Taking your product back from an agency

The decision

Every agency relationship ends. Some end on schedule when you hire your own team, some end when the budget does, and some just break down. Whichever way yours ends, the task is the same: getting the code, the accounts, the infrastructure, and the knowledge into hands you control — while the product keeps running for live users. Done well, it's a few weeks of structured work. Done badly, it's a product nobody can deploy, accounts nobody can open, and a rebuild that costs more than the original build did.

The questions that actually determine it

What do you already control today — before anything is negotiated?

Whose account owns the code repository? The cloud infrastructure? The domain? The app-store listings? Whatever you don't control is leverage the agency holds — usually not out of malice, just because it was faster to set up that way in week one. Answer this today, in writing, item by item. The gap between what you think you own and what you actually own is where handoffs die.

Is the relationship ending warm or cold?

A cooperative ending gets a planned handoff: weeks of paid overlap, recorded walkthroughs, patient question-and-answer. A hostile ending gets a different plan entirely — ownership first, speed over completeness. Be honest about which one you're in, because a warm-ending plan executed in a cold ending leaves you halfway through the transfer when the agency stops answering email.

Who receives the knowledge?

A handoff needs a receiver — your new engineer, an independent engineer hired for the purpose, or a second vendor. Documentation without a reader is a ritual: the agency writes pages nobody checks, you file them somewhere safe, and six months later they answer none of the questions that actually matter. If nobody on your side can receive, fix that before you fix anything else.

What must keep running while ownership moves?

If you have live users, this is heart surgery, not a file transfer. Ask what breaks if nothing is deployed for two weeks: certificates that expire, payment flows that fail, support escalations nobody can act on. A live product needs a keep-the-lights-on lane running alongside the transfer lane, with a named owner for each.

Your options, with honest costs and risks

The planned handoff

Warm relationship, weeks of runway. You pay the agency for 2–6 weeks of overlap at day rates — budget ₹1.5–6 lakh ($2,000–8,000) all-in — with the transfer scoped as named deliverables, not "support during transition." The risk is paying for overlap and receiving vague availability. Buy deliverables with acceptance criteria, and tie the final invoice to the last one.

The cold handoff

Relationship over, goodwill gone. Work a strict priority order: first, account ownership — repos, cloud, domain, app stores; second, a complete copy of the code with its history, plus data and backups; third, credentials rotated so old access dies; fourth, whatever knowledge you can extract in whatever form you can get it. This path is cheap and fast, and you inherit a codebase without context — your next engineer's first month becomes archaeology. Keep the anchor in view: a rebuild after a botched handoff runs 60–120% of the original build cost plus 3–6 months. Everything below is cheaper than that number.

The assisted handoff

An independent engineer receives and audits the handoff on your behalf, for ₹60,000–2 lakh ($750–2,500). They verify the code actually builds, the deploy actually works, a backup actually restores, and the documentation matches reality. For a non-technical founder this is the best money in the whole process, because you cannot verify a handoff you can't read.

The inventory

Whichever path you're on, the same things must transfer:

• Code with full history — repository transfer, not a zip file. History is where the "why" lives.
• Account ownership — repository organisation, cloud account, domain registrar, app-store listings, all in accounts you own with billing on your card.
• Credentials — every password, API key, and signing certificate, transferred and then rotated.
• Infrastructure — servers, databases, DNS, all mapped in writing, your access confirmed by logging in yourself.
• Third-party services — payments, email, analytics, monitoring, with billing moved to you before the agency's card expires.
• Data and backups — current copies, plus one restore actually tested. An untested backup is a hope.
• The deploy runbook — written steps from code change to production, verified by someone following them cold.
• Recorded walkthroughs — screen recordings of the architecture tour and the "how do I…" sessions. Cheap to make, gold later.

What I'd recommend

Start the inventory today, regardless of how healthy the relationship feels. It costs nothing, offends nobody, and every item you secure now is one that can't become leverage later. I've sat on both sides of this handoff — including taking a product back from an outsourced team at my own company — and the founders who suffer are never the ones who started too early.

Fix account ownership this week, not at the end. Repos, cloud, domain, and app-store listings move into your accounts while the relationship is warm. If things are tense, do it before announcing you're leaving — ownership transfers requested after "we're moving on" have a way of slowing down.

Pay the overlap by the day, not lump-sum. Day rates keep the agency's attention on finishing; a lump-sum handoff fee is an invitation to stretch thin effort across the agreed weeks.

And hold one acceptance test above every document: someone on your side deploys a small change to production while the agency watches — watches, not types. Until that has happened, the handoff is not done, whatever the folder of PDFs says.

When this doesn't apply

• Disputed IP or unpaid invoices. If the agency claims ownership of the code or you're behind on payment, this is a legal matter first. Lawyer, then checklist — running the checklist mid-dispute can weaken your position.
• Regulated data. Health, finance, and similar domains constrain how data may move and who may hold credentials. Get compliance advice before the transfer, not after.
• White-label platforms. If your "product" runs on the agency's proprietary platform, there may be far less to hand over than you think — the real question is contractual exit terms, not code transfer.

In any of these, don't self-serve from this guide. Get a human expert on your specific situation.