Bibhash Roy

AI Automation Engineer & Entrepreneur. Building @RapidFlowAutomation — helping freelancers & agencies work smarter with AI, agents & workflows. 30+ yrs in tech.

[Bibhash Roy @bibhashroykol]

I would run the demand test first, because this idea came from your own pain. Find developers who already complain online, in their own words, that they found out about user errors days later. Then talk to a few of them about the last time it happened, before you show them your idea. If they describe the pain without you leading them, the problem is real for more people than just you. This is exactly why I have built Valdorro: it stress-tests an idea before you spend money on it. Free beta seats here: https://t.co/2g2V59MVc8

[Bibhash Roy @bibhashroykol]

A yes from someone who already pays your competitor proves almost nothing. 🤷
The test to run depends on where your SaaS idea came from: your own problem, or a market with paying rivals. 🧭
Both tests, step by step 👇 https://t.co/vTxZoCO0Qu

[Bibhash Roy @bibhashroykol]

@mathiasvrb I would search last month's card statement and my inbox for the product name and the word receipt, then cancel every recurring charge I find. The money is the smaller cost. The bigger one is that the product still feels open, so your attention keeps going back to it.

[Bibhash Roy @bibhashroykol]

🧟 A product you stopped building can still charge your card every month.
📝 The clean shutdown, in 10 steps: kill note, export, refunds, read-only, delete.
👇 Full article below
#SaaS https://t.co/SfAwxLFIWU

[Bibhash Roy @bibhashroykol]

@StackShortlist True. I would build and test the export before I announce, and before any refunds. I would also ask one real user to download it and open every file, so nothing is missing when they leave.

[Bibhash Roy @bibhashroykol]

@i_mika_el True. I would stop monthly renewals on the day you announce, at least 30 days before read-only, so the paid month ends first. For annual plans I would cancel on the read-only day and refund only the unused months after that, so nobody is paying for a locked product.

[Bibhash Roy @bibhashroykol]

@mathiasvrb True. In 2020 they were making $2,844 a month, and GitHub sent less than 2% of visitors but ranked third for trial signups. Today they have more than 22,000 paying subscribers, with no outside investors and no money spent on ads.

[Bibhash Roy @bibhashroykol]

🔓 22,000+ customers pay Plausible to run code anyone can download free.
🧪 Inside: when public code helps a small SaaS sell, when it hurts, and a 7-question test for yours.
👇 https://t.co/wRjSgy0y9b

[Bibhash Roy @bibhashroykol]

@IronRedSandHive What I took from it is simpler than brilliance. The homepage had grown one section at a time, whenever a new feature or question needed explaining, until the useful part was buried. The real work was putting the answers back in the order a stranger asks for them.

[Bibhash Roy @bibhashroykol]

Plausible kept its homepage design and changed only the order and two button labels. 🔀
Free-trial signups rose 84% while traffic barely moved. Here is what changed, and a 6-step check you can run on your own SaaS homepage. 🛠️
👇 https://t.co/j2q5bYTEpJ

[Bibhash Roy @bibhashroykol]

@Pulzivo That's how I read it too. Public code lets a buyer verify the no-personal-data claim by reading the tracking script. What they pay for is not the code itself, but someone else handling the server, upgrades, backups, security, and uptime.

[Bibhash Roy @bibhashroykol]

@i_mika_el I don't have that split. The only self-hoster numbers I found are tips from people who kept self-hosting: six $5 tips over about six months in 2020, and about $300 a month by February 2024, when Plausible already had more than 12,000 paying subscribers.

[Bibhash Roy @bibhashroykol]

@Pulzivo About three months. They changed the page in January 2026 and compared that month with February, March, and April, each of which set a new signup record. It was a before-and-after comparison, not two versions of the page running in parallel.

[Bibhash Roy @bibhashroykol]

@lumopages Copy is worth testing, yes. One correction: Plausible did not test the labels before the layout. They published the new section order, the two renamed buttons, and the job titles on the testimonials on the same day. I would not put the 84% on the labels alone.

[Bibhash Roy @bibhashroykol]

@asif_aleem That's the right split. The reorder can get more people to the form, and the label can keep them through it. It fits Plausible's reading. Just treat it as a strong guess: the changes were not measured separately, so we cannot prove which part did the work.

[Bibhash Roy @bibhashroykol]

You can sell a SaaS before you build it.

What most founders skip is the part that keeps it honest: the promises they write next to the price. 🧾

━━━━━━━━━━

💰 In 2016, a solo founder named Mike Taber took about $2,000 in pre-orders from 11 or 12 people for an email follow-up tool that wasn't built yet.

✅ That part worked. People paid real money for software that didn't exist.

📅 The date didn't. He planned roughly 3½ months. He later called it "just an unrealistic timeline" and said the product needed 8 to 12.

He didn't lie to anyone. He set the ship date before he knew the work.

━━━━━━━━━━

That's the pattern I keep seeing on pre-sale pages:

❌ The page gets built first.

❌ The launch date is picked because it sounds good.

❌ The refund rule gets made up when the first buyer asks.

❌ A "testimonial" or an "only 5 left" counter gets added to make it look real.

Each step feels small. Together they give an honest founder a dishonest page.

━━━━━━━━━━

💳 Two more things most founders find out too late.

A Stripe card hold usually lasts 7 days. So "I'll hold your card until launch" does not work for an 8-week build.

💸 A refund promise is only real if the money is still there. In 2015 the FTC settled with a crowdfunding creator who raised more than $122,000, spent most of it on things like rent and a move to another state, and cancelled the project.

🤔 Think about this tonight: if your first buyer asked for their money back tomorrow, could you pay it the same day?

━━━━━━━━━━

Tomorrow I'm publishing the full breakdown. Inside:

📝 The 7 decisions to make, in order, before your pre-sale page goes live

📅 A ship-date formula with a worked example

💸 The exact refund line, and where to keep the money until launch

🪧 A fill-in-the-blanks wording block for the page

🚫 4 things on a pre-sale page that cross the line into lying

👀 Watch this space.

What has stopped you from asking for money before building: the fear of a no, or the fear of a yes?

#SaaS #BuildInPublic

[Bibhash Roy @bibhashroykol]

@ankurmans Yes. They did not hire a designer, and they kept the same colours, fonts, and images. The work was a reorder and two renamed buttons, which costs almost nothing next to buying another channel. The conversion was already sitting in the traffic they had.

[Bibhash Roy @bibhashroykol]

@Saksham_ux Yes. A button label is a promise about the next page. Get started could mean a tour, a demo, or a form, and a visitor who expected a tour may leave when a form appears. The label has to name what comes next, and the next page has to keep that promise.

[Bibhash Roy @bibhashroykol]

@acesaasdemos Yes. Traffic from visitors who were not already logged in rose only about 2%, but the share who started a trial went from 2.65% to 4.80%. The audience was already there. The lift came from the ask, not from finding more people.

[Bibhash Roy @bibhashroykol]

@RonnyBruknapp Yes. New paying customers followed about a month later, matching the 30-day trial: 652 in February, 953 in March, and 1,156 in April, their best month in seven years. Cancellations stayed about the same, so the extra trials included real buyers, not only easier signups.

[Bibhash Roy @bibhashroykol]

@SantanaFrost17 Plausible can't say for sure, since all four changes shipped together with no A/B test. But the numbers lean to the copy: about 22% more people reached the signup page, while completion jumped from 38% to 57%. "Start free trial" told them a form was next.

[Bibhash Roy @bibhashroykol]

@hellozainstudio Agreed, the feature list under the hero is how visitors saw the product before the long text. The numbers add a twist: only about 22% more people reached the signup page, while completion rose from 38% to 57%. Plausible's best guess is renaming Get started to Start free trial.

[Bibhash Roy @bibhashroykol]

@mailrocketIN Good addition. Check 4 is why: Google's spam rate is per domain, so one Report spam click on a campaign can put OTPs over the 0.3% line. The audit already uses https://t.co/PaT1F3gO6z for auth mail; a separate marketing subdomain keeps that score isolated.

[Bibhash Roy @bibhashroykol]

Users signed up and never came back? Check Spam before you blame your product. 📩🚫

I read Google's own sender rules so you don't have to. 5 checks, $0, about an hour. 👇 https://t.co/k9VPtgjEP4

[Bibhash Roy @bibhashroykol]

@gmc_sammy Of course, Sam, sorry I missed this one. Ask away, here or in DMs. And since email is your world, I'd love to hear whether the list-over-followers point matches what you see with clients.

[Bibhash Roy @bibhashroykol]

https://t.co/yLmdhzOkMi