You shipped something. Maybe a few people paid. Maybe nobody did. Either way, a new idea is already open in another tab. In 2026 that tab is dangerous. AI makes product two feel free. It is not free. It costs the hours that could have made product one work.
Indiecity is for one person or a team under twenty. The decision hangs on who paid, what work you skipped, and whether product two is a real bet or an escape from selling.
Portfolio and focus are different jobs
Focus means one offer, one buyer type, and most of your week on making that offer clearer, selling it, and keeping the people who paid. You still ship. You do not open a second company because Tuesday felt slow.
Portfolio means more than one product that can live at the same time. That only works if dead products die, and live products do not each need a full founder. If every product needs you for sales, support, and roadmap, you do not have a portfolio. You have a stack of unfinished jobs.
People mix the two because they copy public multi-product founders and skip the part where one product actually stuck. The method is search, then double down. Not half-attention forever.
Why product two looks smart right now
Building got cheap. Pieter Levels said it in June 2026: everyone can build apps with AI. Almost nobody has an audience, ad money, or a free way to get attention. Product two is the same trap with a new name. You feel productive because code is moving. You avoid the inbox where product one is waiting. You tell yourself the market for product one is dead when you have not run the sales work yet.
How you get your first 10 paying customers without ads is still the same job on the first product: a price, names, and an ask. If product one launched to silence, start there, not in a new repo.
What Levels actually did with many products
In 2014, Levels set a public rule: launch one startup a month for a year. The point was finishing and shipping, not collecting half-built repos. Later he described the pattern in plain words: shotgun. You shoot a lot of projects and see which sticks. One stuck as Nomad List. Remote OK grew out of that success. The philosophy is not “never focus.” It is “find the one that sticks, then grow it.”
If you copy only the shotgun and skip the stick, you get a pile of launches. That is not Levels. That is starting over every time a product needs sales.
What focus looks like when it is working
Paul Graham wrote in 2009 that nothing kills startups like distractions, and that the underlying cause of death is often lack of focus. He also said it is better to make a few users love you than a lot of people semi-happy. That still holds when shipping is cheap. Love shows up as payment, retention, and introductions. Semi-happy shows up as polite DMs and empty dashboards.
Nothing kills startups like distractions.
In Do Things that Don't Scale, he argues for a deliberately narrow market first: keep the fire small so it gets hot. That is double down language. You do not widen into a second product while the first fire is still cold. You make a few people care, then grow from there.
Focus work this month is boring on purpose: same buyer, same offer, more conversations, clearer price, fewer features that nobody asked to pay for.
The decision rule
Answer these in writing. One page. No roadmap slides.
- Has anyone paid product one in real money (card, invoice, transfer, UPI, cash)?
- In the last 30 days, how many humans did you ask to buy with a clear price?
- If product one has payers: what is the next growth lever you have not done for two solid weeks (outreach, SEO on one phrase, price test, retention fix)?
- If product one has zero payers: did you talk to 20 people who should care, or only polish the product?
- Can product one run for seven days without you coding a new feature?
- Is product two a different buyer and a different paid outcome, or the same escape with a new logo?
Then apply the rule:
- People pay, and you have not run a growth lever for two weeks: double down. Do not open product two.
- Nobody pays, and you have not done 20 real conversations with a clear ask: double down on learning and selling. Do not open product two.
- Nobody pays after 20 real conversations and a clear ask: kill or change the offer. That is the same fork as zero paying customers after launch. A second product is allowed only after you write what failed in plain words.
- People pay, and product one is mostly automated or on a maintenance schedule: a second product can use spare capacity. Still ship one bet at a time until it sticks or dies.
- You want product two because selling product one feels bad: that is not a strategy. That is avoidance. Stay on product one until the market answers.
Write the answer as one sentence: “I am doubling down because ___” or “I am starting product two because ___.” If the blank is “I’m bored” or “launch looked quiet,” rewrite it. The reason has to be about payers, conversations, or spare hours.
What double down means this week
Double down is not “rebuild v2.” It is more effort on the same offer.
- List 30 names who match the buyer. Send ten human notes with a price and a 15-minute ask.
- Talk to every paid user. Ask what almost made them cancel, and who else has the same problem.
- Change one thing on the page using their words, not a redesign.
- Ship only the fix that unblocks a paid user or a booked call.
- Count sent, replied, booked, paid. That is the scoreboard.
Hey [name]. I’m still on [product one outcome] for [who]. Price is [price]. I’m not building something new this month. If [pain in their words] is still on your plate, can we do 15 minutes this week?
If you already have a few payers, doubling down can mean raising price, cutting a free tier that trains people not to pay, or writing one SEO page for the exact phrase buyers used. It still is not a second product.
When a second product is allowed
Start product two only after one of these is true.
- Product one is dead with evidence: real conversations, a real ask, and no path you still believe in. Write the post-mortem in ten lines before you open a new repo.
- Product one pays and can run without your full attention (automation, a simple maintenance block, or a clear “hours per week” cap you respect).
- Product two reuses an audience you already have (same people, same channel) and you can still protect a block of hours for product one if it still needs sales.
When you start product two, keep the Levels constraint: ship fast, put a price on it, and kill it if nothing sticks. Do not run product two as an infinite side quest while product one starves. If product two is the new main bet, say that out loud and put product one on maintenance.
Ramen profitable still matters. Graham’s definition is simple: enough to pay the founders’ living expenses. If product one is close to that, protect it. A second product that delays ramen profitability by six months is expensive entertainment.
A two-week decision plan
- Day 1. Write the decision rule answers. Freeze product two. Put the freeze on a sticky note if you have to.
- Days 2 to 6. Ten personal asks on product one. Track sent, replied, booked, paid.
- Day 7. Read the scoreboard. If you have payers, plan two weeks of the next growth lever only. If you have zero talks, you did not run the test. Extend the freeze.
- Days 8 to 12. Either double down on the lever (outreach, price, retention, one channel) or, if the market already said no, write the kill note and only then open product two as a time-boxed bet.
- Day 14. One sentence again: double down or second product, with a reason that mentions money or conversations. Not mood.
If you kill product one, keep the list of people who talked to you. That list of people is the part worth keeping. Do not throw away the only hard-won part.
What to stop doing
- Starting product two because product one needs sales work.
- Calling five landing pages a portfolio.
- Copying multi-product founders without copying the “what stuck” half.
- Building an AI factory before anyone paid for version one.
- Using “I’m validating in parallel” to avoid one clear ask.
- Keeping a dead product on life support so you never have to admit it failed.
- Opening a new idea every time a prospect says not now.
Pick the path, then do the week
Portfolio vs focus is not which Twitter bio looks cooler. It is whether your hours this month match a rule you can defend. Most people under twenty headcount should double down until the market pays or clearly refuses. Second products are for dead ends with evidence, or for spare hours after something already works.
If you double down and the story is real, write it down. When the product and the name are yours, put it on the map. You can join Indiecity while you run the two weeks. A community does not make the call for you. The decision is still yours: one hard path, run long enough to learn.
Until the rule says otherwise, close the second tab.
FAQ
Questions people get stuck on
Is a portfolio of small products always better for solo founders?
No. A portfolio works when each product either dies fast or can run with little of your time. Running three half-sold products is not a portfolio. It is three unfinished jobs. Pieter Levels’s public pattern is shotgun shipping, then doubling down on what sticks. The product that sticks is the point.
I am bored of product one. Is that enough reason to start product two?
No. Boredom often shows up right when the work shifts from building to selling. That is the hard part, not a signal that the idea is dead. Write down the last ten sales actions you took. If that list is empty, you are not bored of the product. You are avoiding the job.
What if product one has users but nobody pays?
That is still product one. Fix price, buyer, and the ask before you open a new repo. Free users teach you about free users. A second product will not make the first one charge for itself. Put a price on the page and ask ten people who already use it to pay.
How long should I double down before I kill it?
Long enough to run real conversations and a clear ask, not long enough to polish forever. If 20 people who should care will not give you 15 minutes, or nobody pays after a plain price, change the offer or the buyer. Kill after the market answers. Do not kill because week two felt quiet.
Does AI change this in 2026?
It changes how cheap it is to start product two. It does not change how hard it is to get money and attention. Pieter Levels wrote in 2026 that everyone can build apps with AI, and almost nobody has a way to get customers. Shipping another app is easier than it has ever been. That makes focus more valuable, not less.
When is a second product the right move?
When product one is dead after real tries, or when product one already pays and can run without your full week. Then a second bet is capacity, not escape. If product one still needs you for every sale and every support reply, a second product steals the hours that would have grown the first.
Can I keep product one as a side income while I build product two?
Only if product one is mostly automated or scheduled. Paul Graham’s old warning still holds: profitable side work is a distraction when it keeps interrupting the thing with more upside. If product two is the main bet, put product one on a maintenance schedule. Do not let every support ping reset your week.
What does double down mean day to day?
More selling and retention on the same offer, not more features for fun. Talk to buyers. Ask for money. Fix the one blocker paid users hit. Raise price if people say yes too fast. Write in the rooms where buyers already complain. A new dashboard is not doubling down.
Is Levels-style multi-product only for people with an audience?
Audience helps, but the method is not “launch ten brands at once with no follow-through.” Levels forced launches, killed what did not stick, and kept growing the ones that did. Without follow-through on the winner, you only collect landing pages. Audience is not a substitute for that rule.
