# Round 1 Synthesis - Open Mom Test (12 interviews, simulated) *Study lead synthesis, 2026-08-07. Simulated personas: every number below is a directional prior, not validation. Round 1 of max 5. Not converged.* ## Headline finding Nobody in this roster is short of supplier or customer NAMES. All twelve can fill a funnel for free in an afternoon. The money is lost in three places downstream of discovery: (1) the qualification gap, where paper evidence lies and only floor-walked, audit-grade proof counts ($15.2k Dana, $19k Marcus, ~$38k Priya, $9k Greg); (2) speed and follow-through, where deals die because the seller side cannot turn quotes and drawings fast enough ($60k Kelly, $14k Jenny, $3k Tom); and (3) seller-gated access, where the evidence that matters is legally unreachable until the buyer is already $25k+ committed ($22.4k Derek, $400k Aisha). A "matching" pitch aims at the one step nobody is bleeding on. Every persona also priced the concept against a specific scar (ThomasNet, Xometry, a ghosting sourcing agent), which means the pricing model itself is read as a trust signal before the product is. ## Scorecard | # | Persona | Wedge | Fit | Pain | Concrete $ loss | WTP (structure) | Who pays | |---|---|---|---|---|---|---|---| | 01 | Dana, DTC hardware sourcing lead | W1 | strong | 4/5 | $15,200 (+~$40k soft) | $300-500/shortlist, refund if no sample order | Company card, CFO sign-off >$500 | | 02 | Marcus, OEM procurement | W1/W3 | medium | 3/5 | $19,000 expedite + $2.5k labor | $150-250/shortlist, castings only, audit-grade | Dept budget; QE has veto | | 03 | Rita, 8-person DTC (control) | W1 | weak | 2/5 | $6,200 (self-fixed for $150/order) | ~$0; maybe $20-40 one-time audit report | Owner, skeptical | | 04 | Tom, 63, CNC shop owner | W5/W4 | medium | 4/5 | $3,100 + $3,000 labor | $150-250 one-time flat; NO commission | Business account, his call | | 05 | Kelly, 34, fab shop successor | W5 | strong | 4/5 | $60,000 job lost on speed | $0 upfront, 8-10% on close, capped; NO flat fee | Company P&L, her call | | 06 | Hank, foundry sales manager | W3 supply | medium | 3/5 | ~$1,330/dead RFQ, 4-6/quarter | $300-400/mo IF it saves quoting hours | GM sign-off, not his card | | 07 | Priya, OEM sourcing engineer | W3 buy | medium | 4/5 | ~$30,000 tariff + $8k labor | $300-400/shortlist, under $2,500 card limit | Corporate card, no approval | | 08 | Sal, 61, manufacturer's rep (control) | W2 | no | 2/5 | $36,000 commission churn | $0, refuses to fund own disintermediation | n/a | | 09 | Jenny, trading co owner | W2/W1 | medium | 4/5 | $14,000 margin, dead deal | $150-250/dossier, $2-3k/mo cap; vetting only | Company, her call | | 10 | Derek, search fund principal | W4 | medium | 4/5 | $22,400 dead deal + $28k live | $7,500-15,000/deal IF pre-LOI signal | Fund diligence budget | | 11 | Aisha, PE roll-up associate | W4 | medium | 3/5 | $400,000 earnout dispute | $8-15k/deal, to $25k; she champions, MD signs | Deal budget | | 12 | Greg, EMS purchasing (control) | none | no | 2/5 | $9k + $40k + $12k | $0; "the buyer is a process, not a person" | Committee/ERP-gated | Roster shape check: 2 strong, 7 medium, 1 weak, 2 no. Controls behaved as designed (Rita: free tools win at small scale; Sal: hostile incumbent; Greg: no individual buyer exists). The study did not skew positive. ## Cross-cutting patterns **1. Discovery is free; qualification is where the bodies are buried.** Five personas independently said finding candidates is not the job. The expensive step is proving a candidate survives contact with reality (a quality audit, a first article, a first order). > "I can find ten foundries in an afternoon. Finding one my quality guy will > actually sign off on is the eight-week part." - Marcus **2. Paper evidence is already priced at zero. Only witnessed evidence counts.** Every buyer who lost money lost it to a supplier that looked fine on paper: Dana's factory "photographed beautifully," Priya's foundry claimed 40 tons and ran 22, Greg's foundry had no melt-source trail. "Verified" badges are assumed to be pay-to-play until proven otherwise. > "Heartland told us 40 tons, and PPAP told us the truth: they run 22 on a > good day." - Priya > "If you can't show me who actually walked the floor and when, you're just > Alibaba with a nicer font." - Dana **3. The pricing model is a scar map: each persona rejects whichever model burned them, and the scars point in opposite directions.** ThomasNet-burned (Kelly, Tom) refuse subscriptions and flat access fees. Xometry-burned (Tom) and rep-loyal (Hank) refuse commission to software. Agent-burned (Dana) demands a results guarantee on a flat fee. There is no single pricing model that clears the whole roster; pricing must be segmented by which incumbent burned the segment. > "If a percentage-of-revenue pitch walks in my door I'm already picturing > Xometry with a suit on." - Tom > "I'll pay you out of what you bring me. I will not pay you to find out if > you bring me anything." - Kelly **4. It is always company money, and the real price ceiling is the approval threshold, not the value of the loss.** Zero own-pocket WTP across all twelve. The binding constraint is procedural: Priya buys freely under her $2,500 card limit, Dana needs sign-off above $500, Hank needs Denise, Derek's committee triggers at $25k. Price the artifact under the buyer's no-approval line or inherit a second sale. > "Three hundred bucks on my p-card, sure, I could do that without asking > anyone. Three thousand and now I'm writing a justification memo to Dave." > - Priya **5. The relationship layer is explicitly off-limits to software, even among the friendliest buyers.** Jenny (the closest thing to a Phil analog in the roster) will pay for finding-and-vetting but hard-lines any agent speaking in her name. Sal treats the entire concept as an attack on his commission. Hank trusts Gary's gut over any badge. The trust layer cannot be automated head-on; it can only be fed better evidence. > "If your agent picks wrong and it's got my name on the email, that's not a > bad lead, that's my reputation with a factory I've known for a decade." > - Jenny > "I don't pay for leads. I make leads." - Sal **6. Every in-market search was triggered by a dated external shock, and the critical evidence was often externally discoverable in hindsight.** Tariff stacking triggered Dana, Marcus, Priya, Jenny, Greg. Succession events triggered Kelly, Derek, Aisha. Notably, both acquirer deal-killers (the retiring plant manager, Frank's fishing buddy in procurement) left traces outside the target: Derek caught his in a trade publication by accident. > "Nobody put those two things next to each other until I did, by accident, > reading a trade rag." - Derek > "Without the tariff, I wasn't touching this list." - Priya **7. Bus-factor-one is universal on the seller and buyer side alike.** Dana, Kelly, Priya, Jenny, Tom, and even Hank (via Gary) all admitted the whole motion stops if one person is out for a month. The workflow lives in one inbox, one Rolodex, one head. This is both the reason nothing digital plugs in easily and the raw material for any relationship-mapping product. ## Vetoes and dealbreakers (verbatim kill conditions) - Evidence that is repackaged self-reported data: instant kill for Dana, Marcus, Priya, Greg, Aisha ("a nicer PDF of the same self-reported capacity numbers... is garbage"). - Any agent contacting a customer or factory in the user's name: kill for Jenny; existential threat for Sal. - Commission or percentage-of-revenue pricing to shops burned by Xometry: kill for Tom; Hank will not pay software a percentage "the way I would to Gary - Gary's earned that, a tool hasn't." - Flat subscription to anyone burned by ThomasNet: kill for Kelly, Tom. - Tire-kicker leads: one bad first lead kills it for Tom and Hank. - No pre-LOI access to real customer signal: kill for Derek and Aisha; the seller gates access, and repackaged CIM data is worthless. - Liability ambiguity: Marcus opens with "who's liable when the evidence is wrong"; Aisha notes a wrong "low risk" report becomes evidence in an earnout dispute deposition. - No ERP/process integration path: structural kill at Greg-type companies (AS9100/ITAR flow-down cannot be shortcut). ## WTP read (with required skepticism) Simulated WTP is a directional prior at best, and simulated personas underprice friction and overprice their own rationality. What we should carry forward is the STRUCTURE, not the dollars: - Per-artifact flat fees clustered at $150-500 for sourcing buyers, always under an approval threshold, always conditional on evidence quality no vendor has yet demonstrated to them. Treat the numbers as ceilings. - Commission-only appetite exists exactly once (Kelly) and is capped, gated on a closed job in 90 days, and benchmarked against a $150/mo freelancer, not against ThomasNet. That benchmark is brutal and probably realistic. - Acquirer WTP ($8-25k/deal) is the largest and most credible band because it is priced against existing $55-120k diligence line items, but it is 100% conditional on solving an access problem both personas doubt any vendor can solve. - Every stated number came attached to a condition the product does not yet meet (refunds, audit-grade proof, pre-LOI access, demonstrated hour savings). Real-world WTP should be assumed to be materially lower until a human says a number with a card in hand. ## Wedge status after round 1 - **W1 Verified Shortlist: ALIVE, lead wedge.** Dana is the textbook buyer; Marcus, Priya, and Jenny all bid $150-400 conditionally. But the surviving version is narrower than pitched: the evidence bar is floor-walked, audit-adjacent proof (inspection reports, verified production history), not clever data fusion. That implies per-dossier COGS (third-party inspections run $150-600) that threaten margins at a $300 price point. The wedge lives or dies on whether remote evidence can clear the bar or whether inspections can be bundled profitably. - **W2 Agent-amplified rep desk: DEAD as scoped.** The hostile control (Sal) behaved as predicted, but the killing blow came from the friendly side: Jenny, the book-owning principal W2 needs, refuses to let agents front her name at any price. What survives of W2 is back-office grunt work under the human's byline, which is not a wedge, it is a feature of W1. - **W3 Castings/forgings desk: WOUNDED, redefinition required.** Buy side (Priya, Marcus) has real five-figure losses but the bottleneck is PPAP-stage qualification, which a shortlist does not touch. Supply side (Hank) explicitly does not want more leads; he wants bad RFQs triaged before they eat 14 engineer-hours. If W3 survives it is as a qualification-evidence and RFQ-triage desk, not a matcher. - **W4 Succession demand book: WOUNDED, one live pivot.** Both acquirers confirm the pain, the dollars, and the budget line, and both independently name the same structural wall: the seller controls customer access until it is too late. The live pivot is outside-in signal, since both cited deal-killers (a retirement, a personal relationship) that were in principle externally observable. If outside-in durability monitoring can be demonstrated, W4 reopens at the highest price point in the study. - **W5 Commission rep in software: WOUNDED, successor-segment only.** Kelly validates the pricing model but not the product: her lost $60k job died on quote/drawing follow-through speed, not lead volume, and she says the bottleneck out loud. Tom (the older-owner segment) rejects commission outright. W5 survives only as "sales back-office for digital-native successors," which shifts the work from finding buyers to accelerating quotes. ## What we still do not know 1. Frequency: how many shortlist-worthy searches per year do Dana, Priya, Marcus, and Jenny actually run? Per-artifact pricing collapses if the answer is 1-2. 2. Whether any remote/desk evidence package has EVER cleared a quality gatekeeper for these buyers, or whether a physical floor-walk is strictly required (decides W1 unit economics). 3. Whether Hank's dead-RFQ triage pain generalizes across foundries and what data a triage tool would need that Gary's gut currently provides. 4. Whether outside-in customer-durability signals (trade press, retirements, job postings, plant closures) would have actually caught Warrentown and Crescent, and what evidence format an IC or investor committee accepts. 5. Whether Kelly-type successors are a findable, reachable segment at scale, and what share of her 4-week Permian cycle was delegable work. 6. What refund/guarantee structure makes a flat-fee shortlist credible to an agent-burned buyer (Dana's "refund or credit if none of the three pan out" is a contract design question we have not explored). 7. Real WTP. Nothing in this round counts as WTP evidence; it is all prior. ## Convergence call NOT converged. Wedge ranking moved materially this round (W2 died, W3 and W5 redefined, W4 found a pivot), WTP is unbounded conditionality, and five named unknowns would each change the round 4 concept reveal. Round 2 proceeds: deepen the specific pains named above, per persona group, and quantify frequency and delegability.