The notice period is the cheapest part of leaving
Founders costing out an agency switch count one thing: the 30 days of notice they'll pay for work they won't get. Thirty days at ₹75,000 is ₹75,000. Annoying, survivable, roughly what people budget for.
That isn't the cost. The cost is the gap — the stretch of calendar where no content ships, no technical ticket gets raised, no outreach goes out, and the pages you published four months ago stop being reinforced by anything new. On most engagements that gap runs eight to twelve weeks, and about half of it happens before you've given notice at all.
SEO punishes a pause harder than any other channel. Paid media stops and restarts the same day; you lose the spend, not the position. Organic compounds, so the gap costs you the work you didn't buy *and* the compounding it would have started. A quarter of silence shows up two quarters later, by which point everyone has forgotten why.
Where the ten weeks actually go
Break the gap into stages and it stops looking like bad luck. Every stage is predictable, and three of the five are compressible if you sequence them properly.
- Notice and selection can run in parallel. They almost never do, because nobody wants to interview replacements while pretending everything's fine. That overlap alone is worth three to six weeks.
- The re-audit is the most infuriating line. You pay a second agency to rediscover the same redirect chains, because the first agency's findings live in a PDF nobody can act on.
- Link and PR work restarts from zero. Outreach relationships belong to the person who built them, and placements in flight tend to die on your notice date.
| Stage | Typical elapsed time | What's happening | Can you compress it? |
|---|---|---|---|
| Quiet disengagement | 2–6 weeks, uncounted | You've stopped trusting the work but haven't said so. You approve slower, they push less, the cadence dies without anyone announcing it. | Yes — pure waste. Set a decision date and hold it. |
| Notice period | 30 days | Full fee, almost nothing shipped. Nobody starts a three-month content programme in their final month. | Partly. Run your search *during* notice, never after. |
| Selection and contracting | 3–6 weeks | Shortlist, calls, proposals, references, MSA. Longer with procurement or a security review. | Yes — overlap it entirely with the notice period. |
| New agency ramp | 2–4 weeks | Access, tooling, a fresh crawl, a re-audit of what the last agency already audited, keyword remap. | Partly — a proper handover pack roughly halves it. |
| First work actually live | Week 8–12 from the decision | First articles published, technical tickets in a sprint, outreach restarted. | No. This is the floor; everything above decides how early you reach it. |
Get these eleven things out before you give notice
This decides whether your gap is four weeks or twelve, and it has to happen while you're still a paying client. The day you give notice, your account moves — formally or otherwise — to whoever covers leavers. Requests that took a day start taking a week. Nobody is being malicious; they're just no longer paid to care.
Ask for all of it as routine housekeeping. A decent agency hands it over without blinking, and one that stalls has just answered your question about whether to leave.
- Ownership of Search Console, GA4, Business Profile, Google Ads and Bing Webmaster Tools. Check you're the *owner*, not a delegated user. A property verified through their tag can leave with them.
- Everything in their tenancy: Looker Studio dashboards, the Tag Manager container, call-tracking numbers, form integrations.
- The keyword map as a spreadsheet, not a slide. URL, target query, current position, intent, status — in cells you can sort.
- The full link ledger: every placement, with live URL, anchor text, date, contact, and whether money changed hands. If they won't say which were paid, assume all of them were.
- Redirect maps and server or edge rules. A migration redirect map is the most expensive document to reconstruct.
- Unpublished drafts, briefs and outlines. You paid for those.
- Access to anything they built in their own account — a staging site, a microsite, a landing-page builder. This is where assets go missing.
- Schema, template changes and code their developers shipped, plus a note on where it lives.
- Raw data exports. Search Console only holds 16 months in the interface — export the full history before it rolls off, and your rank tracker with it.
- Names and roles of everyone who touched the account. You'll learn who to poach, and whether the work was done by the people who pitched.
- A written note on anything unresolved — a pending disavow, a half-finished migration, a manual action.
Why re-baselining quietly ruins your before-and-after
Here's the part nobody warns you about, and it costs more than a month's fee: the new agency will set their baseline on the day they start.
Think about when that day is. After your old agency spent two months coasting, after 30 days of notice where nothing shipped, after six weeks of your own disengagement. Your trailing-90-day organic lead count is sitting at the bottom of a trough you helped dig. Any competent agency beats that number, because recovering ground you already had is easier than taking new ground.
A second distortion runs the other way. Content published in the outgoing agency's final months hasn't matured — pages typically take three to six months to settle. Those wins land in the new agency's first quarterly report. None of this is fraud; it's what happens when measurement starts at a convenient moment. The fix takes twenty minutes.
- Freeze two baselines. One at handover, one from your best 90-day window in the previous twelve months. Judge quarter one against the first, year one against the second.
- Tag inherited pages. Mark every URL that existed before the switch, so any report can separate what they built from what they inherited.
- Compare year-on-year. That strips out both the trough and most of your seasonality.
- Write the lead definition down before day one. If the old agency counted every form fill and the new one counts sales-accepted leads, your before-and-after compares two different things. Setting an SEO baseline is the same exercise, done once, for both.
When ten weeks is worth paying, and when you're about to waste them
Switching is sometimes obviously correct. It's also close to a coin flip — half the founders describing a broken agency are describing a broken brief, a stalled dev queue, or month four.
Switch. The ten weeks buy you something.
- Nobody ever wrote down a baseline. No agreed number means no possible proof of value in either direction. There's nothing to salvage.
- You can't name the person doing the work, and asking produces a shuffle rather than a name.
- The link profile has bought placements in it. That's not a performance problem, it's a liability on your domain. Get the ledger, then get out.
- Two consecutive quarters of flat impressions on your target queries. Impressions move first, usually inside four to eight weeks.
- Reporting leads with traffic and goes vague on leads. Traffic is easy to inflate with queries that will never buy from you.
- They can't explain a ranking drop — not the exact cause, but the diagnostic path. 'Google updated something' is guessing with your money.
Don't switch. You'll import the problem.
- It's month three or four and you're impatient. The most common wasted switch there is. See why clients fire their agency in month four.
- The recommendations are good and nothing gets deployed. That's your dev queue. A new agency inherits the same backlog plus ten weeks of ramp.
- You never defined a qualified lead. They optimised for what you asked to see, which was traffic.
- Your approvals take three weeks. No agency runs a cadence through that loop, and the next one won't either.
- You want the same service for ₹30,000. You'll get a smaller, different service wearing the same name.
- You haven't had the direct conversation. One honest 45-minute call with the person doing the work fixes more accounts than a switch does.
The four-week handover, in order
This is the sequence that turns a ten-week gap into a four-week one. The trick is entirely in the ordering: everything that can happen while you're still under contract, happens while you're still under contract.
Weeks minus-4 to minus-2 — before anyone knows
- Collect the eleven assets above as routine housekeeping. Say nothing about leaving.
- Verify account ownership yourself, in the settings, rather than by asking.
- Export 16 months of Search Console data and your full rank history.
- Write your two baselines down, dated, in a document with your name on it.
Weeks minus-2 to zero — run the search in the dark
- Shortlist three agencies, not eight. You're buying speed, and eight proposals costs a fortnight.
- Give each the same brief: two baselines, your lead definition, your dev-queue reality, three problems in a sentence each.
- Ask for their first-30-days plan as a deliverable. That plan is what compresses their ramp.
- Get the contract signature-ready before you give notice.
Notice day, and the 30 days after
- Give notice in writing with a handover request attached and a date on it. Ten working days is reasonable.
- Ask for a recorded 60-minute handover call between the two agencies. The good ones agree, and it's worth a fortnight of ramp.
- Don't cancel tool access on day one — you want rank history and crawl data still collecting.
- Keep publishing anything already drafted. An approved queue is the cheapest way to cover a gap.
New agency, week one
- Access on day one, not day nine. Chase it — this is the most common two-week slip.
- They confirm your two baselines in writing rather than setting their own.
- They ship something small and real inside a fortnight. Momentum is easier to keep than to restart.
Five clauses that make the next exit cheaper
You're about to sign a new contract. This is the only moment you'll have real negotiating power over how the *next* switch goes, and almost nobody uses it.
- Asset ownership on termination. Every account, dashboard, document, draft and dataset is yours, in your tenancy, from day one. Not 'transferred on request'.
- A handover obligation with a clock on it. The full pack delivered within ten working days of notice, at no extra fee.
- Thirty days' notice, no auto-renewing annual term. Twelve-month lock-ins protect the agency's revenue forecast. They have never protected a client's results.
- Named delivery people, and notification if they change. A silent swap to a junior is the most common quality drop in this industry, and it's invisible without this clause.
- The metric and the baseline, written into scope. One number, defined, with today's value beside it. What belongs in an SEO contract covers the rest of the paperwork.
How we take over an account, and what we tell switchers
Roughly half our conversations start with somebody leaving someone else, so the first ten working days are a fixed sequence: access and ownership, a full crawl, a link audit looking specifically for paid placements, a keyword map rebuilt from your closed deals rather than from search volume, and both baselines agreed in writing before we touch anything.
We freeze two numbers. Your trailing-90-day count of qualified leads from organic search on day one, and your best 90-day window from the previous twelve months. The guarantee runs against the first — beat it inside 90 days or we keep working free until we do — but we report against both, because the second is the honest measure of whether you're actually ahead of where you were before all this started.
Two honest caveats for switchers. If your site carries a manual action or bought placements, the first quarter is cleanup, and we'd scope that as a recovery project rather than pretend a 90-day lead guarantee is fair to either side. And if you're leaving because nothing your last agency recommended ever got deployed, we'll say so on the first call — we'd be walking into the same queue, and ten weeks of your money would buy a nicer-looking standstill. Our pricing is published, so you can do the arithmetic first.