Business: Turning Products into Companies
A breakthrough demo is not a business. VCs qualify almost every pitch with one blunt question: is this a feature, a product, or a company? The hard work is closing the gap between something that works and something customers buy repeatedly — with a go-to-market, business model, and execution culture that can scale.
Source note: This article synthesises a Harvard Innovation Labs Startup Secrets workshop led by Michael Skok, with case stories from Gregg Favalora (Actuality Systems) and John McEleney (SolidWorks / CloudSwitch / Belmont). It is a practitioner summary, not a transcript. Slides and related workshops live on Startup Secrets / mjskok.com.
Feature, product, or company?
Messaging, photo sharing, check-in, friend directory — each can look like “just a feature.” Twitter was largely messaging. Instagram made photo sharing a billion-dollar company. Facebook started as a directory of friends. Foursquare made check-in a company (whether it lasts is a different question).
The iPad was written off as “a big iPod.” It became one of the most successful consumer devices ever and helped define a post-PC era. Vision (Jobs famously did not ask focus groups) and disciplined execution both work; there is no single formula. The useful habit is to ask of your idea: am I willing to invest my life in building a company around this — or is it a feature someone else will absorb?
Shift happens. Markets — especially mobile — change faster than product plans. Many ventures die in the gap between a great build and real traction: capital spent, life savings at risk, not enough momentum or runway to get across. Skok’s estimate from the venture world: roughly two-thirds never make that crossing. This is bigger than “crossing the chasm”; it starts in how you approach building the product in the first place.
Case study A: Actuality Systems — awesome tech, unpaid market
Gregg Favalora’s Actuality Systems built a crystal-ball-scale volumetric 3D display: floating imagery you could view from all sides without glasses, driven by thousands of light patterns on a rotating screen (~10,000×/second). It started in a Central Square apartment and, over ~12 years and ~$15M raised, became a painful story of technology looking for a market — ending in a patent sale, not a scaled product company.
The overarching lesson
Even with an awesome product you need:
- Deep understanding of at least one market where warm-blooded customers write cashable checks
- More money than you think — especially for hardware (atoms and photons), it always takes more
“If we build this hologram, people will buy it” was the early bet. Mechanical CAD, pharma, luggage scanning, military viz, oil & gas — segments went on a whiteboard; big accounts were pursued. Praise was abundant (“your parents must be so proud”); checks and ROI cases were not. Researchers loved it; that was a mirage market.
Lessons compressed
| Phase | What happened | Takeaway |
|---|---|---|
| Raise & build | 2.5 years to raise ~$1.5M in the portal boom; reverse-engineered TI DLP with no datasheet; 6 people | Hustle (always carry a leave-behind); hardware dependency is existential |
| Fishing expedition | More capital; “grown-up” CEO; OpenGL-style path so apps could drive the display | A component (spark plug) is not a whole product (car); APIs assume expert customers |
| Market map | Whiteboard segments × target accounts; endless “best thing ever, but prove $100K ROI” | Map the flow of money, not just market names and workflows |
| Pivot | Radiation oncology / brachytherapy Machine Vision; still failed when markets tanked (2009) | Product marketing (MRD, bet the company) is as critical as engineering |
| Wind-down | One month of cash; patents held and sold after a near-collapse of the buyer | Angels get tiny checks; nonzero still matters; depression after false exits is real |
Gregg’s bias early on: if you are not writing code or using diodes, you are wasting time. Wrong. Product marketing — deeply understanding a customer, defining the market, writing the MRD the engineers build against — is CEO-adjacent work. Budget for it early.
Hardware constraint: paper prototypes often fail for visual products. People imagine liking a hologram; answers only get real when they see it. That raises the cost of validation — it does not remove the need for a valuable problem.
The product–company gap
Three Startup Secrets pillars fill the gap:
- Go-to-market
- Business model
- Execution
Entrepreneurs often say “it’s all about the product” and point at Apple. Look at the P&L flip instead:
| Stage | Where money goes |
|---|---|
| Start | ~100% of spend in engineering / building the product |
| Next | Proving market acceptance (product marketing, early sales) |
| Scale | Sales & marketing dominate OpEx |
Public Apple at the time of the talk: roughly ~2% of sales on R&D, multiple times that on SG&A. Demandware and peers show the same pattern: from product-heavy burn to sales-and-marketing-heavy spend. Design the company so go-to-market friction falls — so SG&A does not have to boil the ocean — if you want a path to ~20% bottom-line profit with a product that “flies off the shelves.”
Develop foundations: value, then MVP
Build less, and smarter
- Build core only — your exceptional capability (Gregg’s core was creating 3D imagery; displays/DLP belonged to TI)
- Prefer MVP over boiling the ocean; prefer core over even MVP when possible
- Crowdsource / outsource non-core (landing pages, wireframes) — oDesk/Elance-era labour markets, Estonia, etc.
- Be your own customer — if you ship a platform/tool, make internal engineers use it first (Demandware’s pattern). Few companies do this; the good ones make it culture
- Open and extensible — Linux-style; open APIs so others add drivers/integrations; leverage open source (Drupal/Acquia-scale communities) instead of rebuilding Facebook/PayPal/CRM connectors
Valuable ≠ viable
An MVP is irrelevant if it does not solve a valuable problem. Viable (shippable) and valuable (worth paying for) are distinct. Before the first feature: what problem, for whom, how significant?
Qualitative shortcut from the value-proposition workshop: a before / after scenario. Does the customer move from chronic pain to real joy — or get a little novelty? Penicillin vs vitamin.
Validate with dollars and “why”
- Get out of the building (Abby Ficner’s agile validation tools; do not guess)
- Selling is validation — but who and why matter. Researchers writing small checks ≠ a market
- Prefer prepaid customers / Kickstarter-style commitment over surveys alone
- Gain–pain ratio: not only the gain you deliver, but the pain to adopt (price, integration, training). A $100 plug-and-play 3D display opens markets a $100K research appliance never will
Avoid multifaceted value props early (e.g. security that must excite consumers and enable banks/merchants). Pick one audience, one starting value prop. Hunt problems that are unworkable, unavoidable, urgent — and markets that are underserved.
From MVP to minimum viable segment (MVS)
Product–market fit is not one slide. Double-click into go-to-market fit.
The stretch trap
Classic failure mode:
- Define a broad market (“medical”)
- Talk to customers; collect dissimilar needs (dots all over the map)
- Add features for each → MVP becomes a large product
- Burn cash and engineering on inconsistency
Do the opposite. Find one or two customers whose pain and need line up. Then:
- Product footprint stays stable (no constant expansion)
- Messaging, channels, and tactics stay consistent
- You get references and can dominate a small beachhead early
A segment is not always “SMB vs enterprise” or “medical vs government.” It might be “anyone who must trade derivatives internationally” across verticals — or compliance needs shared by pharma and financial services. Define the segment by shared need, not by textbook industry labels.
Minimum viable segment test: can you repeat customer-to-customer without changing the product or the go-to-market? If yes, you found a segment. If needs scatter, throw outliers out to stay on the diagonal.
How you find pain (shoe leather)
John McEleney: there is no clever substitute. Get in front of people. Grade yeses — polite yes, Japanese yes (“I heard you”), vs check-writing yes. Bias is dangerous if only the founder hears the story; a partner who challenges your hearing helps. Computer Vision lore: padlock the lab until people have sat with at least three customers.
Scorecard every need: what exactly is it, what problem does solving it remove, what would you pay, and would you write a check in advance? Skok started companies with customer checks before investor money.
Narrow to a blatant critical need
Example funnel (mobile):
smartphones → mobile professionals → field workers → service technicians → medical equipment → hospitals → critical care (if unserviced, someone dies)
That is a critical need — not “2% of a billion phones at $4.99.” Focus is the number-one missing startup habit. Expand on success one aligned customer at a time; most failures come from trying to do too much and contracting on failure.
Vision vs execution: a product roadmap that becomes a company
Help the founder write the path in advance:
breakthrough → feature → product → solution for a market → company with a business model
Product-side milestones along that path: usability, partners/services/whole product, referenceable customers, segmented beachheads with repeatability, then product line and profitable scale. Validate with customer-based metrics (NPS if you believe in it, retention, upsell, LTV) — not impressing the board with cash-flow theatre. Hitting forecast means little if customers are not buying repeatedly.
Architect to attract: slippery products
Slippery = take friction toward zero across see → try → buy → implement → deploy → own. Skok’s visual: everyone else runs in shoes; you get an ice lane.
Rough checklist (SIMPLE + more):
| Attribute | Intent |
|---|---|
| Simple | Advantage ≈ Innovation × Simplicity. Complex = disadvantage (Microsoft Media Center remote vs Apple’s few buttons; Knowledge Management category vs Evernote) |
| Low / no initial cost | Frictionless trial — but avoid free fall (perceived value → zero, no path to monetize). LinkedIn: free network value, monetize recruiters |
| Installs / integrates easily | Embrace and extend; open APIs; restful payment/CRM/etc. instead of rebuilding. Change is risky, painful, costly — do not invent training cliffs (haptics, 3D instruments) |
| Proves value quickly | Consumer: instant gratification. Enterprise: payback in ~3–12 months. Build self-proving analytics (baseline → savings shown in-product) |
| Progressive disclosure | Reveal capability step by step (Drupal Gardens: live in ~15 minutes → modules on → export to full Drupal; no 10,000-module cliff) |
| Easy to use (UBIE) | Out-of-the-box experience; templates. Siebel’s install pain vs Salesforce’s delightful start |
| Obvious ROI | Revenue up, cost down, or early competitive advantage for visionaries; calculators (e.g. desktop virtualization) help |
| Sticky | Customers cannot live without it (iPad-class dependency) |
North star: disruptive innovation with non-disruptive adoption — VMware-style: huge utilization gains without rewriting apps or replacing hardware. Apperian-style iPad enterprise example: Estée Lauder gained sales lift from iPads without handing personal devices to IT — cloud delivery, no IT touch, gain without classical IT pain.
Packaging, pricing, and whole-product detail live in adjacent Startup Secrets sessions; design for them early even if you do not implement everything on day one.
Case study B: SolidWorks — 1503 and weeds through concrete
John McEleney joined SolidWorks near first sale, later ran it through the Dassault years (~$100M → ~$400M under his CEO stretch; company later ~$600M, highly profitable). Contrast with Actuality: same “3D world,” opposite outcome — because of who paid, why, and how the company sold.
Why SolidWorks existed
PTC (Pro/ENGINEER) made parametric solid modeling valuable — change designs like a spreadsheet. The problem: too expensive, too hard to learn/use. A huge population had “noses against the glass.” SolidWorks: ~$4K (not $20–30K), sold through VARs, usable 3D for people who already understood the benefit but could not afford PTC.
The 1503 culture
Four numbers burned into DNA: 1–5 seats, 0–3 month sales cycle.
| Temptation | What actually happened | Why 1503 won |
|---|---|---|
| Chase Ford powertrain / 60 seats | Love → more demos → pilots → 3 seats after 6 months | Under-capitalized VARs lose money on long cycles |
| Big expected value | Looks smart on paper | VAR reps at ~$65–70K need quick hits, not strategic selling |
Ideal account (“modeling saturation index”): a few Pro/E seats (so the org already believes in 3D) plus many AutoCAD seats that had not moved — because 3D was too hard/expensive. Not “convert 10 Pro/E” and not “teach 10 pure 2D users from zero.”
Strategy: land and expand — weeds through concrete. EMC: never a top-down CIO sale; displacement a few seats at a time until hundreds forced recognition. First-year plan ~$4M; actual closer to ~$13M once the repeatable motion scaled.
Execution detail as culture: profitable on ~$2,400 net to SolidWorks; upgrade boxes engineered to 15.9 oz so UPS did not jump a weight band. Strategy and ops were the same mentality.
Partner ecosystem (whole product)
Focus on core modeling; partners build the rest. Three-step partner play:
- Marketing cold fusion — lightweight partner program, credibility by association/numbers
- Pick #3 in each category, force them to #1 in the SolidWorks world (analysis, etc.)
- Followers follow — category leaders then had to show up; SolidWorks became the platform
Operating mantras
- Only advantage is time — incumbents have customers, capital, megaphone; you have the calendar
- Events force actions — SolidWorks World (~$2M, thousands of users) was worth it if only to force one-day alignment and decisions (April 15 tax logic)
- Perfect is the enemy of the good — ship, iterate
- What you think vs what you know — unpaid beta praise lies; CloudSwitch: people paid for education, not the product they praised
- Price from the channel’s P&L — subscription set from end-user willingness × VAR support capacity × your margin (market was used to ~18% of list; align the middleman)
- Hire people who scare you with confidence; salespeople are coin-operated and recognition-driven — if not, they are not salespeople
- Culture of customers you may never name — mold shops and Nike suppliers, not only logo walls
Analytics → bundling → defense
A “performance monitoring” nugget (aggregate usage/version/add-ins) started as a support problem. It revealed a ski-slope: few customers used many partner apps. In a price war with Autodesk, SolidWorks bundled into Office Professional / Premium, raised ASP (~$4K → $5.4K → $7.5K) and grew volume, raised subscription, paid partners blanket licenses, and made VAR book-of-business sticky enough that competitors could not easily steal the channel.
Failed adjacent bet: 3D content / electronic catalogs for manufacturers. Engineering relationships did not transfer to marketing buyers; needed media salespeople and a separate company mindset. Lesson: data can be more valuable than you capture — but only if you respect a different buying center.
McEleney’s upfront product asks for founders
- Instrument early (usage, versions, attach) — it pays later in packaging and defense
- Design for a profitable, repeatable motion (your version of 1503) that average channel talent can execute
- Align product, pricing, packaging, and culture so every step makes money
Subscription was as strategic as modeling technology: end-user identity through the VAR, recurring revenue, channel lock-in.
Closing the loop
| Actuality | SolidWorks |
|---|---|
| Wow technology; years of market fishing | Clear underserved need (affordable, usable 3D) |
| Component + API ≠ whole product for buyers | Core + partner ecosystem = whole product |
| Mirage markets, late product marketing | 1503 focus, VAR economics, land-and-expand |
| Underfunded for hardware reality | Capital and model matched the motion |
| Patents as residual value | Analytics + packaging as compounding value |
If you take one idea: while you build the feature or product, design the go-to-market methodology that becomes culture — simple enough that a mid-tier VAR rep can repeat it. R&D is a minority of the eventual P&L; sales, marketing, and the slippery path to the customer are where companies are made.
“It’s easy to start a company; it’s hard to build a business.” Events force actions. Perfect is the enemy of the good. The journey is why most entrepreneurs do it — see the world as it should be, then make one narrow beachhead prove you are right.
Recap: what to remember
- Qualify — feature, product, or company? Be honest
- Market before miracles — who writes checks, why they make money, map cash flow
- Expect the OpEx flip — engineering → market proof → sales & marketing
- Core + open — build less; be your own customer; APIs and partners
- Valuable before viable — before/after; gain–pain; prepaid validation
- MVS over sprawling MVP — aligned needs, repeatable motion, beachhead dominance
- Critical need — segment until someone cannot live without the fix
- Slippery — simple, low friction, fast proof, progressive disclosure, UBIE, ROI, sticky
- Non-disruptive adoption of disruptive value
- Your 1503 — one clear, profitable sales motion; culture and packaging that defend it
Discussion
Comments
Share feedback or questions about this page. No account required.
Loading comments…