# Round 2 Synthesis - Deepen, Quantify, Frequency (12 interviews, simulated) *Study lead synthesis, 2026-08-07. Simulated personas: every number below is a directional prior, not validation. Round 2 of max 5. Not converged.* ## Headline finding Every persona with real pain has already built the product, badly, with their own money, and the price they actually paid is the true WTP anchor - and it is far below what they said in round 1. Dana pays a freelance inspector $380-450 a visit. Marcus pays a metrology lab $900 a report. Jenny pays VeriFact $450 on every single factory now, $9-11k a year. Rita's benchmark is $150 and five days. Kelly bought a $240/yr Bluebeam seat at eleven at night. Derek pays $516 a year in trade-press subscriptions and does the rest in unbilled hours. Aisha asked her MD twice for a $5,000 line and got told no, so her price is zero plus Saturdays. The round's second finding partially reverses round 1: it is NOT true that only a floor-walk changes decisions. Independent, third-party evidence generated on a PHYSICAL artifact has now flipped decisions three times with nobody flying anywhere (Marcus's Buckeye CMM report, Rita's Insight inspection PDF, Priya's registrar-verified cert on a low-criticality bracket). The bar is independence plus physicality plus criticality-tiering, not presence. But the floor still catches what nothing else can (Buckeye's undisclosed plating subcontractor, Kevin catching tooling wear, Jenny's rented showroom). The wedge that survives round 2 is not a shortlist and not matching: it is brokering, scheduling, and standardizing the independent physical verification these buyers are already paying incumbents for, one Venmo transaction at a time - and it must beat a guy named Kevin who answers texts within the hour. ## Scorecard | # | Persona | Wedge | Fit | Pain | New $ data this round | Self-fix already bought (real anchor) | 24-mo frequency | |---|---|---|---|---|---|---|---| | 01 | Dana, DTC hardware sourcing lead | W1 | strong | 4/5 | +$3,500 Canton never refunded; $1,900-2,250 trailing inspection spend | Kevin Zhou, $380-450/visit, mandatory over $2k deposit, 4x since June 2025 | 4 searches | | 02 | Marcus, OEM procurement | W1/W3 | medium | 3/5 | $900/report x 4 = $3,600 this year, self-instituted | Independent CMM lab report before any PPAP sign-off | 3 searches (2 completed) | | 03 | Rita, 8-person DTC (control) | W1 | no | 2/5 | $470 total inspection spend in 24 months; $9k exposure killed by a $150 report | $150 inspection rule, never skip again | 4 searches (~1.5/yr) | | 04 | Tom, 63, CNC shop owner | W5/W4 | medium | 4/5 | Great Lakes = $420k/yr (32% of revenue); buyer retiring "in a year or two" | $750 non-refundable quote deposit, invoked 2x, ~$1,500 protected | 3 customer chases | | 05 | Kelly, 34, fab shop successor | W5 | strong | 4/5 | Trinity near-miss: ~$1,800 wrong steel from delegated takeoff error | $240/yr Bluebeam + Chris's $150/mo triage script (worked) | 9 chases in 14 months | | 06 | Hank, foundry sales manager | W3 supply | medium | 3/5 | $5,320/quarter in Deb's dead-RFQ hours (56 hrs x $95), first time summed | None bought; asked Denise for headcount, told "show me the number," never went back | 5 dead RFQs in Q2 alone | | 07 | Priya, OEM sourcing engineer | W3 buy | medium | 4/5 | $18,000/quarter live PTO-housing tariff bleed until Oct 1 qualification | Free registrar call + own-picked references, ~2-3 hrs/candidate, $0 | 2 searches | | 08 | Sal, 61, manufacturer's rep (control) | W2 | no | 2/5 | Fairview save ~$22k/yr account preserved via 8-month human tip | $800-1,000/yr in lunches for the number-two at every account | 5 new accounts | | 09 | Jenny, trading co owner | W2/W1 | medium | 4/5 | $2,150 Litong cash loss; $2,400 + $2,700 tariff re-qual sprints | VeriFact now on EVERY factory, no size exception, $9,000-10,800/yr run rate | 22-24 qualifications | | 10 | Derek, search fund principal | W4 | medium | 4/5 | Higgins now $34,500 committed; durability work still $0 billed | $516/yr subscriptions + 3-4 hrs/wk DIY monitoring (~$2,240-3,400/deal) | 2 deep deals + 3 light passes | | 11 | Aisha, PE roll-up associate | W4 | medium | 3/5 | Briarwood: $18k deposit + $40k QoE sunk, killed by her own free Googling | $0 - MD declined a $5k/deal budget twice; unpaid overtime | 4 deals through diligence | | 12 | Greg, EMS purchasing (control) | none | no | 2/5 | None new; $40k hit confirmed soft (PM estimate, not invoice) | Free one-line traceability ask in Denise's RFQ template | 2 searches | Roster shape check: 2 strong, 7 medium, 3 no. Controls behaved as designed and are now essentially spent (Rita's benchmark bounded, Sal's principal-mandate path tested and dead, Greg's loss confirmed ownerless). Study still not skewing positive. ## Cross-cutting patterns **1. The self-fix price is the real WTP, and it is a fraction of the stated number.** Ten of twelve responded to their burn with a purchased or procedural fix, and the prices are now on the table: $150 (Rita), $240/yr (Kelly), $380-450 (Dana), $450 (Jenny), $516/yr + own hours (Derek), $750 (Tom), $900 (Marcus), $0 (Priya, Aisha, Greg). Round 1 stated numbers ($8-15k/deal, $10-15k/deal, $300-500/shortlist) should be discounted toward these. Tom is the one inversion: his revealed price ($750 to protect Dave's quoting hours) is HIGHER than his stated $150-250, which says owners underquote vendors and overpay themselves. > "That saved me about nine grand, and it cost me a hundred fifty bucks and > five days. Beat that number and I'll listen. I don't think you can." - Rita **2. Independent + physical evidence DOES clear the bar without a floor walk - round 1's absolutism was wrong.** Three decisions flipped on documents alone this round, and all three documents shared the same properties: generated by a party with zero stake, on a physical artifact (a sample part, a production batch, a registrar's live database), and matched to the criticality of the part. Vendor self-reported paper stayed at zero for the twelfth straight persona ("A supplier's own capability study has never once changed Dave's mind. An independent lab's report on an actual part has, twice." - Marcus). But the floor visit still finds the category of fraud nothing else touches: subcontracting, showroom fronts, tooling wear. > "Dutchman's bracket got approved off a cert and a registrar confirmation > alone, no site visit, because Renee doesn't treat a cosmetic trim bracket > the way she treats a gearbox housing. Not every part needs the same bar." > - Priya **3. Frequency splits the roster into two different markets.** High cadence: Jenny (22-24 qualifications/24mo), Kelly (one chase every 6 weeks), Hank (5 dead RFQs a quarter), Dana (4-6 inspection events). Low cadence: everyone else runs 2-4 real events in 24 months. Per-artifact products only compound for the first group; for Priya, Marcus, Derek, and Aisha, even a perfect product sells 2-3 units a year per account. > "Two real searches in two years, plus this PTO thing that's still open. > That's not a lot of shots for anyone trying to sell me something per > search." - Priya **4. Nobody fights for the money, and nobody walks the number upstairs.** Dana never escalated the Canton refund past one email. Hank computed $5,320 a quarter live in the interview and still has not taken it to Denise. Aisha absorbed two burns as overtime after her MD said no twice. Derek will not even file a $516 expense report. The losses are eaten quietly and converted into private rules. Implication: there is no internal champion motion here; the only viable buyer is the person whose card already moves at a price under their approval line (Dana, Jenny, Priya under $2,500). > "Nobody told me to add that number up. You're the first person who's made > me do the math out loud... Now I have the number. I still haven't gone > back to her." - Hank **5. The delegation boundary is now mapped: triage yes, judgment no.** Kelly ran the natural experiment. Handing phone-lead triage to Chris via a script worked and saves ~5 hrs/month; handing a technical takeoff to her foreman nearly ordered $1,800 of wrong steel and she has not delegated one since. Jenny independently drew the same 40/60 line: paperwork, scheduling, cert cross-checks are delegable; reference calls and the walkthrough judgment are not, because "they do it because it's me asking." Any agent product lives strictly on the pre-qualification side of that line. > "Chris screening the phone leads saved me real hours. Nobody else I've > tried to hand anything to has saved me anything, they've cost me a night > double-checking their work." - Kelly **6. The outside-in signal thesis got its strongest evidence yet - and its weakest payer evidence.** Warrentown's killer was in a public 8-K and a trade journal six weeks early. Aisha found the fishing-buddy tie in 40 minutes across LinkedIn, a newspaper walleye-tournament photo, and a Rotary roster PDF, and found Briarwood's brother-in-law in a wedding announcement. Derek is running the product by hand right now (LinkedIn title watches, WARN lists, news alerts) and has one live flag he cannot resolve. But the same round showed no one pays: Ansel refused the scope, the MD refused the budget, and Derek prices his own labor at $2,240-3,400 per deal. > "It's not that the information doesn't exist, it's that nobody's job is to > go get it before it's too late." - Aisha > "Nobody sends me an invoice that says 'customer durability check, $4,000.' > It's zero dollars until it's twenty-two thousand dollars." - Derek ## Vetoes and dealbreakers (added or hardened this round) - Beat-the-incumbent-benchmark or silence: Rita's $150/4-5 days, Dana's Kevin ($380-450, answers a text in an hour), Jenny's VeriFact ($450), Marcus's Precision Metrology ($900). These are now the comparison set, not ThomasNet. A product priced above them must catch something they miss. - A flag you cannot resolve is a liability, not a product. Derek has a LinkedIn title change he cannot act on ("a flag and no way to resolve it, which is its own kind of frustrating"); Sal says a probabilistic retirement flag "stops working" the moment it is a screen instead of a person. Signal without a sanctioned next step will churn. - Vendor self-reported anything: still an instant kill, twelve for twelve. - The relationship layer stays off-limits: Jenny's 60% (reference calls, walkthrough judgment) is non-delegable even to her own trained assistant; Sal re-confirmed hostile. Unchanged from R1 and now quantified. - Owners will not document their book pre-sale: Tom trashed the broker letter unread and has never written down the story behind 80% of his revenue. W4 has no seller-side entry; it is buyer-side only. - Do not sell aggregation of diffuse losses to the person who will not carry the number upstairs (Hank). The $20k/yr is real and organizationally invisible by choice. ## WTP read (with required skepticism) Simulated WTP remains a directional prior; this round replaces stated numbers with revealed ones, which are better priors but still simulated. - The revealed per-event band for verification work is $150-900, paid on company cards, under approval thresholds, to incumbents who earned trust through one catch. Round 1's $300-500 shortlist band sits inside this only if the product replaces the inspection, not if it stacks on top. - Jenny is the only compounding account: $9,000-10,800/yr run rate on audits at current cadence, with a stated desire to push spend down or coverage up. That is the single best revenue shape in the roster. - Acquirer WTP has collapsed from R1's $8-25k headline to a $2,240-3,400 DIY-labor anchor (Derek) and a twice-refused $5,000 budget line (Aisha). The R1 band was "save the deal" framing; the isolated diagnostic work is priced at an analyst-week. Assume low four figures per deal, not five. - Kelly's commission structure survives but the delegable scope shrank to triage; what remains commissionable is thin until R3 tests it. - Real WTP still unknown. Nothing here is a purchase; it is all prior. ## Wedge status after round 2 - **W1 Verified Shortlist: WOUNDED, redefined again.** The "shortlist" half is dead (discovery confirmed free, twelve for twelve); the surviving job is verification orchestration - arranging independent physical evidence (source inspections, sample CMM, registrar checks) faster, cheaper, or in geographies the buyer's incumbent cannot reach. It now competes with named incumbents (Kevin, VeriFact, Insight, Precision Metrology), not with nothing. Alive mainly on Dana (coverage risk: one freelancer, no backup) and Jenny (spend-compression on $10k/yr). Frequency caps it for Marcus and Priya at 2-3 units a year. - **W2 Agent-amplified rep desk: DEAD, confirmed.** The principal-mandate path was the last live route and Corland's $15k/yr BDR experiment produced zero pipeline through three reps before quietly lapsing. Control spent. - **W3 Castings/forgings desk: WOUNDED, trending dead as a standalone.** Buy side (Priya, Marcus) has fully merged into the W1 verification story - same job, castings-flavored. Supply side triage now has a number ($5,320/qtr) but its owner will not spend it upstairs, the two killer day-one questions are a free form field, and the shopped-print signal lives in Gary's peer network, which no tool can ingest. One more round without a payer and W3 folds into W1. - **W4 Succession demand book: WOUNDED, evidence up, price down.** Outside-in discoverability is now proven three ways (8-K, wedding announcement, Rotary roster) and Derek is hand-running the exact product weekly. But there is no existing line item to displace ("I'd be creating a new one"), the two people closest to paying were refused or refuse to ask, and the honest anchor is analyst-week pricing. Seller-side entry is confirmed closed (Tom). Survives as a buyer-side, low-four-figures diligence artifact pending R3 behavior data. - **W5 Commission rep in software: WOUNDED, narrowed to pre-qual triage.** Kelly's experiment split the job: triage delegation works, technical follow-through delegation does not, and she self-fixed the takeoff bottleneck for $240/yr. The 11 dead days on Permian were buyer-side waiting no vendor touches. What is left to commission is qualifying and warming inbound before it reaches her - real but much thinner than R1's "sales back-office." Tom's segment confirmed unreachable for commission pricing (unchanged). ## What we still do not know 1. Where the incumbent workarounds FAIL: Has Kevin ever been unavailable, wrong, or outside his geography (Dana is sourcing Vietnam, Mexico, Ohio - Kevin covers Chinese provinces)? What does VeriFact structurally not check that has cost Jenny? Precision Metrology missed Buckeye's plating subcontractor - did Marcus notice that gap? The wedge lives exactly in the incumbents' failure modes, and we have not mapped them. 2. Whether Jenny's audit spend is compressible or extendable: would she pay the same $450 for broader coverage, or only less for the same? 3. Whether any budget owner above Derek/Aisha would fund outside-in monitoring if the artifact were IC-memo-ready and deposition-safe (the Crescent language weaponization suggests a legal-defensibility angle no one has priced). 4. Whether Kelly would pay Chris-plus rates for expanded pre-qual triage, and what share of her nine chases the Chris script could have owned. 5. Whether Hank's $5,320 number ever moves Denise, and whether the two day-one questions get adopted as a form (free) or need enforcement (product). R3 should just ask what he did after this interview. 6. Whether Tom does anything at all about Ron's retirement in the next round; his behavior, not his words, decides if succession risk is sellable to owners in any form. 7. Real WTP. Still zero purchase evidence; all numbers remain priors. ## Convergence call NOT converged. The wedge board moved again (W1 redefined around verification orchestration, W3 trending into W1, W4 re-priced an order of magnitude down, W5 narrowed to triage), and the surviving wedges all now depend on unmapped incumbent failure modes and unresolved payer questions that R3's cadence-and-behavior arc is designed to answer. Round 3 proceeds.