How to Refine a Business Idea | Sharpen It with Five Lenses

Hello, this is Ryuta Hamamoto from TIMEWELL.

When I take questions about starting a new business, the people who stumble at the entrance tend to fall into two camps. Either "I just can't come up with a good idea," or "I thought of something, but it feels like I've seen it before and I can't build any confidence in it." These look like two different problems, but they grow from the same root. They are only looking at the idea from one angle.

When an appraiser examines a gemstone, they turn it over again and again, viewing it from above, from the side, and held up to the light. A stone can sparkle from one angle and look cloudy from another. A business idea is the same: viewed from a single perspective, it always comes out weak. People who only talk about the market overlook their own pain as fuel. People who have fallen in love with their own technology never check whether customers are truly troubled. They set off running with one side left in total darkness. In this article I'll share how to hold the stone of your idea up to five different angles and polish it into a shape that shines no matter which way you turn it. This is one piece in a series that walks through how to build a new business step by step; the overall map is laid out in the Complete Guide to the New Business Framework. If you'd like to measure how much room your own business still has for using AI before you dive in, running the AI Literacy Check first will make the AI techniques in the second half click into place.

A Business Idea Isn't a "Flash of Inspiration"—It's a "Polished Result"

Let me set one premise first. A strong business idea is not a flash of inspiration that falls from the sky. It is a "polished result," an ordinary seed re-carved from changing angles. The very first idea that comes to you in a café almost never becomes a business as-is. So there is no need to lament that "I can't come up with a good idea." More often, you simply don't have a process for polishing one.

Before you start polishing, take an honest look at how weak the idea in your hands currently is. Skip this and you'll begin polishing without knowing where to cut, and you'll only add ornamentation. The following check is the one I run first in every sparring session. Mark the items that honestly apply.

Symptom of a weak idea	In a nutshell
People might say the existing alternative is enough	Excel or the current service already covers it
Nobody is desperately troubled	It stops at "nice to have"
You have no advantage	A big company could copy it overnight
The scale is small	Even selling to everyone won't reach 100M yen in revenue
You can't explain why now	It could just as well be last year or next year
It's unclear who it's for	As vague as "small and mid-sized businesses in general"
The value doesn't land in one sentence	It takes three minutes to explain
It stops at an improved version of an existing product	Just faster, cheaper, more convenient
Technology is the subject	AI or DX comes first, the customer is bolted on
The founder isn't genuinely troubled	You aren't the one experiencing the problem

If three or more apply, don't be discouraged. It's actually proof that you have rough stone that could shine once polished, so consider this where the real work begins. Conversely, if none apply yet you still aren't moving forward, the problem is probably not a weak idea but insufficient validation. I'll get to that in the second half.

Five Lenses: With a Single Viewpoint, an Idea Is Always Weak

What I use is a method for shining light on an idea from five angles. I call each of them a lens. I line them up by initial as A, B, C, D, and E, but knowing that there are five entry points matters far more than memorizing the names.

Lens	Entry point	Note for beginners
A Market-out	From distortions in the market, or frustrations customers put up with	Read behavioral traces, not surveys
B First principles	By breaking down what the industry takes for granted with "why"	Doubt common sense, dig to the physics or economics
C Break-the-bias	By reversing the industry's unspoken rules	Reassemble by dropping, reducing, increasing, adding
D Seeds	From the skills, network, experience, and data you have on hand	What's obvious to you can be a weapon seen from outside
E Lived experience	From what is troubling you right now	Your own pain is the fuel least likely to run dry

Most people hold only one of these. People who can talk market are all A; engineers are all D. Someone with only one lens can shine light from only one angle, so they run with one side left dark. The point of passing through all five is to leave no dark face behind.

There are two important ways to use this. The first is order. Rather than running through A, B, C, D, E in that sequence, I recommend the order E to D, then A, B, C. Your lived experience (E) and the seeds you hold (D) are the fuel least likely to run dry. A business is something you sustain for years. Market trends shift, but your own pain and the things only you can do don't disappear so easily. Solidify a foundation that starts from yourself before letting outside forces in, and you can run for the long haul.

The second is how to set your main axis. The rule is: "Pass through all five, but choose only one as your main axis." Pour 40 percent of your time into the main axis and cover the other four at 10 to 20 percent each. Try to dig equally deep into all of them and every one ends up half-baked. Where to place the main axis is determined to some degree by your own position.

Your position	The lens to set as your main axis
New-business lead at a large company	D (seeds): you can weaponize your company's assets
About to go independent or start a company	A (market-out): enter from the market's frustrations
Student or in your twenties	E (lived experience): your own troubles are a life-sized seed
Second-generation heir in traditional crafts or a local industry	C (break-the-bias): break what the industry takes for granted
Employee starting a side business	E (lived experience): pain you feel firsthand alongside your day job

Skipping the four secondary lenses entirely is forbidden. Build the skeleton with the main axis and erase the blind spots with the other four. Once you can divide the work this way, the idea becomes dramatically more three-dimensional.

The Two Lenses You Dig from Yourself: E and D

Following the order, start with the two that begin from yourself. So many people treat this part lightly, but I believe this is where it's won or lost.

E: Lived Experience—Start from Your Own Pain

The strongest seeds lie in what is troubling you right now. Try to recall something you muttered "what a hassle" about three or more times in the past week. A seed is usually buried beneath that annoyance. To borrow the writer William Gibson's words, the future is already here, it's just not evenly distributed. In an area where you happen to hurt a little earlier than others, you are unknowingly "living in the future."

A common pitfall here is to decide on your own that "surely only I am troubled by this" and throw the seed away. Try searching for people in the same position and it's not unusual to find ten times more of them than you imagined. As a rule of thumb, if the number of people who share the same pain looks likely to exceed 10,000, it's worth moving on to validation. The moment you take your own pain lightly, you throw the fuel least likely to run dry down the drain. This is the one place where you must resist self-censorship and look honestly.

D: Seeds—Take Stock of the Weapons You Already Hold

Next is ideation from what you already possess. Take inventory of the work you've done for 1,000 hours or more over the past three years, the network you've built, the data piling up, and the equipment you can use. What you're looking for is "the part that's obvious to you but surprises people from the outside." I call this "what's obvious to you is a miracle to others."

When people hear "seeds," they tend to picture patent-grade, dazzling technology, but that's a misconception. Plain business processes or a network that exists only in a particular industry are actually harder for competitors to copy and let you fight for the long haul. Not flashy, but it takes years to imitate. For a large company's new business, its customer base or sales channels become the moat; for a local industry, decades of trading relationships. If you feel you have "no real weapons," it may simply be that your inventory is still shallow.

The Three Lenses You Dig from the Market: A, B, and C

Once your self-originated foundation is in place, widen your view outward. The next three are tools for finding distortions on the side of the market and the industry.

A, Market-out, reads distortions in the market from behavioral traces. What you want to watch out for is mistaking this for "asking in a survey." Customers can't put into words what they want. So don't ask; read the traces. What I recommend is reading 30 to 50 one-star reviews of an existing competing service all at once. Amid cries like "it doesn't fit our use case" or "in the end we rebuilt it ourselves," you'll start to see the behaviors customers are reluctantly putting up with. Cancellation reasons and support inquiries are the same vein of ore. "I'd use it if it existed" from a survey is unreliable, but low ratings are the true feelings of people who actually paid money, so they don't lie.

B, First principles, breaks down what the industry takes for granted with "why." Write out ten of the industry's premises and hit the most suspect one with "why" five times. What matters here is not stopping partway. For example, when digging into "why does hiring cost so much," if you stop at the third level with "because of labor costs," you end at "so it's hard, then." Dig two more levels and you reach the principles of information asymmetry and labor intensity, and you see a fresh starting point: "then what if we could fill that asymmetry with data?" Stop at three levels and it becomes "so it's impossible"; dig to five and it turns into "so now it's possible." That difference is enormous. Doing this sparring alone is surprisingly tough, so it works well to have an AI dig one question at a time. I'll introduce the prompt in the second half.

You are a first-principles thinking coach. Working with me one-on-one, repeat "why" and break the industry's assumptions down to indivisible principles. Do not deliver a complete answer all at once; always ask back one question at a time.

[The industry premise I want to doubt]
[e.g., In this industry it's normal for a quote to take two weeks]

How to proceed
1. First, ask only one question about my premise: "why is that so?" When I answer, ask "why" about that answer too. Repeat this at least five times.
2. If I start to stop at "because it's an industry custom" or "because it's always been that way," cut in with "then which principle does that custom come from: physics, economics, or human psychology?"
3. Once you've dug through all five levels, articulate the indivisible principle you reached in a single sentence.
4. Starting again from that principle, present three "how it should really be" ideas and contrast how each differs from the existing approach.

C, Break-the-bias, reverses the industry's unspoken rules. First, make visible "what everyone in this industry advertises in the same words" and "the unwritten rules no one doubts." Having a physical storefront, keeping salespeople, annual contracts as the norm, delivering a finished product, and so on. Then reassemble them with four moves: drop, reduce, increase, add. The barbershop service that specialized in around-10-minute cuts and dramatically lowered its price dropped the premise of "the full-service barber." The spread of home-sharing, staying in someone else's spare room, broke the unwritten rule that "lodging is something hotels provide." Neither was being clever for its own sake; something got better for the customer in exact proportion to the unwritten rule they broke. Put the other way around, a reversal for which you can't write, in one line, the name of the unwritten rule you broke and what improves for the customer is merely eccentric, not differentiated. A reversal and mere eccentricity are two different things.

Once Polished, Validate Within 48 Hours

After running through the five lenses, integrate the insights and rewrite the idea. The rule here is "always write three versions." V1 is an extension of the existing; V2 is a middle-ground option; and V3 is "the most cutting-edge option that industry people will laugh at but could become an innovation." The reason you can't write the third one isn't a lack of talent, it's the self-censorship of "I don't want to be laughed at." And the gold is usually buried in V3. At the end of a sparring session, I always make sure we squeeze out a V3.

Once rewritten, score the chosen version on five axes. Rate Pain Depth, Market Size, Uniqueness, Feasibility, and Timing (why do it now) out of 5 each, look at the total out of 25, and treat 18 or higher as the passing line. The one that tends to get inflated here is Pain Depth. Pain at the level of "that's a bit annoying" is only a 2 or a 3. Mistake it for subscription-worthy pain and you get a business with a fascinating structure that no customers come to. Use whether there are three or more cases of someone actually paying money in the past as your yardstick for judgment.

What underpins Timing is Why Now, the explanation of why now. Screen this on three elements: changes in technology cost, changes in regulation or policy, and shifts in customer behavior. Check whether you can state each with concrete numbers and dates. "Because AI is coming, so now" is the vague phrasing most disliked in this screening, and it's a zero. For instance, you only approach full marks when you can describe the change in concrete terms, as in "the unit cost of using major generative AI has dropped substantially over the past two years, making processing that used to be uneconomical now viable."

Here's a prompt you can hand the whole job to, from the scoring above to designing the next step.

You are a screener conducting a pre-investment stress test. Score my polished idea and design the next step. Direct your criticism at the idea itself, not at me personally.

[Polished idea (one sentence)]
[To whom, what, and how differently you provide it]

1. Screen Why Now on three elements. Examine strictly whether each of (1) change in technology cost, (2) change in regulation/policy, and (3) shift in customer behavior is backed by concrete numbers, dates, or statistics, and score vague phrasing like "because AI is coming" as zero. If two of the three elements aren't present, cap Timing at 3 points.
2. Score on five axes (Pain Depth / Market Size / Uniqueness / Feasibility / Timing, 5 points each). Add a one-line reason for each score and give the total out of 25. The passing line is 18.
3. If it doesn't pass, use the lens that corresponds to the lowest axis (A if Pain Depth is low, C if Uniqueness is low, B if Feasibility is low) to give concrete instructions on what to reconsider.
4. Pick one validation action that can be started within 48 hours from the following, and design it fully with owner, deadline, deliverable, and pass/fail criteria: one landing page plus an ad buy / DMs to five prospective customers / reading 50 cancellation reviews / three sales calls with a one-page deck / a 30-minute interview using a Figma mock / running a concierge MVP manually for one day / publishing a pre-order sales page.

And the final gate is validation. Even if you clear 18 in scoring, that's still only a story on your desk. Polishing without validation is, unfortunately, delusion. Make it a rule to start at least one action within 48 hours. The trick when choosing is to pick "the action you've been most afraid of and avoiding." Being afraid is proof that there's a fact you don't want to confirm, and that's where learning is highest. Declare the owner, deadline, deliverable, and pass/fail criteria on the spot before you move.

Words alone are hard to grasp, so let me run a fictional business all the way through from start to finish.

The founder is a former registered dietitian at a care facility who cared for their grandfather at home. The first idea went like this: "A nutritionally balanced meal-delivery service for the elderly. Healthier than existing boxed-lunch delivery, and 1,000 yen a month cheaper." Run it through the symptom check and you get four hits: existing alternative is enough, no advantage, an improved version of the existing, and competing on price. A textbook weak idea.

Now apply the five lenses. Looking back through E (lived experience), what was truly hard about home care wasn't the nutrition of the boxed lunches. It was that the family had to judge, at every meal, the degree of thickening and mincing to match the grandfather's swallowing (deglutition, the ability to swallow) level, which changed day to day. There are several hundred thousand caregiving households nationwide facing the same pain. Taking inventory with D (seeds), out come the know-how for adjusting food by swallowing-difficulty stage cultivated at the facility, and a network of care managers. This is an asset ordinary families absolutely do not have. Reading the one-star reviews of existing boxed-lunch delivery with A (market-out), you find lined up: "too hard for my parent," "too soft," "in the end the family remakes it." Breaking down the premise that "meal delivery means delivering a finished product" with B (first principles), you see that what customers really want isn't a boxed lunch but "the result of a meal my parent can safely eat given today's condition." Breaking the unwritten rules with C (break-the-bias), you see a reassembly: drop the fixed menu, reduce the family's judgment workload, increase the fit to swallowing level, and add a mechanism where sending today's condition automatically adjusts tomorrow's menu.

Here's the integrated V3: "For home-caregiving families, a swallowing-level-linked semi-prepared meal kit plus LINE consultation with a registered dietitian. When the family sends the parent's condition for the day through an app, the next day's menu and the degree of thickening and mincing adjust automatically. Fixed monthly fee." Scoring it: Pain Depth is a 5 because they'd pay to avoid the every-meal judgment stress, Market Size a 4, Uniqueness a 5 because it reassembles the meal-delivery industry into a condition-linked service, Feasibility a 3 because it requires a dietitian setup, Timing a 4, for a total of 21, which passes. Why Now can be told on three elements: individualized menu optimization has become cheap with generative AI, policy is shifting toward home medical and home nursing care, and it has become normal for even elderly family members to use LINE. The 48-hour validation was set as a single move: post "Is there anyone who wants a service like this?" in an online community where caregiving families gather, and confirm from the number of responses whether the pain really exists. It's already a completely different thing from the first weak idea.

If you'd like to grasp how to build a new business a little more systematically, reading the Value Proposition Canvas, which digs into the value proposition itself, and How to Calculate TAM, SAM, and SOM, for nailing down the Market Size axis with numbers, alongside this will raise the precision of your scoring. If you'd like your team to develop a feel for making AI a partner in building a business, or you want a companion for the launch itself, we do exactly this kind of sparring and validation design together through our AI consulting service, WARP.

Making AI Your Sparring Partner for the Five Lenses

Reading this far, you may have felt "running through five lenses on my own sounds like a lot of work." In fact, this theme has an excellent affinity with AI. Have Claude or ChatGPT play, in turn, the market-out analyst, the first-principles physicist, and the break-the-bias strategist, and even alone you can hold a pseudo-workshop.

There is one rule to declare before you start. Direct criticism "at the idea, not the person." Being torn apart here and now is far cheaper than being torn apart after you launch. So let the AI break the idea without holding back, while you make sure you don't feel your character is being denied. Communicate this boundary at the outset and you'll get feedback that digs in deep.

First, here's the main sparring prompt that runs through all five lenses end to end.

You are an expert in evaluating business ideas. Refine my business idea thoroughly with the "five lenses." Direct your criticism at the idea itself, not at me personally. No need to hold back.

[Current idea (within 200 characters)]
[To whom, what, and how you provide it]

Be sure to run through all lenses in the following order (the order polishes from self-originated to market-originated).
1. Lens E (lived experience): Ask back what my own pain behind this idea is, and estimate together how many people in the world are likely to share the same pain.
2. Lens D (seeds): Ask five questions to draw out the work, network, data, and equipment I've done longer than others that could become a weapon.
3. Lens A (market-out): List five behaviors customers in the target market reluctantly put up with, and three weaknesses likely to be attacked in low ratings of existing alternatives.
4. Lens B (first principles): List ten premises, hit the one most worth doubting with "why" five times, and dig to an indivisible principle.
5. Lens C (break-the-bias): Make visible five of the industry's unspoken rules and reassemble them by dropping, reducing, increasing, and adding. For each item, add one line each on the name of the unwritten rule broken and what improves for the customer.

Finally, integrate the insights from the five lenses and rewrite the original idea into three versions (V1 = extension of the existing / V2 = middle-ground option / V3 = the most cutting-edge option that industry people will laugh at but could become an innovation). Be sure to sharpen V3.

When running through all of it at once is too much, entering from a diagnosis makes it easier. It will inspect the symptoms of a weak idea and even choose the main-axis lens that fits your position.

You are my new-business sparring partner. First, diagnose whether my idea matches the "typical symptoms of a weak idea."

[Idea (one sentence)]
[In one sentence here]

For each of the following ten items, judge with reasons whether it applies or not, and count the number of hits.
(1) People might say the existing alternative (Excel or the current service) is enough (2) Nobody is desperately troubled (3) You have no advantage (4) The scale is small (even selling to everyone won't reach 100M yen in annual revenue) (5) You can't explain why now (6) It's unclear who it's for (7) The value doesn't land in one sentence (8) An improved version of the existing industry (9) Technology-first with the customer bolted on (10) I, the founder, am not genuinely troubled.

After the diagnosis, taking my situation into account (e.g., new-business lead at a large company / about to go independent / student / second-generation heir in a local industry / employee with a side business, write whichever applies), recommend, with reasons, one of the five lenses (A: market-out / B: first principles / C: break-the-bias / D: seeds / E: lived experience) to set as the main axis. Finally, give three questions I should answer first with that main axis.

Running through it this way, you realize AI never tires of asking "why" over and over and properly points out the unwritten rules you're trying to dodge. Catching a human sparring partner every time is hard work, but AI will keep you company even at midnight. That said, don't swallow the AI's third option whole. You are the one who makes the final call; treat the AI strictly as a tool that adds angles.

If I had to sum up how to refine a business idea in one line, it comes down to "never being satisfied with a single viewpoint." Shine light in the order E to D, A, B, C, choose one main axis, rewrite into three versions, and move to validation within 48 hours. Run this round trip at least three times and it grows into an idea completely different from the first draft. And since polishing and validation go back and forth, if validation misses, just return to the diagnosis. If any axis is under 7 points, make the corresponding lens your next main axis and go one more round. Rather than waiting for a perfect first draft, starting to turn the loop even roughly gets you stronger faster. When you'd like to talk through where to start given your own situation, feel free to reach out through a WARP individual consultation. I hope your rough stone becomes one that shines no matter which way you turn it.