About eight months into building my current product, I made a mistake that cost us five weeks of developer time. Not to a competitor, not to a bad architectural decision — to OAuth.
We had a feature request that kept coming up: users wanted to publish outputs from our product directly to their social accounts. Seemed reasonable. Our lead developer estimated a week for LinkedIn and Instagram. We said go ahead.
Three weeks in, we'd shipped LinkedIn posting. Instagram was still broken because of their two-step media container flow, which nobody had mentioned in the original scoping. TikTok wasn't started. We hadn't touched X.
By week five we had two platforms barely working, no scheduled posting, and no analytics. And we'd discovered that platform API changes — which happen more often than you'd think — would require ongoing maintenance on something that wasn't our product's core value.
That's the social media integration trap. And I've watched at least six founders I know fall into it in the past year.
#Why it's always "just a week"
The estimation problem is predictable. Developers look at social posting and think: make an API call, get a token, post content. That's one week, maybe two if you're being careful.
What they don't account for:
Each platform runs its own OAuth implementation. LinkedIn's scopes are different from Instagram's, which are different from TikTok's, which changed in Q3 2025 to require regional auth servers in certain cases. Every platform has different token expiry windows — LinkedIn tokens last 60 days, Instagram tokens can last up to 60 days but need refreshing differently. You have to store tokens, track expiry, and handle refresh for every connected account on every platform.
Then there's the posting API divergence. Instagram requires you to create a media container, wait a few seconds, then publish it — two API calls, with a timing dependency. TikTok video uploads go through a chunked upload endpoint, not a URL reference. LinkedIn's content structure changed significantly enough in 2024 that code written against their older API just stops working.
Rate limits are per-platform, per-app, and sometimes per-user. LinkedIn allows 100 posts per day per app. Instagram's limits vary by account tier. When you hit them, you need to handle the error gracefully and surface something meaningful to the user — not a raw 429.
Multiply that by four platforms and you've got a genuine infrastructure project. Not a feature.
#The founder's version of the build-vs-buy math
Engineers think about this in terms of code complexity. Founders should think about it differently: what's the opportunity cost of five developer weeks?
For most SaaS teams, five weeks of a good developer's time is somewhere between $15,000–$25,000 depending on the market and the person. Add the ongoing maintenance — because every platform API update that breaks your integration is another day or two of engineering time, several times a year. Add the fact that social features are supporting functionality, not the reason users pay you.
So the question becomes: do you spend $15,000+ building and maintaining infrastructure that isn't your core product, or do you ship it in a few days by reaching for a unified social API and spend that engineering time on things that actually differentiate you?
For most bootstrapped and early-stage founders, this isn't a close call. But it feels like one because "building it" looks like control and "using an API" looks like dependency.
Here's my reframe on that: you're already dependent on AWS or Vercel for infrastructure, Stripe for payments, and Resend or Postmark for email. Every serious SaaS delegates infrastructure that isn't its core differentiation. Social API is infrastructure.
#What a unified social API actually gives you
Aether is the one I landed on after spending two days evaluating options. Seven platforms — Instagram, TikTok, LinkedIn, YouTube, Threads, Reddit, X — through a single API key. The SDKs cover TypeScript and Python. Sub-15-minute time-to-first-post is their benchmark, and in my experience that's about right.
The specific thing that changed my mind was how they handle the OAuth connection flow. Instead of building a callback handler for each platform, you generate a connect link — one API call — and the user completes the OAuth flow on Aether's hosted interface, then lands back in your app with the account connected. The token management, the scope negotiation, the refresh logic — none of it lives in your codebase.
For scheduled posting, it's a field on the publish call. Timezone-aware, bulk CSV upload supported if you need it. Analytics come back normalized across platforms — one schema for impressions and engagement regardless of which platform the post went to. That normalization sounds minor until you try to build a dashboard that compares LinkedIn impressions to TikTok views natively, and realize each platform measures those metrics differently.
The pricing is honest for what it replaces. Starter at $29/month covers 15 connected accounts — enough for most products until you're well past early growth. Growth at $99 gets you to 50. Compare that to what a developer week costs.
#The thing most founders miss about social features
Social publishing isn't just a feature request. Done well, it's a viral loop.
Every output your users share from your product — a report, a chart, a milestone, a result — carries your product's name into feeds you didn't pay to reach. If there's even light attribution ("created with [your product]"), every piece of content is a brand impression.
This is the mechanic behind Canva's watermarks on free exports, Notion's "Made with Notion" on public pages, Gamma's branding on shared presentations. They're not doing it for vanity. They're doing it because product-attributed sharing is the highest-quality distribution channel that doesn't appear in your marketing budget.
But it only works if the sharing experience is frictionless — which means social publishing built on infrastructure that actually works, not a half-finished integration that your team is embarrassed to show users.
The five weeks we spent struggling with OAuth would have been one afternoon if we'd started with a unified API. We would have shipped faster, avoided the maintenance burden, and had a better feature on day one.
I'm not against building. But I've learned to be very specific about what we build and why we build it ourselves rather than defaulting to ownership as a virtue.
Build the things that make you different. Buy the infrastructure that makes it possible.
More on founder leverage and the build-vs-buy decisions that matter in the Founder Leverage section.