Every decision is a hypothesis. We built the notebook for it.
optimiser.pro is an experimentation OS for product-ops teams — run A/B tests, ship feature flags, and watch funnels without a $40k/yr enterprise contract.
↘reads like a lab journal, ships like a deploy tool
Experiments · live
14across 3 products
Avg. lift shipped
+4.2%last 90d
Time to significance
6.3days median
Flags in prod
822 stale
The state of experimentation · 2026
The tools cost more than the experiments.
Most product-ops teams are stuck choosing between a $40k Optimizely contract, a clunky homegrown setup, or no experiments at all. We think there's a fourth option.
1.
Enterprise tools price out the people who need them most
The teams with the tightest margins — indie products, bootstrapped SaaS, in-house growth squads — are the ones who most need to test before shipping. $40k annual seats don't scale down.
2.
Homegrown experiment frameworks rot in six months
Every product team builds its own. None of them get maintained. The person who wrote it leaves. The CSV of results sits in someone's Dropbox.
3.
"Just use GA4" isn't an experimentation strategy
Event analytics tells you what happened. It doesn't tell you what to ship next. Significance, holdouts, and feature flags need their own layer.
4.
Dashboards optimised for data scientists, not PMs
The tools that do exist are built for statisticians. PMs want to know: ship it, kill it, or keep running? Not read a confidence-interval essay.
We wrote this from the other side of the keyboard.
optimiser.pro was built by a product-ops team (sanjow.ventures) for its own products — topup.ae, wepdf, checkin. We were the team with a homegrown setup that rotted. We're the ones who couldn't justify Optimizely.
So we built the thing we wished existed. Fast, opinionated, priced like you're running experiments — not signing procurement deals.
“We had three different experimentation tools stitched together. The CSVs lived in Dropbox. We shipped two bad variants in a month because nobody knew which one had won. optimiser.pro fixed that in a week.”
— srikant, product ops @ sanjow
Four steps · four minutes
Hypothesis toshipped variant, without ceremony.
A CLI, a dashboard, and a JS snippet. That's the whole surface area. You don't need to learn a new statistical framework — just describe the change and watch it run.
01 · WRITE
Write the hypothesis
Two lines in the CLI. Name the experiment, describe the variant. We generate the feature flag and tracking automatically.
# write it like a note to yourself
$ optimiser new "cta_copy"
variant: "Start free →"
hypothesis: "action verb lifts CTR"
02 · SHIP
Ship the flag behind a line of code
Wrap the variant in a flag check. One import, one function call. Works with React, Next.js, vanilla JS.
Real-time lift, confidence intervals, and a plain-English verdict. No spreadsheet exports. No "talk to our data team."
# live view, updated every 60s
$ optimiser watch cta_copy
lift: +4.8% · p=0.02
verdict: ship it
04 · ROLL
Roll to 100% (or kill it)
One command. Gradual rollout, kill-switch, holdout group. All the grown-up controls. None of the grown-up pricing.
# ship the winner
$ optimiser ship cta_copy
✓ rolled to 100%
✓ losing variant archived
Product tour · the dashboard
Dense when you want it. Quiet when you don't.
Every surface is designed for the PM's question — “can I ship this?” — not the statistician's essay. Significance is shown, not buried. Annotations are first-class. Kill-switches are one click.
sanjow / topup.ae·experiments · active
live · updated 4m ago
Activeexperiments
14 running · 3 ready to ship
Running
14
↗ +3 this week
Ready to ship
3
95% CI reached
Avg. lift · 7d
+4.8%
↗ vs +2.1% last wk
Losing
2
↘ kill candidates
Experiment
Lift
Significance
Status
Note
exp_04 · topup_cta_copyvariant B · 50/50 split · 8d
We shipped 14 experiments in our first month on optimiser.pro. Our previous stack got us 3 in six months. Same team, same roadmap — just no friction anymore.
SS
samresh shrivastava
art director · studio krygn
“
The CLI is the killer feature. I can spin up an experiment faster than I can open a Jira ticket to ask our data team to build one. That's the whole pitch.
RV
riddhi varma
co-founder · homelore
“
Finally a tool that treats "ship it or kill it" as a first-class question. We stopped having 45-minute meetings about whether a lift was real. The verdict is on the dashboard.
SK
srikant k.
pm · sanjow ventures
FAQ
The questions we get most.
If yours isn't here, ping us on hello@optimiser.pro and we'll answer in the open.
Not really. PostHog is a Product OS — analytics, replay, flags, experiments, surveys, the full suite. optimiser.pro is narrower on purpose: experiments and flags, with just enough funnel + cohort tooling to run them well. If you need the full PostHog suite, use PostHog. If you just want to run experiments without renegotiating your annual contract, use us.
Because every experiment you run is a hypothesis, and every result is a finding. The lab-notebook frame reminds teams that the dashboard isn't a spreadsheet — it's a record of what you tried, what happened, and what you decided. Dashboards that look like Google Sheets encourage teams to treat experiments like reports. We want them treated like science.
You pay for the events we track — pageviews, conversions, custom events. Not for seats, not for workspaces, not for the size of your logo. First 100k events every month are free forever. After that, it's €5 per 100k on the Operator plan, €3 per 100k on Studio. No surprise invoices. You can see your current bill live on the dashboard.
No. The verdict column on every experiment is plain English: "ship ready", "keep running", "kill candidate". The statistical machinery (Bayesian by default, frequentist available) runs in the background. PMs can read the dashboard and make shipping decisions without asking anyone. Your data team can still open the raw significance data if they want to audit.
You can export everything — experiments, flags, events, verdicts — as CSV or JSON any time. If you cancel, we keep your data for 30 days in case you change your mind, then it's purged. We never sell your data, never train AI models on it, never share it with third parties. It's yours.
Yes, on the Studio plan. We ship a Docker image that runs the full optimiser.pro stack — experiments engine, dashboard, event ingestion — on your own infrastructure. Most customers don't want this (managed is easier), but it's there for regulated industries and paranoid CTOs.
Three products under the same philosophy. optimiser.pro (this site) is the experimentation OS. optimiser.world is a community marketplace for AI agents and skills. optimiser.art is a premium creative brand consultancy. Different products, different audiences — same belief that the best businesses are built one well-run experiment at a time.
Ready to ship
Stop deliberating. Start experimenting.
Free forever for solo builders. 14-day trial on every paid plan. No credit card, no “jump on a call”, no procurement forms.