The expected-value framing is sound if the $50 tests are treated as cheap market-learning options, not as proof that a scalable SaaS has been found. The math supports disciplined spend because Brian can cap the downside of each hypothesis at roughly $50 plus a few hours, while his engineering skill keeps product-development cash cost low. But the 80% assumption is only plausible if “success” means “one or more ideas produce a qualified signal worth manual follow-up”: a demo booked, a buyer conversation with a named urgent workflow, a paid/concierge pilot, or a waitlist lead who survives follow-up. It is not plausible if “success” means “a scalable SaaS business emerges.”
For Brian specifically, the bottleneck is not building. He is a senior fullstack engineer at Dripos in NYC (YC S20), so he can prototype cheaply. The bottleneck is market discovery: picking painful enough niches, writing buyer-specific copy, doing follow-up, asking for money, running enough sales reps, and refusing to build beyond the evidence.
The base formula is:
EV = P(success) × payoff - validation spend - follow-on spend/opportunity cost
For this question:
The math is favorable only when the hit produces enough monthly contribution profit to justify 8 months of focus. It is unfavorable when “success” is merely click curiosity or an email list with no buyer urgency.
Assume validation spend is $500 and follow-on opportunity cost is temporarily ignored. This isolates the value of the experiment portfolio.
| Monthly contribution after first customer | 8-month payoff if hit | EV at 80% hit probability | Read |
|---|---|---|---|
| $250/mo | $2,000 | $1,100 | Worth testing, not worth 8 months alone |
| $500/mo | $4,000 | $2,700 | Positive but fragile |
| $1,000/mo | $8,000 | $5,900 | Clearly positive if follow-on time is cheap |
| $2,000/mo | $16,000 | $12,300 | Strong small-business outcome |
| $5,000/mo | $40,000 | $31,500 | Very strong, but much harder to reach |
This table shows why $500 of validation spend is almost never the economic problem. The real swing factor is whether the post-hit work can plausibly produce real paying customers.
Holding 8-month contribution constant and ignoring opportunity cost:
| Monthly contribution | P=20% | P=40% | P=60% | P=80% |
|---|---|---|---|---|
| $250/mo | -$100 | $300 | $700 | $1,100 |
| $500/mo | $300 | $1,100 | $1,900 | $2,700 |
| $1,000/mo | $1,100 | $2,700 | $4,300 | $5,900 |
| $2,000/mo | $2,700 | $5,900 | $9,100 | $12,300 |
| $5,000/mo | $7,500 | $15,500 | $23,500 | $31,500 |
A useful way to sanity-check the 80% portfolio hit assumption: with 10 independent tests, an 80% chance of at least one hit requires about a 14.9% per-test hit rate. A 40% portfolio chance requires about 5.0% per-test. A 20% portfolio chance requires about 2.2% per-test. So 80% is not mathematically absurd for “one qualified signal across 10 narrow tests,” but it is too high for “one scalable SaaS business across 10 ads.”
If Brian treats the 8-month follow-on as nights/weekends with low cash opportunity cost, the tests look attractive. If he prices his own time like senior engineering labor, the hurdle rises sharply.
| Monthly contribution | EV with $0/mo opportunity cost | EV with $2,500/mo opportunity cost | EV with $5,000/mo opportunity cost |
|---|---|---|---|
| $500/mo | $2,700 | -$13,300 | -$29,300 |
| $1,000/mo | $5,900 | -$10,100 | -$26,100 |
| $2,000/mo | $12,300 | -$3,700 | -$19,700 |
| $5,000/mo | $31,500 | $15,500 | -$500 |
This does not mean “don’t test.” It means the $50 tests should be used to avoid wasting the 8-month follow-on window. The validation budget is an option premium. The costly decision is committing months to an idea without buyer pull.
Churn compresses the 8-month payoff. If a customer starts paying immediately, the expected customer-month multiplier over eight months is:
| Monthly churn | Expected paid customer-months over 8 months | Practical implication |
|---|---|---|
| 0% | 8.00 | Clean math, but optimistic |
| 2% | 7.46 | Healthy B2B SaaS-ish retention |
| 5% | 6.73 | Still workable if CAC is low |
| 10% | 5.70 | Growth treadmill; needs strong acquisition |
| 20% | 4.16 | Usually a leaky bucket unless price/CAC is tiny |
Lenny Rachitsky’s churn benchmark discussion argues that churn is a major “tax on growth,” and that cohort retention / net revenue retention are more informative than blended monthly churn. For Brian’s early MicroSaaS tests, this means the first customer is not enough. A pass should include evidence that the workflow recurs and that the buyer expects to keep paying.
If the 8-month post-hit period includes time before money arrives, the payoff shrinks fast.
| Time to first revenue after hit | Revenue months left in 8-month window | Effect |
|---|---|---|
| 0 months | 8 | Best case: paid pilot or immediate concierge |
| 1 month | 7 | Fine if follow-up is tight |
| 2 months | 6 | Common for B2B; EV needs higher price |
| 3 months | 5 | Requires discipline and bigger ACV |
| 4 months | 4 | Half the window gone |
| 6 months | 2 | Validation has mostly failed for this strategy |
A landing-page click has weak value if it creates a 3-month detour before revenue. A paid pilot or concierge service compresses time-to-revenue and reveals real willingness to pay.
The paid-ad economics are noisy at $50. The goal is not to calculate durable CAC from tiny samples; it is to find whether intent exists cheaply enough to justify direct outreach.
| CPC | CAC at 1% conversion | CAC at 2% conversion | CAC at 5% conversion | CAC at 10% conversion |
|---|---|---|---|---|
| $1 | $100 | $50 | $20 | $10 |
| $3 | $300 | $150 | $60 | $30 |
| $5 | $500 | $250 | $100 | $50 |
| $10 | $1,000 | $500 | $200 | $100 |
| $20 | $2,000 | $1,000 | $400 | $200 |
WordStream’s 2025 Google Ads benchmark says search advertising costs have risen for years and CPC increased for 87% of industries in its dataset. So $50 may buy useful directional traffic in some narrow niches, but in expensive B2B categories it may buy only a handful of clicks. This reinforces the policy: use ads as a signal generator, then do manual sales; do not infer scalable CAC from a $50 smoke test.
Customer-development literature supports the shape of the strategy, but not the lazy version of it.
Steve Blank’s Customer Development model separates customer discovery, customer validation, customer creation, and company building. He emphasizes that startups need a process for discovering markets, locating first customers, validating assumptions, and growing the business; Customer Discovery tests hypotheses in front of customers, and Customer Validation develops a sales model that can be repeated and scaled. That maps directly to the distinction Brian needs: landing-page tests belong in discovery, not proof of a repeatable sales model.
Kromatic’s landing-page smoke-test guide is especially aligned with this report. It describes landing-page tests as quantitative, remote, behavioral tests of whether a value proposition is compelling enough for visitors to take a concrete action. It explicitly warns that an email signup is not a purchase, and willingness to pay is better tested with simulated purchase, pre-order, pre-sales, or concierge tests. It also gives a rough cost range of $30-$600 and notes that trustworthy conversion reads need enough visitors, usually about a week of ads.
Paul Graham’s “Do Things that Don’t Scale” is the strongest corrective to founder avoidance. He says the most common unscalable startup work is recruiting users manually: founders cannot wait for users to come to them. For Brian, a paid search smoke test should therefore create a sales queue, not a permission slip to build in private.
Lenny Rachitsky’s CAC payback discussion is useful for the scaling phase. It defines payback period as CAC divided by gross profit, and warns founders not to use revenue instead of gross profit. A MicroSaaS that costs $500 to acquire a customer at $100/mo revenue but $80/mo gross profit has a 6.25-month payback, not 5 months. This matters because small paid-ad channels can look profitable until support, hosting, refunds, and founder time are counted.
MicroConf’s State of Independent SaaS page is credible directional evidence that bootstrapped SaaS is benchmarked by ARR stage and that many founders operate pre-revenue or sub-$10K ARR for a long time. The accessible page is mostly a report landing page rather than the full dataset, so it should not be over-read; the relevant takeaway is that indie SaaS has a long grind distribution, not a guaranteed quick ramp.
High Alpha’s 2025 SaaS Benchmarks page is more mature-company oriented, but it reinforces the scaling point: growth quality depends on retention, expansion, CAC payback, and efficient acquisition. Those concepts apply even to a tiny MicroSaaS, just at smaller scale.
The 80% assumption is plausible only under a narrow definition:
Why? The first is a learning event. The second requires multiple gates: pain, buyer identity, urgency, budget, reachable channel, credible offer, sales conversion, onboarding, retention, support burden, pricing power, and repeatability. Ads can help test the first few gates. They do not solve the rest.
So the better phrasing is: “A disciplined 10-test portfolio may have a high chance of finding one or more promising market signals. The chance that a signal becomes a durable business is much lower and must be modeled separately.”
Validation success means one of these happens:
Business success means more gates clear:
The $50 tests can validate signal. They cannot by themselves validate business success.
Brian’s engineering skill changes one term in the EV equation: product-development cash cost. It does not remove market-discovery risk. In fact, being able to build cheaply creates a specific danger: building too soon because code feels like progress.
The EV bottleneck for Brian is therefore:
A good use of engineering skill is not “build the SaaS after a few clicks.” It is “ship test pages fast, instrument them, make credible demos, and build only the smallest thing needed after a buyer has committed.”
Define the hypothesis in one sentence:
Do not run generic “AI tool for X” tests. Run narrow tests like “recover underbilled change orders for 10-50 person trade contractors” or “turn denied insurance prior-auth packets into a resubmission checklist for small clinics.”
A $50 test passes if it creates at least one of these:
For low CPC niches, require more volume: e.g. 20-50 clicks and at least 5-10% high-intent CTA conversion. For high CPC niches, a single qualified call can be meaningful, but only if followed by a real conversation.
A test fails if:
Stop an idea after one $50 test if there is no qualified signal and the copy/channel was coherent. Run a second test only if there is a clear fix: better ICP, clearer pain language, better keyword intent, or a more concrete CTA. Kill after two weak tests unless non-ad outreach contradicts the result.
Stop the whole portfolio temporarily if Brian is not doing follow-up. More ad spend without sales behavior becomes founder avoidance.
Do concierge/manual service when a buyer has the pain but the product shape is uncertain. Offer a narrow paid pilot such as:
Concierge is the bridge between validation signal and product requirements. It also tests whether the buyer will tolerate operational friction for the promised outcome.
Build only after at least one of these gates clears:
The first build should be an internal tool plus a thin buyer-facing surface, not a platform.
Scale ads only after:
Do not scale ads from click-through rate alone. Scale from paid intent, sales conversion, and retention.
The EV case for modest ad spend is strong as a disciplined learning strategy. Ten $50 tests are a rational way to buy market information, especially for someone who can build cheaply and wants to avoid months of product work on unvalidated ideas. But the real EV depends less on the $500 and more on the policy after the signal: follow-up speed, manual pilots, sales reps, price testing, and stopping rules.
The best version of the strategy is:
1. Spend $50 to test narrow pain/ICP/offer combinations.
2. Treat ad results as weak evidence until a human follow-up confirms urgency.
3. Convert promising signals into concierge/manual paid pilots.
4. Build only after repeated buyer language or money appears.
5. Scale ads only after CAC/payback and retention are grounded in real customers.
Under that policy, the math supports the strategy. Without that policy, the same $50 tests become a cheap way to generate misleading confidence.