Depth bets that one person who knows a niche's record systems will screen faster and ask better. Breadth bets that the best candidates are scattered across many books and a short briefing is enough to spot them. Both can be true in different corners of one firm, so the test gives the arms one shared block and two zones only one arm can reach.
Who the client is, and what either model asks your people to do
The client problem is the same under both models. An established company may hold years of records it created itself, such as dispatch exceptions, project histories or support tickets, that model developers can use for training and evaluation under a direct license, and most such companies have never heard of the option. Your firm's part is narrow: recognize a likely fit, ask the CEO, introduce the company if the CEO wants that, describe the program accurately and step back while SourceX runs its review. Nobody at the firm requests files, samples or logins.
| Dimension | Specialist depth | Broad recruitment |
|---|---|---|
| Who spots candidates | One named person in one niche | Ten to fifteen client leads, each in their own book |
| What they know | The niche's systems, record types and usual contract limits | Their clients well; the program only from a briefing |
| Who asks the CEO | The relationship owner, after the specialist's fit check | The client lead who owns the relationship |
| Set-up cost | Time to build one niche screen | A briefing per recruit, plus a coordinator |
| Typical failure | The niche runs out of companies that fit | Colleagues name the same company, or oversell the program |
Large partner programs meet the same fork. Microsoft's Partner Center documentation describes two ways to handle Joint Planning leads: a designated referrals admin who receives every lead and updates its status, or partner sellers who work the leads they are tagged on while an admin takes the rest. How niche knowledge surfaces candidates filed under other industries is shown in the sector specialist method and, for accountants, the niche practice guide.
Recruiting outside professionals instead of colleagues is a different proposition: each needs their own partner account and signed SourceX Channel Partner Agreement, and the terms describe no share for whoever recruited another partner. It can serve a professional referral group; under the published terms it is not income for your firm.
How the program's rules tilt the comparison
Four rules in the terms bear harder on one arm than the other.
- First valid introduction, one partner. Within the 90-day attribution window, the first valid introduction is credited once the company submits a verified application. In a breadth arm two colleagues can name one company in the same week, so decide first whether the firm registers as one partner or each adviser separately; that turns a repeat into either a coordination slip or a contest between your own people. Any internal reward split belongs in your own written agreement.
- Earlier contact counts against everyone. Section 3 treats any earlier lead, account or email correspondence between SourceX and a company as an existing relationship, not credited to a later partner. You learn this only after introducing, and a wider net finds more such companies. Count them as a cost of breadth.
- Corporate groups. A parent and its subsidiaries count as one referred company: the USD 50,000 floor applies once, to the group's first qualifying transaction, and later qualifying transactions first agreed within the 12-month Repeat Window pay 15%. No published rule settles credit when a sister company came in through another partner, or when the parent once dealt with SourceX; SourceX resolves such cases one by one in duplicate review. Log the group link beside the name and leave the outcome to that review.
- Dated records decide. Attribution rests on dated evidence such as written introductions; a link click alone does not decide credit. One shared, dated name register is the backbone of the test, and the pre-introduction checklist covers the 48 hours before a name goes out.
The depth-versus-breadth experiment plan
Fix every element below before day one, in writing. One idea comes from experimental statistics: the NIST/SEMATECH handbook explains that blocking removes the effect of the most important nuisance variables and randomization reduces the contaminating effect of the rest. Here the biggest nuisance variable is the niche. A specialist in their own niche against generalists everywhere else compares niches, not models, so the one niche both arms can work becomes a block, split at random.
| Element | Fixed before day one | Why it matters |
|---|---|---|
| 1. Question | Per 10 hours, which arm yields more CEOs who want a conversation and more verified applications, with fewer collisions and incidents? | Answerable with counts |
| 2. Specialist arm | One named person, one niche the firm already advises in | Depth needs a single owner |
| 3. Breadth arm | 10 to 15 client leads, one 30-minute briefing each, at most two names per lead, one coordinator | Caps the noise a broad call creates |
| 4. Hours | The same budget per arm, for example 40 hours, with briefing and coordination time charged to breadth | Unequal hours decide the result in advance |
| 5. Shared block | List the niche accounts both arms could work and flip a coin for each: heads to the specialist, tails to its own client lead | The only fair comparison in the test |
| 6. Outer zones | The specialist's own network outside the client book; breadth leads' clients outside the niche | Reported as found, never compared |
| 7. Name register | One dated register; every name entered before any ask; the first entry keeps it; a repeat is logged as a collision | Mirrors how attribution is decided |
| 8. Window | 90 days to match the attribution window; read-out on day 90; a second look on day 180 under the same rules | Later stages outrun the test |
| 9. Counts | Names, names passing the four fit facts on metadata, CEOs asked, CEO yeses, verified applications, companies SourceX already knew, collisions, incidents, hours by task | Each count keeps its denominator |
| 10. Client answer | One question to every CEO asked: was this worth raising with you? | Usefulness, judged by the client |
| 11. Stop rule | A request for client records, or a CEO approached without the relationship owner's agreement, pauses the arm where it happened | Clients come before the data |
Money stays out on purpose. No reward is payable until handoff is complete and the Acquisition Partner has issued its final written acceptance of the complete, validated dataset, and the deadline sits 60 calendar days after whichever of those came later; SourceX's published material contains no conversion rates or payout averages either. Neither arm will have a reward by day 90. Also write down the smallest result you will act on: with fewer than five CEO yeses across both arms, call it inconclusive, because three against one is an anecdote.
A 90-day read-out from one firm (Illustrative)
Illustrative: an invented 60-person advisory firm with a transport and logistics practice. Its block held 12 logistics client accounts, six to each arm by coin flip. The specialist also named three logistics companies from their own network, and twelve client leads covered the rest of the book.
| Count at day 90 | Specialist arm | Breadth arm, block half | Breadth arm, rest of book |
|---|---|---|---|
| Companies named | 9 (6 block, 3 own network) | 6 | 23 |
| Passing all four fit facts on metadata | 6 | 2 | 7 |
| CEOs asked | 5 | 2 | 6 |
| CEOs who wanted the conversation | 3 (2 from the block) | 1 | 2 |
| Verified applications | 2 | 0 | 1 |
| Already known to SourceX | 0 | 0 | 3 |
| Collisions in the register | 0 | 1 | 4 |
| Incidents | 0 | 0 | 1: a sample export was requested and stopped before anything was sent |
| CEOs who said it was worth raising | 4 of 5 | 1 of 2 | 3 of 6 |
| Hours logged | 38 | 14 | 27 |
Inside the block, two CEO yeses from the specialist's six accounts against one from the breadth half's six is inconclusive under the firm's own rule. Across whole arms, efficiency is nearly level: 38 hours for three yeses against 41 hours for three. The rest separates them. Breadth brought five collisions, three companies SourceX already knew and one incident, though its two yeses outside the niche are companies the specialist could never have reached. The pre-written tie rule favored fewer incidents and collisions, so the firm kept the specialist and turned breadth into a referral-in route: client leads pass names to the specialist, who checks fit at metadata level before the relationship owner asks.
Effort, terms, trust and money in each arm
| Dimension | Specialist depth | Broad recruitment |
|---|---|---|
| Effort | One person's hours and growing skill; the niche sets the ceiling | Briefings, a coordinator and follow-up on uneven names |
| Terms | Channel Partner Agreement signed before submitting; one partner per company | The same, plus a shared register and an internal agreement on credit |
| Trust | One accurate voice the niche learns to recognize | Many voices, any of which may tell a client "you qualify" before any review |
| Money | Unforecastable; no cap, and the USD 50,000 floor once per company | Same formula; more companies at most, never a richer rate |
More recruits cannot make any single introduction worth more. Each company's default reward on its first qualifying transaction is the greater of $50,000 plus 5% and 15% of Actual Supplier Payout, with no cap; the USD 50,000 floor applies once per company, and further qualifying transactions first agreed within 12 months of the first Closing pay 15%. Illustrative: a $250,000 first-transaction payout produces a $62,500 reward, a $400,000 payout $70,000 and an $800,000 payout $120,000. Breadth pays only if more companies pass the fit facts, which do not loosen as recruits are added.
Coordination is the cost that surprises firms. In PartnerStack and Wynter's survey of 100 senior leaders at B2B SaaS companies with $50 million or more in revenue, charted in September 2025, alignment between teams was the most common blocker to partner-driven revenue, at 37%. Those are software vendors, not advisory firms, but a broad arm is an alignment exercise by design; the boundary role-plays make a reasonable briefing kit.
Which companies either arm may put forward
Both arms screen against the same published facts, at metadata level only: which records exist, roughly how many, over how many years, in which system and owned by whom. A company meets the baseline when it reached 50 or more full-time employees at peak, contractors excluded; keeps documented operating history it created itself; holds the rights to license it; and has a CEO who wants the conversation. The CEO authorizes a licensing discussion and signs; a controller who owns the records can help build the case but does not replace that yes. 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. For a staffing firm, only its own full-time employees count, not workers placed with clients; give both numbers and introduce the firm anyway when its records are strong.
Neither arm can know before SourceX's review whether a company's records match a current program, whether customer contracts limit reuse or what a license would cover; a company that misses one program may fit another later. Records that belong to a client's customers, scraped data and introductions the decision maker does not know about are usually not a fit. Firms tempted to run licensing work themselves should first read refer or build a licensing practice.
Pilot, maintain, expand or decline: the rules to sign on day one
| Day-90 result | Decision | Reason to record |
|---|---|---|
| Fewer than five CEO yeses across both arms | Pilot again with a longer window, or keep introductions occasional | The test could not tell the models apart |
| Specialist hours per CEO yes at least a quarter lower, with no incidents | Maintain depth and stop recruiting broadly | Niche knowledge is doing the work |
| Breadth hours per CEO yes at least a quarter lower, no record incidents, collisions under one in five names | Expand broad recruitment and keep the central register | The firm's fits are scattered, and a briefing was enough |
| The two arms within a quarter of each other | Keep the arm with fewer incidents and collisions; consider breadth naming and the specialist screening | Efficiency is a tie, so risk decides |
| Most CEOs asked say it was not worth raising | Decline both | The client problem is not present in this book |
| A record request or an unapproved CEO approach happens twice | Stop that arm | These people cannot run this model safely |
The one-quarter margin and the one-in-five collision ceiling are our choices, not published benchmarks. Set your own before day one and hold to them when the result disappoints.
Four limits belong in the read-out itself:
- Small numbers and one person. A few companies decide everything, and a strong specialist result may belong to an individual who can leave. Repeat before generalizing.
- Niche saturation. Depth can look strong for 90 days and run dry by month nine; count the niche's remaining fits at the end.
- Late duplicates. You learn that SourceX already knew a company only after introducing it, so breadth's collision count runs low until review catches up.
- Moving demand and a watched test. SourceX checks relevance to current programs in review, and people in a visible test work harder than usual, in both arms.
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.
Once the plan is written and both arms know their hours, register for a partner link, check each name against who qualifies and read the attribution and duplicate sections of the program terms before the first CEO is asked. More partner decisions are collected in the guides hub.