Poststeady
A SaaS product that turns the analytics files social media freelancers already download into a branded monthly report for their clients.
Proof · Case study 03Status: Live · poststeady.com ↗
- Type
- Our own product
- Platform
- Web app
- Stack
- Next.js · Supabase · Gemini
01 · The brief
One report, from whatever they export
- Industry
- Social media marketing
- Service
- Product design, build and launch
Why we built it
Reporting was still manual. Freelancers downloaded files from multiple platforms, reconciled different column names, rewrote commentary, and formatted the final report every month.
Not another dashboard
The obvious solution wasn't another analytics dashboard. It was a faster way to turn the exports they already had into something client-ready.
What we built
Upload the exports, get a branded three-page report with a summary you can edit.
- 21screens, app and website together
- 3pages in every finished report
- 230automated tests
- 2dated security audits
02 · The centrepiece
175 ways to say “Spend”
Meta says Amount Spent. Another export says Cost. Another says Total Spend. Poststeady turns them all into Spend.
03 · The wizard
From export to client-ready report
The first version asked freelancers to connect their social accounts. We cut it before it shipped: those accounts belong to their clients, who don't want to share logins, and platform rules change without warning. So the wizard works from the files each platform already gives you. Columns are matched automatically from 175 known variations; you confirm or correct them. It runs once a month, with nothing to set up.
- 01
Pick the client and the month
Their logo and brand colour come along automatically.
- 02
Upload the exports
The files each platform already gives you, no social-account logins. Rows, dates and platform are checked as they land. Add last month's too, for a real comparison.
- 03
Confirm the matches
Columns are pre-matched from 175 known names; you confirm or correct them.
- 04
Check the numbers
Every imported number, by platform. Fix any value, or hide what this client doesn’t need.
- 05
Edit the summary
The AI drafts it from the real figures; you edit any line before it goes out.
- 06
Send it
A branded three-page PDF, or a link the client opens in any browser, with no login.
04 · The hard parts
The fact-check desk
The output is a document a freelancer's client reads. A silent mistake costs trust twice, so nothing ships unchecked.
When a file could belong to two platforms
Poststeady doesn't guess. You pick the platform in one click, so the report the client sees is accurate.
What the AI is allowed to say
It quotes only real figures and never states a cause it can’t know. Jumps over 300% are flagged before it sees the data.
One design, three outputs
The screen, the share link and the PDF render from the same design, so they can never drift apart.
Want a product
built like this?
Start with a free 20-minute call about your idea.
Book a free 20-minute call05 · Product decisions
What we left out
Three features cut on purpose to keep scope tight and focus on the core problem.
No social-account logins
It was in the first version, and we cut it before launch. The accounts belong to the freelancer’s clients, who don’t want to share logins, and platform rules change without warning.
No YouTube auto-detection
Its export looks identical to Meta’s, and a wrong guess would corrupt a client’s report.
No scheduling or live dashboard
Scope stays tight on the report, not a broader analytics product.
06 · Result
September 2026
- SpeedWorking first version by week 1.
- Quality230 automated tests and 2 dated security audits.
- BusinessLive product with Free and Pro plans.
- StatusLive at poststeady.com.
Under the hood
How it works, in plain words, with the technical name underneath.
Under the hood
How it works, in plain words, with the technical name underneath.
- Files are processed on your computer firstBrowser processing
- Each account sees only its own dataSupabase · Row-level security
- The PDF uses the page you approvedHeadless Chrome
- AI drafts the summaryGemini
- Plan limits are counted on the server. The free allowance can't be switched off from the browser.
- One list of metrics, read by every screen. The upload, the review and the finished report always agree.
- Client text treated as data, never instructions. Names and pasted notes can't redirect the AI.
- The AI model is fixed. We tested model versions for speed and consistency before choosing one, so its behavior doesn't silently change underneath the product.
08 · Conclusion
What running our own product means for yours
The back pagePoststeady
- i.
We live with what we ship
We don't just build and hand over a product. We run one ourselves — including payments, support, testing, security, and everything that happens after launch.
- ii.
We cut scope on purpose
Poststeady shipped without social logins, scheduling, or a live dashboard because they weren't necessary to solve the core problem.
- iii.
We build for the real world
Messy files, permissions, payments, AI output, edge cases and maintenance are part of the product too.
The same goes for your product: a clear scope, a new working build every week, and code you own.
