transactions, roughly 5.3 per agent per year Fact. Agents with fewer than three transactions a year accounted for 77 percent of attrition Reported, and 63 percent of departing non-productive agents leave the industry entirely Reported. Agent count fell 5 percent in 2024 while output per agent rose, and the company eventually reframed its own reporting around productive agent count. Network size and network output decoupled completely.
In March 2021 UWM required brokers to stop submitting loans to two named competitors as a condition of continuing, with liquidated damages of $5,000 per loan or $50,000, whichever was greater Fact, court filings. Of roughly 12,000 brokers in the network, UWM said only about 600 refused Reported. Rocket said its wholesale volume rose roughly 20 percent in 2021 and that hundreds of signatories returned within a quarter Reported, Apr 2022, and UWM has litigated repeatedly against brokers who continued sending loans elsewhere Fact. Exclusivity buys share where the vendor's economics already make leaving costly. It does not create production, and compliance leaks.
Sunnova's incoming CEO stated that by early 2025 the company could not pay its dealer network, that contractors stopped work on in-process projects, and that this further reduced incoming cash flow Reported, Jun 2025. Sunrun, moving the other way, cited deliberate reductions in its affiliate channel, the bankruptcy of a major installation partner, and a slower than expected ramp onboarding in-house reps, with subscriber additions down 31 percent year over year Reported, Aug 2026. Independent distributors are businesses with their own solvency, and replacing them is slow even for a well-capitalized incumbent.
An early-1990s program awarding points to the individual agent rather than the agency. Approximately 52,000 agents enrolled by October 1993 at a rate near 350 per day, with Radisson's agent bookings reported up roughly 60 percent and no material change in other marketing Reported, US Patent 5,483,444. Treat with caution. The source is filed by parties with an interest in the claim, there is no control group, and the period coincides with rising agency automation Unresolved as to effect size. Included because compensating the individual decision-maker rather than the firm is a distinct lever from raising the rate, and because it is one of the few claimed step changes in the cohort.
Commonly cited figures hold that about 80 percent of channel-sourced revenue comes from roughly 20 percent of partners, that only about 11 percent of partners reach the thresholds for significant incentives Estimate, and that fewer than 30 percent of registered partners ever generate a deal Estimate. Every one of these comes from a firm selling channel enablement services and none publishes its method Unresolved as to precision. They are directionally consistent with the disclosed numbers above, which is the only reason they appear here, and they should not be read as independent measurement.
Zenefits, which competed with brokers rather than distributing through them. Herbalife and Avon, where compensation for recruitment itself confounds the production question. Automotive finance and insurance attach rates, where the sellers are employees rather than independents. Zillow Premier Agent, where the published data measures agent spend rather than agent production.
Medicare Advantage removals, the FIA disruption, Sunnova's dealers stopping work when payment stopped. The only mechanism with documented, fast, repeatable effect. The evidence is asymmetric and the asymmetry is the finding.
Twenty years of display-bias regulation and the halo effect literature. Not availability. Ordering.
Territory Partners, annuity IMOs, and Aflac's stated reason for moving toward brokers.
Aflac's broker pivot. eXp's finding that agents on teams are 78 percent more productive than individuals Reported.
UWM. Buys share, generates litigation, leaks at the edges.
Radisson, single uncontrolled source. Aflac's use of contests to reactivate lapsed producers points the same way.
Likely powerful and not measured cleanly here. The Medicare case suggests it supplements economics rather than substituting for it Unresolved.
Evidence against is the Primerica productivity band. Evidence for is selection-biased, since distributors who complete certification are the ones already committed. Vendor lift figures for certified partners describe who self-selects, not what certification does Inference.
Every strong compensation result in this study is a removal. That asymmetry is not a gap in the evidence to be filled in later by assumption. It is the finding.
The relevant comparison is never the payout against zero. It is the payout against what the distributor would earn using the same minute of the same conversation on something else.
Signed something. Drop-off is administrative. Primerica converts roughly 12.5 percent of recruits to licensed in a quarter Fact, computed from the Q1 2026 disclosure.
Licensed, appointed, integrated, trained. Drop-off is that nothing in the distributor's day changed. No demonstrated relationship to later production anywhere in this cohort.
The primary gate appears to be occasion rather than stated willingness. Aflac's own conversion definition is cleared by 16 percent of new agents Reported. Many first sales are relationship-driven, sold to family or to a client the distributor was already mid-conversation with.
The first sale that had to win on its merits against the distributor's alternatives. Lost to a competing use of the same minute. eXp's productivity threshold sits near here, with sub-three-transaction agents accounting for the overwhelming majority of attrition Reported.
The product enters default presentation. Everything in this cohort that purchased this stage purchased it with economics, position, or contract. Nothing purchased it with education.
The largest absolute loss is between enabled and first sale. The largest strategic loss is between first sale and repeat, because that group was already proven capable Inference.
The distributor is willing and the product fits no moment in their existing work, so selling it requires manufacturing a conversation. The most common silent failure, and from outside it is indistinguishable from indifference.
Either the vendor cuts it or a competing product starts paying more for the same minute. The distributor does not announce the switch.
Sunnova. Independent distributors do not extend credit to suppliers.
The FIA channel seized when the layer that made the sale compliant and explicable was disqualified.
Independent sellers price license exposure heavily, because the license is the business.
Because production concentrates, losing a few distributors removes a large share of volume.
Mechanism plausible, measurement absent. See H7, which remains unresolved.
Producing ratio. Distributors who transacted in the last ninety days, divided by distributors appointed.
Concentration. Share of volume from the top decile of producing distributors.
Test the occasion. For a distributor doing an ordinary week, in which specific transaction does this product appear without them having to remember it. If the answer requires the word "remember," the constraint is occasion and no incentive change fixes it durably.
Test economics against the alternative, not against zero. Earnings per unit of distributor effort on this product versus the product they would otherwise sell in the same minute.
Test explanation burden. How long a competent distributor takes to explain the category to a customer who has never heard of it. If that exceeds the length of the transaction it attaches to, the constraint is explanation, and the fix is a layer or a product change rather than a training deck.
Test post-sale burden. Count the actions required after compensation is earned. Each is a decay point.
Only then test customer demand. If distributors produce where demand exists and not where it does not, the constraint belongs to marketing rather than to channel management.
Both ratios are diagnostic heuristics, not established breakpoints. Nothing in this study identifies a threshold at which channel behavior changes, because no company in the cohort discloses the underlying distribution.
As rough orientation drawn from disclosed averages and from industry estimates of uncertain quality: a producing ratio in the low tens of percent and top-decile concentration in the high seventies or above appear to be ordinary rather than pathological in mature independent channels. A company's own trend in both numbers is far more informative than its level against any external benchmark.
| Symptom | Likely constraint |
|---|---|
| Low producing ratio, high concentration, stable producers | Not supply. Recruiting more will not help. |
| Many first sales, few second sales | Repeat production. Check economics versus alternative and post-sale burden. |
| Sales cluster among distributors on one platform | Position, and workflow friction for everyone else. |
| Distributors quote it, customers decline | Demand or product, not activation. |
| Distributors will not quote it despite customer interest | Economics or explanation burden. |
| Production drops immediately after a compensation change | Economics, confirmed. |
| Production drifts down with no policy change | Post-sale burden, or a competitor paying more for the same minute. |
| Volume concentrated in the first ninety days after onboarding | Incentive-driven activation that never became habit. |
A reader could look at the same evidence and conclude that only one condition is doing real work.
The Medicare Advantage case is the cleanest data here and it is purely an economics story. The plans were unchanged, still in the enrollment platforms, still inside the workflow, still being actively chosen. Only the commission changed and enrollment reversed. Carriers now use this deliberately as a margin management tool, meaning the practitioners closest to the question have concluded that money alone moves independent distributors within weeks. UWM points the same way, since exclusivity is an economic instrument and roughly 95 percent of a twelve-thousand-broker network accepted a restriction on their own product choice. So does the FIA case, where the highest-commission product captured the channel.
If that reading is right, position and explanation layers are downstream of price, and a company with weak production should raise the payout before anything else.
Primerica converts about one in eight recruits to a license and averages roughly two policies per licensed representative per year. Aflac recruits two to three times its producing base annually. eXp reframed its own reporting around productive agent count.
The operative variable is position inside an existing transaction rather than integration availability. Airline display regulation is the strongest evidence in the study and it concerns ordering, not access.
Plans growing 54 percent and 112 percent declined when compensation was removed while remaining available and unchanged. Revised: what beats vendor push is anything that removes the distributor's need to persuade. Customer demand is one route to that, not the only one, and it does not survive a compensation cut.
Economics is asymmetric rather than diminishing. Below the clearing threshold, production collapses. Above it, no documented durable effect in this cohort. This correction is the study's most useful output.
eXp's attrition composition, Aflac's stable producing core against volatile recruitment, and industry concentration estimates all agree.
The distinct requirement is a paid layer carrying the explanation. Trupanion's Territory Partners, the annuity IMO layer, and Aflac's stated reason for preferring brokers.
No case in this cohort discloses compliance rates on post-sale distributor obligations over time. The supporting logic is strong, since the obligation arrives after the compensation, competes with revenue-generating work, and is rewarded by nothing in the distributor's economics. The supporting data does not exist in the public sources searched. This is the largest gap in the study, and closing it would require private data or a vendor that discloses renewal-action compliance rates. It should not be treated as supported by analogy to the other findings.