Your first 90 days with an SEO agency

Week one: you get access, not output

The first week looks alarming if you're expecting deliverables. Nothing gets published. No rankings change. What actually happens is handover — and the speed of that handover sets the pace of the entire quarter.

The list below is what a competent agency asks for on day one. If they ask for less, they're planning to guess. Note that CRM access is on it: the number being measured is qualified leads, and leads live in your CRM, not in your analytics.

• Google Search Console — the domain property, not just one https://www. variant. Old properties too, if you've migrated.
• GA4 admin, plus whatever tag manager container is actually firing.
• CMS and staging access, and the name of the person who can deploy to production.
• DNS or CDN access, or at minimum the name of whoever holds it.
• Server or CDN logs, if your stack keeps them. Most Indian mid-market stacks don't, and that's worth knowing early.
• CRM read access and the definition of a qualified lead your sales team already uses.
• History — past migrations, past penalties, past agencies, and any period where traffic changed shape.

The one thing week one must produce: a frozen baseline

Before anything on the site changes, the number you'll be judged against has to be written down and agreed. Not estimated later. Written down, in the contract, on day one.

Ours is the trailing 90 days of qualified leads from organic search. Yours might be organic revenue, or booked demos, or footfall enquiries. The metric matters less than the freezing. An unfrozen baseline drifts — definitions loosen, seasonality gets re-explained, and by month six nobody can prove anything either way.

1. The metric, named exactly. "Organic traffic" is not a metric; "non-brand organic sessions that submitted the enquiry form" is.
2. The source of truth — which system's number wins when two disagree.
3. The window — trailing 90 days ending the day before work starts.
4. The definition of qualified, and who decides in a borderline case.
5. The exclusions — branded search, paid traffic, existing customers, obvious spam.
6. The read date, so nobody negotiates the deadline in month three.

What deliberately doesn't happen, and when it does

Most onboarding anxiety comes from silence that looks like neglect. Here's the schedule of absences, and the reason behind each one. If your agency is behind these dates, ask why. If they're wildly ahead of them, ask harder.

Weeks two to six: the fixes nobody will compliment you on

This stage produces the least impressive status report and frequently the highest return in the whole engagement. On an older Indian site, the single most common finding is a replatform from three years ago whose redirects were never fully mapped — traffic you already earned, quietly leaking into 404s.

None of it is glamorous. All of it caps everything that comes after.

• Redirect chains and loops collapsed to single hops.
• Canonical tags that point somewhere sensible. See what a canonical tag does if you've never had to care.
• Indexation reconciled — published pages versus indexed pages, and a reason for every gap.
• Orphan pages wired into the internal link graph, or deleted.
• Thin and duplicate pages merged, redirected or improved. Rarely all three.
• Core Web Vitals, in that order of severity: layout shift, then interaction delay, then load time.
• Schema that describes what the page actually is, not a copy-paste of someone else's markup.

Why nothing new goes live before week five

Three reasons, and the third is the one that matters to you commercially.

First, quality is assessed at site level as well as page level. Adding ten new pages to a site full of thin, duplicated, half-indexed ones doesn't average upward. It just gives the crawler more evidence.

Second, a page published before its cluster exists lands as an orphan. It has no internal links pointing at it, no supporting pages, and no reason for the crawler to come back. You get an article and no ranking, then everyone concludes content doesn't work.

Third — and this is the commercial argument — if content ships in week two alongside the technical fixes, you can never separate the two. When traffic moves you won't know which lever did it, so you'll fund the wrong one for the next twelve months.

The rule bends in exactly one case: if your site is technically clean and your actual problem is that you only have four pages, content can start in week two. A good agency will tell you that on the kickoff call rather than pad the schedule.

Why link outreach waits until week seven

Outreach spends a resource you can't top up: the contact. A journalist, an editor or a partner will take one pitch from you per quarter before they start filtering. Spending that on a page that isn't ready is expensive in a way that never shows on the invoice.

There's also a pattern problem. A domain that has published nothing for two years and suddenly acquires forty referring domains in six weeks looks exactly like a domain that bought them. Slow accumulation isn't a stylistic preference — it's what earned links look like.

Before the first pitch goes out, four things need to be true.

• There's a page worth citing — original data, a genuinely useful tool, or a position nobody else has stated plainly.
• The site loads fast enough that a journalist clicking from their phone doesn't bounce.
• There's an author or about page that survives thirty seconds of scrutiny.
• The claim is checkable. Anything unverifiable gets you one placement and no repeat.

Day 90: what's realistic, and what a guarantee can't cover

At the ninety-day read, this is a fair set of expectations for a site that started with normal amounts of technical debt.

Realistic: impressions up on your target query set, average position improving on mid-tail terms, all new pages indexed, the first handful of new referring domains, and — if your sales cycle is short — the first attributable leads. Not realistic: head terms in a competitive national market, a full revenue return, or a recovered position on a domain carrying a manual action.

This is also where the honest version of a guarantee lives. We don't promise a position for a keyword, because nobody controls Google's index and anyone claiming they do is either lying or buying. What we guarantee is movement against your own frozen baseline: beat that trailing-90-day qualified lead count within 90 days, or we keep working free until we do. It's the reason we take three clients a month — you can't carry that exposure at volume. The mechanics are on our SEO page.