The Right Way to Do Customer Discovery During an Accelerator: A Practical Framework from Interview to Conversion

The first quarter of 2025 has exposed a brutal asymmetry in the accelerator landscape. While Y Combinator’s W25 batch saw a record 28,000 applications for roughly 250 slots — a 0.89% acceptance rate — the more telling statistic is the cohort’s post-demo-day conversion rate. According to PitchBook’s Q1 2025 US Venture Monitor, only 34% of YC W24 graduates had closed a priced round within six months of demo day, down from 48% for the W22 cohort. For the hundreds of corporate and university-affiliated accelerators in Asia, the figure is materially worse. The root cause is not a lack of founder quality or product-market fit, but a systematic failure in customer discovery during the programme itself. Accelerator timelines compress the classic “get out of the building” process into 8-14 weeks, forcing founders into a choice between speed and rigour. The 2025 data shows that choosing speed without a structured interview-to-conversion framework is a direct path to a flat demo day. This article provides that framework, grounded in the mechanics of how early-stage B2B and B2C ventures in Hong Kong, Singapore, and Taipei can convert discovery conversations into verifiable revenue signals before the programme ends.

The Structural Problem: Accelerator Timelines vs. Discovery Depth

The fundamental tension in any accelerator is the mismatch between the programme’s fixed duration and the open-ended nature of customer discovery. A standard 12-week accelerator allocates roughly 8 weeks for customer development before the first demo day rehearsal. This is insufficient for a founder to conduct the 30-40 structured interviews that the lean startup methodology prescribes, let alone iterate on a prototype based on those interviews.

The 10-interview trap. Data from the 2024 Global Accelerator Report, published by the International Business Innovation Association (InBIA), indicates that the median number of customer interviews completed by accelerator graduates before demo day is 12. This is below the statistically significant threshold of 20 interviews identified in the 2020 study “The Impact of Customer Discovery on Startup Survival” published in the Journal of Business Venturing (Vol. 35, No. 4). Below 20 interviews, the probability of a false positive — a customer saying “yes” to a concept but never paying — exceeds 60%. The trap is that founders mistake conversational warmth for purchase intent. A founder who has done 12 interviews and heard “that’s interesting” from 10 of them believes they have validated the problem. They have validated only that people are polite.

The conversion lag in B2B cycles. For B2B startups, the problem is compounded by procurement cycles. A typical enterprise sales cycle in Hong Kong or Singapore, even for a SaaS product under HKD 200,000 per annum, is 8-12 weeks from first meeting to signed contract. An accelerator that starts on week 1 and expects a signed LOI by week 12 is asking a founder to compress a 12-week sales cycle into 8 weeks of discovery and 4 weeks of closing. This is mechanically impossible unless the founder has pre-existing relationships or the product is priced at a frictionless level — typically under HKD 10,000 per annum, which itself signals a lack of enterprise readiness. The 2024 HKEX consultation paper on the new Chapter 18C listing rules for specialist technology companies explicitly noted that one of the key failure modes for pre-revenue biotech and deep-tech applicants was an inability to demonstrate “meaningful third-party validation” — defined as binding commercial agreements, not letters of intent. The SFC’s 2023 “Thematic Inspection of Sponsor Work on Pre-IPO Investments” (published December 2023) reinforced this, stating that sponsors must verify that “customer discovery processes were documented and that the resulting commercial arrangements were legally enforceable.” A polite conversation is not a commercial arrangement.

A Practical Framework: The Interview-to-Conversion Pipeline

The solution is to treat customer discovery not as a research exercise but as a structured sales pipeline with specific conversion gates. This framework, developed from observing the top 5% of graduates from the HKSTP Ideation Programme and the Cyberport Creative Micro Fund over the 2022-2024 period, maps directly onto the 12-week accelerator timeline.

Week 1-2: Problem validation interviews (Gate 1). The objective is not to sell a solution but to diagnose a problem. Each interview must be structured around a single question: “Can you describe the last time you encountered [problem X] and what you did to solve it?” The interviewer should not mention their proposed solution. The conversion metric at this gate is not “interest” but “recall specificity.” A valid signal is when the interviewee can name the date, the cost (in time or money), and the failed workaround. According to the “Customer Discovery Handbook” published by the National Science Foundation I-Corps programme (2024 edition), the minimum bar for a problem to be worth solving is that the interviewee has attempted at least two different workarounds in the past 90 days. If fewer than 70% of interviewees meet this bar, the problem is not acute enough to support a priced product in the accelerator timeframe. The founder should pivot the problem statement by the end of week 2.

Week 3-5: Solution validation interviews (Gate 2). Once the problem is validated, the founder presents a minimum viable product (MVP) — a landing page, a wireframe, or a clickable prototype. The conversion metric here is the “willingness to pre-commit.” The founder must ask: “If I could deliver this solution for [price] by [date], would you be willing to sign a pre-order agreement or place a deposit?” This is a binary question. A “maybe” is a no. The target is that at least 30% of the interviewees from Gate 1 convert to a written pre-commitment. For B2B ventures, this pre-commitment should be structured as a non-binding term sheet, not a handshake. The reason is regulatory: under the SFC’s Code of Conduct for Persons Licensed by or Registered with the SFC (paragraph 17.6), any representation of customer interest that is not documented and verifiable is inadmissible in a due diligence context. A Hong Kong sponsor reviewing a pre-IPO investment will require written evidence of commercial validation. A verbal “I’ll buy it” from a friend at a networking event is not evidence.

Week 6-8: Pilot conversion (Gate 3). The highest-signal validation is a paid pilot. The objective in weeks 6-8 is to convert the term sheets from Gate 2 into signed pilot agreements with payment terms. For a B2B SaaS product, the pilot should be priced at no less than 50% of the expected full annual contract value (ACV). If the full ACV is HKD 120,000, the pilot should be HKD 60,000 for a 3-month term. This price point forces the customer to allocate budget, which in turn forces internal procurement processes. The founder must understand the customer’s fiscal year and budget cycle. In Hong Kong, the majority of corporate budgets are set in Q4 (October-December) for the following calendar year. An accelerator running from January to March is operating in a budget-freeze period for many enterprises. The founder must either target companies with discretionary budget (typically under HKD 50,000 per decision-maker) or find a departmental budget that has not been fully allocated. The conversion metric is simple: a signed contract with a bank transfer or credit card payment. No cheques. No promises.

Data-Driven Interviewing: Avoiding the “Polite Yes”

The single biggest destroyer of accelerator outcomes is the “polite yes” — the interviewee who says they love the idea but never takes a subsequent action. The data from the 2024 “State of Customer Discovery” survey by the Startup Genome project (published January 2025) shows that 68% of founders who reported “positive customer feedback” at demo day had zero paid customers within 6 months. The polite yes is a statistical certainty in Asian markets, where direct confrontation is culturally avoided.

The cost-of-inaction question. The most effective countermeasure is the “cost of inaction” question. Instead of asking “Would you use this?”, the founder asks: “If you do not solve this problem in the next 90 days, what is the quantifiable cost to your business?” The answer must be a number. If the interviewee cannot provide a number — “it would be annoying” or “we might lose some customers” — the problem is not a priority. A valid response is: “We lose HKD 50,000 per month in overtime wages because we manually reconcile invoices.” This number becomes the basis for pricing. If the cost of inaction is HKD 50,000 per month, a solution priced at HKD 10,000 per month is an easy sell. If the cost of inaction is zero, the solution is a vitamin, not a painkiller, and will not convert within the accelerator timeframe.

The reference-ability test. A second data point is reference-ability. At the end of each interview, the founder should ask: “If I were to build this, would you be willing to introduce me to three other people in your company or industry who face the same problem?” The conversion metric is the number of introductions made within 48 hours. If an interviewee says yes but does not send the introductions, the “yes” was polite. The founder should discount that interviewee’s positive feedback by 100%. The only valid signal is action. According to the 2023 “Founder’s Guide to Customer Discovery” published by the Hong Kong Science and Technology Parks Corporation (HKSTP), the top-performing startups in their accelerator programme averaged 2.3 introductions per interview from Gate 1, compared to 0.4 for the bottom quartile.

Converting Discovery into Demo Day Metrics

Demo day is not a pitch competition; it is a data presentation. Investors in the 2025 market, particularly family offices and institutional investors in Hong Kong and Singapore, are demanding evidence of “traction” that goes beyond monthly active users (MAU) or total addressable market (TAM) slides. The SFC’s 2024 “Guidelines on the Use of Alternative Data in Investment Decisions” (published March 2024) explicitly warns against over-reliance on vanity metrics like MAU without conversion data. The guidelines state that “any metric presented to a professional investor must be auditable to the underlying transaction record.”

The conversion funnel slide. A founder’s demo day deck should contain exactly one slide for customer discovery: a three-stage funnel showing (1) number of problem validation interviews conducted, (2) number of solution validation pre-commitments secured, and (3) number of paid pilots signed. Each stage must show the conversion rate. A typical top-quartile funnel from the Cyberport Creative Micro Fund 2024 cohort would be: 40 interviews → 12 pre-commitments (30% conversion) → 4 paid pilots (33% conversion of pre-commitments, or 10% of total interviews). The total value of paid pilots should be stated in HKD or USD. A founder who presents this funnel has demonstrated that they understand the difference between a conversation and a customer.

The pricing signal. The pricing data from the pilots is itself a validation signal. If all four pilots are at the same price point, the founder has found the market-clearing price. If the prices vary by more than 20%, the founder has not yet found product-market fit and is still experimenting. The investor will interpret the variance as risk. The founder should either standardise the pricing before demo day or be prepared to explain the variance with a clear rationale — e.g., “Pilot A is a volume discount for a 12-month commitment; Pilot B is a single-department pilot at list price.” The HKEX’s Listing Decision HKEX-LD-2024-001 on pre-revenue biotech applicants noted that “inconsistent pricing across early commercial agreements is a red flag for lack of commercial discipline.” The same principle applies at the accelerator stage.

Closing: The Minimum Viable Pipeline

Customer discovery during an accelerator is not about learning; it is about converting learning into revenue before the programme ends. The framework above is designed to force that conversion. Three actionable takeaways for a founder entering an accelerator in Q2 2025 or later:

1. Set a minimum of 30 problem validation interviews in weeks 1-2, and do not advance to solution validation until at least 70% of interviewees demonstrate two failed workarounds in the past 90 days. This is your Gate 1 kill switch.

2. Require a written pre-commitment (non-binding term sheet or deposit) from at least 30% of Gate 1 interviewees before building a full pilot, and treat any “maybe” as a definitive no. Verbal interest is not a pipeline metric.

3. Aim for at least 3 paid pilots with signed contracts and bank transfers by week 8, priced at no less than 50% of expected full ACV, and be prepared to present the exact conversion funnel on demo day. A pilot without payment is a free trial, not a customer.