Company acquires · consumer pays
Direct + consumer paid
The hardest structure at high transaction values. Most companies here pivoted, exited low, or died.
Farewill · Talkspace before 2022 · Wevorce · tax resolution firms
Counterexample: LegalZoom.
01 / The problem
Some customers disappear as soon as you acquire them.
The archetype
- 01The consumer needs the service once, or a handful of times, in a lifetime.
- 02Demand is created by a triggering event the consumer did not schedule.
- 03The consumer has no prior awareness of the category, let alone the provider.
- 04The purchase requires unusual trust, often at a moment of impaired judgment.
- 05The transaction is high value relative to the consumer's normal discretionary spending.
- 06Some third party is already present at or before the trigger.
A company need not have all six properties to belong.
The economic consequence
- Brand investment has less opportunity to compound, because the audience turns over completely.
- Retention may barely exist.
- The moment of intent is narrow, contested, and simultaneous across every competitor.
- Acquisition cost often has to be recovered inside a single transaction.
The lifecycle
- Trigger
- Need
- Search / Referral
- Decision
- Transaction
- Customer disappears
This is a structural business problem, not a marketing complaint. The customer leaves the market entirely — and takes the return on every acquisition dollar with them.
02 / What we expected
The obvious answer was distribution.
If someone already knows the customer needs the service, perhaps the company should stop trying to find the customer itself.
- Insurers
- know when claims occur.
- Employers
- already have employees.
- Financial institutions
- already hold customer relationships.
- Doctors
- know when care is needed.
- Funeral homes
- know when a death occurs.
- Banks
- know when a mortgage closes.
The expected pattern
Direct to consumerExpensive and inefficient
Institutional distributionScalable and durable
That was the hypothesis the research was designed to test.
The historical record didn’t support that binary.
03 / The first finding
Separate acquisition from payment.
The strongest pattern wasn’t who acquired the customer. It was whether the consumer bore the full cost.
Company acquires · someone else pays
Direct + someone else pays
The fastest-scaling and least-discussed pattern in the study. Attracts regulators.
The Knot / XO Group · A Place for Mom
Institution acquires · someone else pays
Institution + someone else pays
The most established and most studied institutional model.
Progyny · Empathy · Talkspace after 2022 · employer legal plans · property restoration
Counterexample: Accolade.
Institution acquires · consumer pays
Institution + consumer pays
The weakest documented structure. Referral relationships form easily and produce little.
Avvo Legal Services · most estate-planning referral arrangements
Not every company fits perfectly, and the quadrants describe structure, not causation. LegalZoom is an important counterexample to the consumer-paid thesis: it scaled direct, consumer-paid acquisition at a $99–$500 price point and went public in 2021. FACT The direct-and-consumer-paid model appears viable at low transaction values and not at high ones — the exact threshold is not established by these cases. INFERENCE
04 / The overlooked quadrant
You can acquire the customer directly without making the customer pay.
Two of the largest successes in this study acquired consumers directly, at scale, through content and search, in categories that are once-in-a-lifetime and event-triggered. Both were paid by the supply side rather than by the consumer.
Weddings · XO Group
The Knot
- Trigger
- Engagement
- Consumer acquisition
- Direct — content, search and brand
- Consumer price
- Free
- Payer
- Wedding vendors
~75%
of US couples planning a wedding reached, per XO Group SEC filings FACT
27,000
paying local partners FACT
$152M
revenue in 2016, at a 21% adjusted EBITDA margin FACT
$933M
reported merger with WeddingWire FACT
Why the payer participated
Wedding vendors already spent money acquiring customers — XO's own filings note US small businesses generally spend 5 to 10% of revenue on marketing FACT. The Knot did not need to convince them to create a new budget. It redirected an existing one.
Senior living placement
A Place for Mom
- Trigger
- A senior care crisis, usually sudden
- Consumer acquisition
- Direct — search and brand
- Consumer price
- Free
- Payer
- Senior living communities
14,000+
communities and home care providers in the network, described as the largest of any senior care referral service FACT
1 month
of rent — the referral commission commonly paid by communities on a successful placement FACT
$3.5–12K
per move-in, depending on market and care level (2026 industry source) FACT
Why the payer participated
Move-in occupancy is the single metric senior living operators manage. The referral fee is priced against a month of rent on a resident who may stay years.
The cost of the model
Supply-side payment can create a structural conflict between the consumer’s recommendation and the source of revenue — and regulators eventually address it.
- A Washington Post analysis reviewed inspection reports for 863 communities on the company's 2023 and 2024 “Best of Senior Living” lists and found 324 — 37.5% — cited for serious violations affecting resident care. FACT
- The chair of the US Senate Aging Committee opened an investigation in June 2024, requesting three years of revenue data and sample partnership agreements. FACT
- More states now require disclosure of financial arrangements, and an FTC advance notice of proposed rulemaking addressing referral fee disclosure was in play as of 2026. FACT
The customer and the payer do not have to be the same person.
05 / The second finding
Don’t find the customer owner. Find the metric owner.
Partners distribute against metrics they already manage.
Progyny
Employer / health plan
Multiple-birth cost, medical cost, talent retention
Property restoration
Insurance carrier
Claim cost, cycle time
A Place for Mom
Senior living community
Occupancy
The Knot
Wedding vendor
Customer acquisition, marketing ROI
LifeLock breach channel
Breached enterprise
Liability and remediation
Papa
Health plan
Cost of care, plan performance
Employer legal plans
Employer
Benefit competitiveness, enrollment
Estate-planning platforms
Wealth / advisor channel
Relationship depth, retention
- A partner having access to the customer is not enough.
- A partner thinking the service is useful is not enough.
- A partner liking the brand is not enough.
The strongest historical channels connected the product to a number the distributor already monitored, budgeted against, or was accountable for. In every distributed case examined, the payer’s motivation traced to a number it already tracked. INFERENCE
“Good for our customers” is not a distribution strategy.
06 / The third finding
A partnership is not a channel until something initiates the handoff.
The research initially assumed “embedded” distribution would solve activation. It didn’t. Passive services sitting inside partner systems can have extremely low utilization.
The evidence instead produced an activation ladder — four tiers, ordered by observed yield.
1
Automatic / contractual
The system initiates the handoff. No human decides whether to refer.
Highest observed activation
Property restoration TPA dispatch · LifeLock breach channel · embedded checkout insurance
2
Scheduled / proactive
A recurring process or an outbound action initiates.
Moderate to strong
Employer open enrollment · Papa proactive outreach
3
Discretionary professional
A trusted human has to remember — and choose — to refer.
Meaningful, but substantial leakage
Physician referrals · bank credit-linked products
In a peer-reviewed study of 12,282 referral and test orders, completion within the designated time was 52.9% for resident physicians and 58.4% for attendings. FACT
4
Passive availability
The service exists inside the partner's system. The customer has to find or remember it.
Lowest observed activation
Employee benefit point solutions · passive partner portals
The average employer offers roughly 30 to 40 benefits programs; the share of employees actually using them is under 5%. FACT
Embedded does not mean activated.
A product can sit inside a partner ecosystem and still have almost no distribution. Discretionary professional referral — tier 3 — outperforms passive embedding by an order of magnitude, even though tier 4 is “embedded” and tier 3 is not. That is the finding that breaks the original hypothesis.
The cleanest commercial illustration is vendor-published and should be treated as promotional: insurance presented inside a checkout flow is reported to attach at 3 to 20%, while post-purchase offers for the same product, same partner, and same customer typically see 1 to 2%. VENDOR-SOURCED
07 / Two leaks
Initiation and completion are different problems.
The handoff, measured
- Eligible customer
- Initiation rateHandoff
- Completion rateDelivered customer
Some channels fail because nobody initiates the referral. Others initiate successfully but lose the customer between referral and delivery. The interventions are different.
The clean evidence
18%
Before
73.3%
After
Denver Health's closed-loop referral rate, before and after systematic tracking was implemented — saving approximately 498 staff hours per year. FACT
This is a healthcare referral case. It is used here as evidence about handoff mechanics — that instrumenting completion quadrupled closure without changing referrer behavior — not as proof that commercial referrals will produce identical results.
If neither partner can see what happens after the referral, you don’t have a measurable channel.
08 / The trade
Better distribution usually means less control.
Property restoration is the mature end state of institution-controlled distribution. Automatic carrier and TPA dispatch produces reliable customer flow. In exchange, the distributor sets the terms.
What the distributor controls
- Pricing — pre-negotiated
- Quality requirements — externally audited
- Allocation — assigned by the carrier or TPA
- Payment terms — commonly 30 to 60 days or more
- Performance measurement — satisfaction scores and cycle times
- Removal from the network — probation or delisting for missed benchmarks
Property restoration
40%
Single-carrier concentration above 40% of revenue reduces a restoration business's valuation multiple by one to two turns in 2026 M&A practice. Dependence is priced as a discount, not an asset. FACT
Talkspace
70% → 54%
Gross margin during the institutional transition, partly on B2B mix shift and continued clinician-network investment. The revenue model improved; the margin did not. FACT
Papa
~36
Payer and employer relationships not renewing for 2024 — in a single cycle, with declining Medicare Advantage supplemental funding as a contributing cause the vendor did not control. Strong activation did not prevent it. FACT
Avvo
8 states
Ethics opinions concluding attorneys could not participate in Avvo's fixed-fee service. Automatic routing did not protect the channel from professional-conduct rules aimed at the distributor, not the company. FACT
Activation and durability are separate problems.
The strongest channel may also become your strongest source of dependency.
09 / The decision model
Four decisions before you build the channel.
01
Decide who should pay
- Does the consumer need to bear the full price?
- Could another party economically benefit enough from the transaction to subsidize or fund it?
- Could the supply side pay for demand?
- Could an institution pay because the service reduces another cost?
Do not assume changing acquisition alone fixes the economics.
02
Find the metric owner
- Who already loses money because this problem exists?
- Who has unused capacity?
- Who incurs servicing cost?
- Who risks losing the customer?
- Who has liability?
- Who already spends money acquiring this demand?
- What measurable number does the service improve?
03
Design the initiation
- Exactly what causes the customer to encounter the service?
- Is it automatic?
- Triggered by an event?
- Attached to an existing transaction?
- Scheduled?
- Proactively initiated?
- Or dependent on someone remembering to make a referral?
Do not accept “Our partner will tell customers about us” as a channel design.
04
Instrument completion
- Can the partner see the handoff?
- Can the company see whether it was accepted?
- Can both sides measure conversion?
- Can leakage be identified?
- Can volume be attributed to the partnership?
- Can the process improve over time?
10 / Channel scorecard
A simple test for a potential distribution partner.
Five dimensions, each read as a spectrum. A channel is evaluated across all five — not by asking whether a company “has the audience.”
Payment
Who pays?
- Weak
- High consumer out-of-pocket cost
- Strong
- Partner or supply-side economics fund the transaction
Incentive
Why does the partner care?
- Weak
- Customer experience — nice to have
- Strong
- Moves a metric with an owner and a budget
Initiation
What starts the handoff?
- Weak
- Someone remembers to refer
- Strong
- Automatically triggered by an existing event or workflow
Completion
Can the referral be tracked?
- Weak
- Lead sent, introduction made
- Strong
- Closed-loop visibility from trigger to transaction
Dependency
What control does the partner gain?
- Weak
- Partner controls pricing, allocation and customer access
- Strong
- Channel adds volume without becoming existential
11 / Case index
The historical record.
A condensed index of the researched companies. Two further recorded cases — tax resolution firms and Wevorce — could not be verified and are not relied on in this artifact.
Farewill
Wills and probate, UK
- Trigger
- Death
- Acquisition
- Direct — content and brand
- Payer
- Consumer
- Outcome
- Largest UK will writer at ~10% share by 2020; sold to Dignity for £12.9M, October 2024
Talkspace
Mental health
- Trigger
- Distress episode
- Acquisition
- Direct paid acquisition; pivot to payers from 2022
- Payer
- Health plan, employer
- Outcome
- Consumer revenue under 8% of total by end of 2025; acquired for ~$835M, March 2026
LegalZoom
CounterexampleLegal services
- Trigger
- Legal need
- Acquisition
- Direct, at scale
- Payer
- Consumer
- Outcome
- IPO 2021 — scaled at a $99–$500 price point
The Knot / XO Group
Weddings
- Trigger
- Engagement
- Acquisition
- Direct — content, search and brand
- Payer
- Wedding vendors
- Outcome
- $152M revenue in 2016 at 21% adjusted EBITDA; $933M reported merger
A Place for Mom
Senior living placement
- Trigger
- Care crisis
- Acquisition
- Direct — search and brand
- Payer
- Senior living communities
- Outcome
- 14,000+ network; Senate Aging Committee investigation, June 2024
Progyny
Fertility
- Trigger
- Infertility
- Acquisition
- Pivot to employer-paid benefit, 2016
- Payer
- Employer
- Outcome
- ~$1.1B revenue in 2024; ~6.7M covered lives
Empathy
Bereavement
- Trigger
- Death
- Acquisition
- Institution-first
- Payer
- Life carriers, employers
- Outcome
- $162M raised; partnerships with eight of the ten largest US life carriers
Papa
Elder companionship
- Trigger
- Aging, isolation
- Acquisition
- Institution-first — Medicare Advantage, employers
- Payer
- Health plan, employer
- Outcome
- ~36 payer and employer non-renewals for 2024
Accolade
CounterexampleCare navigation
- Trigger
- Health need
- Acquisition
- Institution-first
- Payer
- Employer
- Outcome
- $414M revenue with ~$100M net loss; sold for ~$621M, 2025
Employer legal plans
MetLife Legal Plans, ARAG, LegalShield
- Trigger
- Any legal event
- Acquisition
- Employer enrollment cycle
- Payer
- Employer, employee premium
- Outcome
- Operating since 1977; category stable for nearly fifty years
Property restoration
Servpro, Belfor, ServiceMaster, TPA layer
- Trigger
- Fire, flood
- Acquisition
- Carrier and TPA dispatch
- Payer
- Insurance carrier
- Outcome
- PE consolidation; single-carrier concentration above 40% priced as a discount
LifeLock
Identity theft
- Trigger
- Data breach
- Acquisition
- Four concurrent channels, including breach-paid
- Payer
- Consumer and breached enterprise
- Outcome
- $650M trailing-twelve-month revenue; $2.3B exit to Symantec, 2017
Avvo Legal Services
Fixed-fee legal
- Trigger
- Legal need
- Acquisition
- Platform-routed
- Payer
- Consumer
- Outcome
- Shut July 31, 2018, after eight state ethics opinions
Atrium
Legal services
- Trigger
- Legal need
- Acquisition
- Direct sales
- Payer
- Consumer
- Outcome
- Wound down March 2020; lawyers left and took clients
Estate-planning platforms
Trust & Will, Wealth.com, Vanilla
- Trigger
- Planning
- Acquisition
- Advisor and enterprise channel
- Payer
- Advisor firm, consumer
- Outcome
- LPL, Vanguard and Schwab relationships
Addiction treatment
Category-level case
- Trigger
- Crisis
- Acquisition
- Paid search and lead generation
- Payer
- Consumer, insurer
- Outcome
- Google removed the category in 2017; certification gate since 2018
12 / What the record doesn't prove
The evidence has limits.
PastBehavior distinguishes what happened from what we infer from what happened. These are the major limitations, stated plainly rather than footnoted.
- Survivorship bias is severe and runs one direction. Institution-paid successes are documented because they scaled, filed, and were acquired. Companies that attempted institutional distribution and failed quietly generate no coverage — the true base rate is worse than these cases imply.
- Institutional failures are less likely to be documented at all. Partnership announcements are abundant; post-mortems are close to nonexistent.
- A funding-environment confound overlaps several outcomes. Talkspace's pivot, Farewill's down exit, Papa's non-renewals, and Accolade's decline all cluster between 2022 and 2025, when capital repriced.
- The embedded-insurance conversion figures are vendor-sourced. The direction is corroborated across independent vendors and non-vendor cases, but the specific multiples should be treated as promotional.
- Clean before-and-after data on a commercial referral partnership is scarce. The best instrumented case in the set — Denver Health — is clinical, not commercial.
- Healthcare referral completion may not generalize. The strongest completion evidence comes from clinical settings with mandatory documentation, which is why the data exists there and nowhere else.
- The exact consumer-price threshold remains unresolved. LegalZoom at a few hundred dollars scaled direct-and-consumer-paid; Farewill at a few hundred pounds did not.
- Signed partnerships rarely disclose actual referral volume publicly. Attribution could not be examined at all.
The decision
When the customer only needs you once, don’t start with the partner list.
- FirstWho should pay?
- ThenWho already has an economic reason for this transaction to happen?
- ThenWhat automatically or proactively initiates the handoff?
- ThenCan you see whether it actually completes?
The historical record does not say every episodic consumer business should become B2B2C.
It says something more useful.
The companies that built durable channels did not simply find organizations with access to their customers. They found a payer with an economic reason to act, connected the service to a metric that already mattered, and designed a mechanism that caused the handoff to happen.
A partnership agreement creates permission. Distribution begins when the customer actually moves.
Methodology and source notes
- Case selection
- Eighteen historical cases across ten event-driven consumer categories — legal services, fertility, senior care, mental health, insurance, weddings, property restoration, bereavement, elder companionship, and identity protection among them — plus a targeted follow-up study of partner activation covering thirteen activation cases, from embedded insurance to physician referrals. Both research documents were compiled in August 2026 from public sources.
- Evidence labels
- FACT — filing, press release, or on-record statement. INFERENCE — conclusion drawn across cases, labeled as such. VENDOR-SOURCED — published by a company selling the solution being measured; treated as directional, never neutral. UNVERIFIED — recorded but not confirmed; not relied on.
- Source hierarchy
- SEC filings and primary documents first, then credible third-party reporting, then company-reported figures, which are marked as such. Figures published by companies selling the solution being measured are labeled vendor-sourced and treated as directional at best. Two recorded cases — tax resolution firms and Wevorce — could not be verified and are not relied on anywhere in this artifact.
Known weaknesses
- Two of the most important patterns rest partly on company-reported figures: Empathy's growth, retention, and covered-lives numbers are company statements, and Progyny's multiples-rate comparison is company-reported.
- Evidence on discretionary referral is strongest in healthcare and benefits rather than commercial partnerships. The direction is consistent across the cases, but clean commercial before-and-after data remains scarce.
- No case produced public data on how a distributor measured referral contribution, so attribution — plausibly the thing that decides whether a channel survives its first budget review — could not be examined.
Building a distribution channel?
PastBehavior studies what has already happened in adjacent markets, identifies the routes with the strongest historical evidence, and helps companies turn that evidence into partner strategy and execution.
Discuss a distribution problem