Month four isn't random. It's the arithmetic of a first quarter.
Agencies talk about month-four churn like it's weather. It isn't. Three clocks converge in the same 30 days, and once you see them lined up the cancellation stops looking like a surprise and starts looking scheduled.
The first clock is the contract. Most Indian SEO retainers carry a three-month minimum term, ours included. Month four is the first month a client is allowed to leave, so it collects every doubt accumulated since kickoff.
The second is the invoice count. At ₹75,000/mo, three invoices is ₹2,25,000 — ₹2,65,500 once GST lands. That has stopped being an experiment and started being a visible line in the P&L that a co-founder or a CFO will ask about by name.
The third is the patience budget. Almost every founder is told SEO takes three to six months. What they hear is three.
Underneath all three sits the shape of the work. The heavy, unglamorous spend lands in months one and two — audit, baseline, technical fixes, keyword mapping. The return compounds from month five onward. Month four is the trough between them: structurally the worst possible month to judge an SEO engagement, and the month the contract invites you to judge it.
The timeline of a typical break, from both sides of the table
Here's the sequence we see when we inherit an account from another agency. Neither column is a lie. Both parties are describing the same four months accurately. That's what makes the argument unwinnable when it finally happens.
- The agency is reading leading indicators and they are genuinely moving.
- The client is reading a lagging indicator and it genuinely hasn't.
- Nobody agreed, in writing, which one the engagement would be judged on.
| Month | What the agency saw | What the client saw |
|---|---|---|
| 1 | Full crawl, analytics audit, keyword mapping, baseline set, roughly 40 hours of setup before a single new page ships. | A PDF, a spreadsheet and a call. Nothing on the site changed. |
| 2 | Redirect chains cleaned, canonicals fixed, three template-level issues resolved, first two pages published. | Two blog posts about topics they didn't ask for. Traffic flat. |
| 3 | Impressions climbing on target queries. New pages entering positions 20–50. First outreach replies landing. | Search Console looks busier. The phone rings the same amount it did in January. |
| 4 | Content stack compounding, two links placed, average position improving week over week. This is going well. | ₹2.25 lakh spent. Zero attributable leads. Time to have the conversation. |
The reporting gap, and the four-month delay built into it
There is a mechanical delay between an SEO action and a rupee. It isn't agency spin; it's the sequence a page has to survive. It gets crawled. It gets indexed. It climbs to a position that earns clicks. It gets clicked. It converts. Then, if you sell to businesses, a human takes 30 to 90 days to sign.
Stack those and the honest picture for a competitive site is leading movement inside 4–10 weeks, first organic leads in months four to eight, and revenue you'd defend in a board meeting from months six to twelve. Those are ranges rather than promises, because two variables dominate them and neither belongs to the agency: how competitive your keyword set is, and how long your sales cycle runs.
This is the table we now put in front of every prospect before they sign, because a client who saw it in week one doesn't fire anybody in week sixteen.
| Signal | Typically moves | What it proves |
|---|---|---|
| Fixes deployed, pages shipped | Weeks 2–4 | The agency is working. Nothing more. |
| Indexation of new pages | Weeks 3–6 | The pages are eligible to rank at all. |
| Impressions on target queries | Weeks 4–10 | Google understands what the pages are for. |
| Average position on target queries | Weeks 8–16 | The pages are competitive, not just present. |
| Organic clicks on commercial pages | Months 3–6 | Searchers are choosing you over the alternatives. |
| Qualified leads from organic | Months 4–8 | The business is getting paid. This is the number. |
| Closed organic revenue | Months 6–12 | Mostly a function of your sales cycle, not the SEO. |
The failure happened in week one
By the time the cancellation call happens, there is nothing useful left to argue about. Both parties have data, both sets are real, and neither can settle the question because the question was never defined. The post-mortem always lands in the same place: the kickoff.
A kickoff that prevents month four produces five artefacts. Not a deck. Five specific written things, and if your agency didn't hand you these, the month-four fight was scheduled the day you signed.
- A frozen baseline number. Your trailing-90-day count of qualified leads from organic search, written down before anyone touches anything. Not traffic. Not rankings. The number that pays salaries. We cover the mechanics in how to set an SEO baseline.
- A definition of 'qualified'. Does a brochure download count? A WhatsApp enquiry? A form fill from a student? Two reasonable people will define this differently and both will be certain they were obvious.
- A named judgement date. Ours is day 90 against the baseline. Whatever yours is, it should be a date on a calendar, not "once things pick up".
- A written list of what will look bad, and when. Month one will look like nothing happened. Month two will produce content you didn't request on topics you don't rate. Month three will show impressions and no leads. Say it in week one and it reads as competence in week twelve.
- An agreed answer to 'what if it doesn't work'. This is the question that separates the field. If the answer is a shrug and a renewal invoice, you already know how month four ends.
The half of it the client owns
This part gets left out of most agency post-mortems, for obvious commercial reasons. It shouldn't be, because the arithmetic is brutal.
Take a three-pages-a-month content plan. Each page needs a brief approved and a draft approved. If a brief sits nine days waiting for a founder who is genuinely busy, and the draft sits another eleven, you've added about three weeks to that page's clock. Do that across a quarter and you've lost roughly a month of the compounding window — landing exactly where you planned to judge the results.
The dev queue is worse, because it's invisible. A template fix specified in week three that ships in week fourteen did nothing for the quarter you're assessing. The agency reported it as "delivered" in week three, which is true of their side and useless to yours.
Then there's tracking. If organic leads aren't tagged in your CRM, the lagging indicator can't be measured at all, and month four becomes two people arguing about a number that exists nowhere. No deck has ever settled that fight.
- A 72-hour approval SLA, both ways. Ours applies to us too. Anything unanswered after 72 hours gets escalated or the agency proceeds on the documented default.
- A standing developer slot. Two hours a fortnight, booked. Not "we'll raise a ticket". The single most common cause of a technically-competent SEO engagement producing nothing.
- One SME hour a month. Someone who actually knows the product, on a call with the writer. It is the difference between content that ranks and content that reads like it was written by a stranger, because otherwise it was.
- A source field on every form, and organic tagged in the CRM. Ten minutes of setup that decides whether month four is a conversation or a fight.
- One decision-maker. Content reviewed by a committee of four arrives late and averaged into blandness. Both of those cost rankings.
The two-report format that prevents the fight
One report cannot serve both audiences, and every agency that tries produces the same hybrid: a monthly deck where good news about impressions sits next to silence about leads. It satisfies nobody and it buries the one number that matters under eleven that don't.
Split it. Two documents, different cadences, different audiences, and a hard rule that they never merge.
| Report A — monthly work report | Report B — quarterly business report | |
|---|---|---|
| Audience | Whoever manages the agency day to day | Whoever signs the invoice |
| Length | One page. Five numbers. | Two pages, mostly prose |
| Contents | What shipped, what's blocked, indexation, impressions and average position on the target set, referring domains | Baseline versus today on qualified organic leads, what changed and why, what we got wrong |
| Verdict | None. It's a status update. | A written recommendation: continue, change, or stop |
| Failure mode it kills | "I have no idea what they've been doing" | "They showed me traffic charts for a year" |
The clause that makes Report B honest
Report B has to permit the word stop. An agency that has never once recommended a client reduce or end a retainer is not an agency with unusually good clients; it's an agency whose quarterly review is a renewal pitch wearing a hat.
We write it into the engagement: if the leading indicators are flat at 90 days and we can't explain why, we say so in writing and we tell you whether to continue. Sometimes the honest answer is that the site is fine and the product has no search demand, which is not an SEO problem and cannot be fixed by buying more SEO.
Why the monthly report should be boring
Five numbers, same five every month, no design. The moment a monthly report becomes a designed artefact, someone is spending three of your paid hours in Canva and the report starts optimising for how it looks rather than what it says. How to read an SEO report covers what to strip out of the one you're currently getting.
When firing in month four is genuinely the right call
Month four is a bad moment to judge outcomes and an excellent one to judge process. Some engagements should absolutely end there. The test: are the leading indicators dead, or are you impatient for the lagging one?
Fire them
- No baseline exists. Nobody wrote down where you started, so nobody can ever prove you moved. This alone is enough.
- Impressions on your target queries are flat at 90 days. Not leads — impressions. If Google hasn't noticed the work after three months, the work isn't working.
- Nothing shipped. Count the pages published and the technical fixes deployed. If month three looks like month one, you bought a subscription to meetings.
- The links came from a network. Ask where. "Our own network of sites" is a private blog network and the penalty risk sits on your domain, not theirs.
- Reporting leads with traffic and won't discuss leads. Traffic is trivially inflatable with irrelevant queries. Refusal to talk about leads is a decision, not an oversight.
- You still haven't met anyone who touches your account. Four months is long enough.
Don't fire them
- Revenue hasn't moved but impressions and positions have. That's the system working in the correct order. Switching now means paying the setup cost twice — the real cost of switching SEO agencies puts a number on it.
- A competitor with a ten-year domain still outranks you. They were always going to for a while.
- The content isn't what you'd have written. Fair criticism, fixable in one call. Not a firing offence unless it's factually wrong.
- You read a thread saying SEO is dead. It said that last year too.
What we changed on our side, and what it costs us
We rebuilt the front of our engagement around this specific failure, and it isn't charity — month-four churn is the most expensive thing that can happen to an agency. You absorb the whole setup cost and earn three invoices against it.
So: we freeze your trailing-90-day count of qualified leads from organic search on day one, before anything else happens. If we haven't beaten that number in 90 days, we keep working free until we do. We don't promise a ranking position for a keyword — nobody controls Google's index, and anybody promising position one is either buying links or buying time.
That guarantee is why we only take three clients a month. You can't carry that exposure at volume, and an agency offering it across forty accounts either isn't honouring it or has defined the metric so loosely it can't be failed.
We're month-to-month after the first quarter, 30 days' notice, and you keep every asset. So month four is still the month you're allowed to leave — we've just tried to make it a month where you have a real number in front of you instead of a feeling. See how the guarantee works or what it costs.