ICP Discipline Before Scale: The WTP Discovery Filter

Decagon

L3

Decagon's founding was itself a sales process. Before co-founders Jesse Zhang and Ashwin Sreenivas wrote a single line of product code, they ran more than 100 structured discovery interviews with operations, support, and sales leaders at enterprise companies. The critical methodological choice: they asked directly "How much would you pay for this?" rather than "Is this interesting?" or "Would this be useful?" The softer questions generate enthusiasm without generating the WTP signal that actually predicts whether a sale will close. The discovery sprint tested multiple departments simultaneously. The contrast was stark. Sales and operations leaders gave conditional, delayed, low-dollar WTP: "Maybe I would pay a thousand dollars a month...maybe next quarter." Support leaders gave immediate, specific, large-dollar WTP: "People were like, yes, if you can deploy this thing, I will sign a $150,000 check immediately, right? And this happened repeatedly." (Sreenivas, PMF Show, January 2026.) Three signals confirmed simultaneously: (1) the budget authority was real — support leaders could sign without committee; (2) the pain was severe enough to prioritize now; (3) the deal size was enterprise-grade, not tooling-grade. The negative evidence was equally important. The $1,000/month "maybe next quarter" signal from other departments was not a product failure — it was an ICP boundary. Decagon could have entered a $500/month per-seat FAQ chatbot market if they had followed the enthusiastic but low-WTP signals. The discovery sprint identified the customer segment that would pay enterprise prices for enterprise automation before the company existed. This is the WTP discovery filter in its cleanest form: the question "how much would you pay?" asked directly and early is a qualification signal, not a pricing conversation. The discovery interviews also served as a soft pipeline. Some interviewees became early customers (Oura Ring, Eventbrite, HeartSpace as the first three bespoke implementations); others made introductions. By the time the MVP was built, Decagon had a pre-warm prospect list of buyers who had already told them they would sign a $150K check. The first sales cycle was not a persuasion exercise — it was a fulfillment of a commitment already made in the discovery phase. The discipline held at scale. Decagon's 4-week pilot structure (pricing agreed before the pilot starts, success metrics defined at kickoff, no post-pilot renegotiation) is the operational extension of the same filter: only buyers who have already mentally solved the budget equation enter the pilot. The Duolingo English Test case confirms the model: a prior vendor had failed to get chat live after a full year; Decagon went live in one month with 80% chat deflection. The ROI is visible in 4 weeks because the ICP was selected to produce visible ROI in 4 weeks.

Gong

L2

Gong's ICP discovery is one of the cleanest documented examples of using product data to debug a failing sales motion in real time. In the first 3–5 weeks of commercial selling, Amit Bendov and the team targeted sales operations people — a reasonable assumption about who would have budget authority for sales tooling. The response was consistent: "This is such a cool technology but we're busy right now, it's not in the budget." Heard 20–40 times in 3 weeks. The diagnostic mechanism was itself a product validation moment: Gong used its own call recordings to identify the pattern. What would normally take months of analysis collapsed to weeks. The diagnosis was clear: sales ops people did not have budget authority for the outcome Gong improved. The pivot target was CROs and VPs of Sales — people whose job title literally contained the word "revenue" and who owned the pipeline outcomes that Gong's data demonstrated. The category rename to "Revenue Intelligence" in October 2019 completed the ICP alignment: the product name matched the buyer's job title, making every piece of outbound content self-qualifying. Amit Bendov's hiring rule captures the ICP discipline in operational form: do not hire a VP Sales until 2 people are selling successfully. The validation period — founders running all sales personally until repeatability is demonstrated — is the ICP discovery phase in practice. When he hired Jameson Yung as SVP Sales, Yung's first action was to raise quotas and restore the pilot motion (Gong had moved away from pilots; Yung recognized pilots as a core conversion driver). The ICP had been validated; the infrastructure could now be built. The alpha program itself was a WTP discovery sprint in product form. Eilon Reshef recruited 12 companies from the founders' personal network. At trial end, Amit ran a "trial close" — directly asked each to pay. Eleven of twelve agreed immediately. Eilon's PMF signal was behavioral, not attitudinal: "9 out of 10 complaints were how come you didn't even record this call?" — users were angry when a call was not captured, meaning the product had already become essential before the commercial conversation began. True WTP is expressed as outrage at absence, not enthusiasm at presence. Gong found it in the alpha program.

Harvey

L2

Harvey's ICP discovery operated at two levels simultaneously: individual lawyers and law firms. The insight that took the founders the longest to articulate explicitly — and which Winston Weinberg only described in full in January 2026 — is that these were two different buyer personas with two different pain points, requiring two different pitches. The individual lawyer ICP was discovered first, through founder-run sales at scale. Weinberg sent tens of thousands of LinkedIn messages personally. The PACER demo tactic was the WTP filter in action: "I would basically download the last thing that they submitted to court. And then I would try to come up with prompts that were like, 'This is bad.' And because they're a litigator and I'm basically attacking something that they just wrote — they would instantly read the screen." (Weinberg, Long Strange Trip, January 2026.) The lawyers who responded intensely to this — who leaned in, who debated the output — were the ICP. Those who were politely interested were not. The individual lawyer pain was precise: "I don't want to do this particular piece of my job. Like, I don't want to go through tens of thousands of documents and do this massive closing checklist. I don't want to do that. That's not the fun part about being a lawyer." This is WTP language — not feature enthusiasm, but relief from a specific, named, ongoing burden. Harvey's B2C2B motion emerged from this discovery: sell to individual lawyers on personal pain, let them demonstrate value inside their firms, then convert the firm. The wedge into Allen & Overy was not a firm-level pitch — it was individual lawyers already using Harvey who created pull from inside. The corporate in-house legal ICP was discovered later, and it has a distinct WTP profile. Corporate legal teams do not perceive AI as threatening to their role (legal is not the corporation's core competency), so adoption is seen as operational efficiency rather than workforce substitution. By Q4 2025, 42% of Harvey's revenue came from Fortune 100 corporates — a composition shift that was only possible because the individual lawyer ICP discovery had produced the law firm relationships that introduced Harvey to corporate clients. ICP discovery does not end at launch; it continues as a production system.

Sierra

L3

Sierra's ICP discovery mechanism was structurally embedded in the design partner program itself rather than in a pre-product interview sprint. The mechanism was the 10–20% TCV payment required upfront — a filter that forced prospects to demonstrate budget authority and urgency before any product was built. This is a WTP filter in commercial form: it does not ask "how much would you pay?" It requires payment before answering that question. Logan Randolph's explicit framing of the filter: "In some conversations, it became clear that prospective partners were interested in 'playing with AI' and building internal prototypes." These companies were screened out. The ICP was operationally defined as companies with acute CX strain that could demonstrate financial commitment to solving it. The resulting design partner cohort (WeightWatchers, SiriusXM, Sonos, OluKai, and two others) shared three properties: annual revenue above $1B, a support operation under measurable cost pressure, and a decision-maker with authority to sign a contract before the product existed. The design partner program ran a 4-step selection process: discovery (30 minutes, no selling) → live demo → security and technical deep dives → verbal commitment plus paid contract. This sequence is itself an ICP discovery funnel. The discovery conversation surfaced whether the problem was real and urgent. The live demo tested whether the buyer could see the solution concretely. The security review identified whether enterprise procurement gates were navigable. The paid commitment confirmed WTP. Companies that failed any step were disqualified, not pursued. The outcome of the ICP discipline: 100% conversion from design partner to paying customer. WeightWatchers achieved 70% containment rate and 4.5/5 CSAT in the first week — a result that is only possible when the ICP is correctly identified and the product is correctly scoped. Sierra launched publicly on February 13, 2024, with proof already in hand from recognizable brands. The post-launch sales motion (Reggie Marable's PEER methodology and Paid PoC structure) was a scaled version of the same design partner discipline: select for WTP, structure for proof, convert through demonstrated outcomes rather than persuasion.

Legora

L3

Max Junestrand's ICP discovery tactic is one of the most elegantly structured in this cohort because it solved two problems simultaneously. He cold-DMed lawyers on LinkedIn offering to pay them at their own hourly billing rate for 30-minute discovery interviews. This accomplished two things: it demonstrated respect for time-as-money (the core value metric in legal), generating goodwill that cold outbound from an unknown founder could not generate; and it pre-qualified respondents, because a lawyer who responds to a paid-interview offer is a lawyer who takes the problem seriously enough to invest time in discussing it. The result: 60+ structured interviews in weeks, before any product existed. The WTP signal Legora found was framed in revenue terms, not cost terms: "$5M potential additional billing per 100 lawyers." This is the correct WTP language for legal: lawyers are not trying to reduce costs, they are trying to capture more billable hours without burning out. A product that recovers 30% of time lost to rote work is a product that converts unbillable time to billable time. This framing — revenue recovery rather than efficiency improvement — positioned Legora's ICP as any firm where rote work was eroding billable capacity. The live demo at a major Swedish legal conference validated the ICP at scale in a single event. While every other startup showed slides, Legora ran a live product demonstration. The result: approximately 150 demo requests from one event. This is a behavioral WTP signal: lawyers who actively request a demo after seeing a live product are expressing purchase intent, not curiosity. The conference event produced the first clear evidence that the ICP was real at population scale in the Scandinavian legal market. Mannheimer Swartling's design partnership — Sweden's most prestigious law firm — then converted the ICP discovery into a distribution mechanism. Jonathan Rintala, Legora's early GTM lead, described the market effect: "Suddenly, every law firm wanted to talk. Lawyers are quite FOMO-driven, similar to VCs and other high-status industries." The prestige anchor created a self-reinforcing ICP signal: if the most demanding buyer in the market is using Legora, every firm in the peer network treats it as confirmed WTP evidence. Legora reached $100M ARR in 18 months from founding — the fastest enterprise business to reach that milestone per Bessemer Venture Partners.