SEO for startups: a 90-day content sprint plan

Startups systematically underinvest in SEO, and the reason is respectable: with eighteen months of runway, a channel that pays out in months four through twelve loses every planning argument to ads that pay out on Thursday. But the arithmetic that defers SEO forever is the same arithmetic that has the company, at month eighteen, buying every single customer at full price while a competitor's year-old content compounds for free. The way out of the deadlock isn't faith. It's reframing what the first quarter of SEO is for.

A 90-day content sprint does not buy traffic. It builds three things: the operating machine (instrumentation, process, cadence), ownership of the small set of queries closest to your revenue, and, by day 90, leading indicators honest enough to decide whether this channel deserves quarter two. That's a deliverable a startup can actually evaluate, on a startup's clock. Here's the plan.

Before day 1: three prerequisites

The sprint assumes three things exist. First, a validated offer: you know who buys, roughly why, and what words they use, because content built on a guessed positioning gets rewritten along with the positioning. (Pre-product-market-fit teams should spend these 90 days talking to users, not ranking for them.) Second, an owner: one named person with five to eight hours a week, protected. Content sprints without an owner become week-two casualties of the roadmap. Third, a defined conversion: the action that counts, signup, demo request, trial, so results have a unit before the first post exists.

Days 1-15: foundation

Two weeks of unglamorous setup that everything else stands on.

Instrument. Search Console verified, GA4 running, the conversion wired as a key event, Google's term for the action you've marked as mattering, and the two tools linked. The conversion-tracking setup is an afternoon, and doing it now means even the sprint's first trickle of visitors produces learning.

Make the machine legible. The technical basics of a new site: crawlable pages, a sitemap, sane titles, fast enough to not embarrass yourself, and a homepage that states in plain sentences what the product is, for whom, and what category it lives in, because that page is what machines, search engines and AI assistants alike, read to learn what you are.

Map the money queries. Now the strategic heart of the sprint. List every query a person types when they're close to buying what you sell: your category plus "tools"/"software"/"services", "[competitor] alternatives", "[competitor] vs [competitor]", "how to solve [the exact problem]", "[category] pricing", "best [category] for [your niche]". Check real volumes, but don't be scared off by small numbers, a query with 90 searches a month and pure buying intent outranks a 5,000-search trivia term for a startup's purposes, buyer intent is the entire filter. Then search each one and study who currently ranks: thin listicles and dated posts are open doors; entrenched deep guides go to the back of the queue. Output of week two: one bounded cluster of 15 to 20 pages, ordered by intent, that the next ten weeks will build. One cluster, deliberately: coverage depth on a single topic is what makes pages lift each other, and scattering the same effort across three themes forfeits that.

Days 16-75: the production sprint

Eight weeks, two substantial pages a week, in a strict order most startups get backwards.

Bottom of the funnel first. The instinct is to start with big thought-leadership topics. Resist it. The sprint publishes in reverse-funnel order: comparison pages and competitor-alternatives pages first, then use-case and how-to-solve-it pages, then pricing and evaluation content, and only then the broader educational layer. Two reasons. The intent closest to money converts from the smallest traffic, first: a hundred visitors comparing alternatives are worth more than ten thousand reading a trend piece. And these pages are individually winnable early, since they target specific, moderately contested queries rather than the category head terms. The click market you're entering is brutally top-weighted, position one takes 27.6% of clicks and the top three over half, and you get to the top of *narrow* results pages years before broad ones.

The weekly loop. Each page: a real brief carrying your product knowledge, customer language, and honest opinions, machine-drafted, then human-edited and fact-checked. AI assistance is what makes the cadence feasible for a team of one, the controlled research shows faster *and* better output, and the ranking data shows top pages are exactly such blends, but the human pass is non-negotiable: comparison pages containing fabricated claims about competitors are a legal and credibility problem, not just a quality one. On comparison pages specifically: be scrupulously fair. Honest "when competitor X is the better choice" sections convert better, age better, and are the difference between a resource and an ad.

Link as you go. Every new page links to its siblings and, from day one, to the signup-adjacent pages. Mid-sprint, around week seven, write the cluster's hub, the overview guide that links to everything, easiest to write once its spokes exist, per the cluster architecture.

One velocity warning: two good pages a week is the sprint. Twenty generated pages a week is a different, documented experiment whose subjects got removed from the index, and Google's scaled-content policy is explicit that automation-at-volume without value is spam regardless of tooling. Startups are constitutionally tempted by the growth-hack version. The graveyard is well marked.

What a sprint backlog actually looks like

For shape, here's a plausible 18-page backlog for a startup selling scheduling software to trades businesses, in publish order.

Weeks three to six, the closest-to-money layer: "[Big incumbent] alternatives for small trades businesses", "[Incumbent A] vs [Incumbent B] for tradies: an honest comparison", "Best scheduling software for electricians", "Best scheduling software for plumbing companies", "Job scheduling app pricing: what you actually pay", and "Scheduling software for a two-person crew: what you need and what's bloat". Six pages, every one targeting a query someone types with a card within reach.

Weeks six to nine, the problem layer: "How to stop double-booking jobs", "How to schedule recurring maintenance jobs without a whiteboard", "Dispatching jobs to a small crew: three systems compared", "How much time does job admin actually take? (what we measured)", and "Paper diary to digital scheduling: a switch plan that doesn't blow up your week". Five pages meeting buyers one step earlier, each ending naturally at the product.

Weeks nine to eleven, the evaluation layer: "What to look for in trade scheduling software: 9 questions", "Scheduling software implementation: what the first two weeks really involve", and "Do you need scheduling software or just a shared calendar?", that last one deliberately honest about who shouldn't buy, which is exactly the page that earns trust and links.

Week seven, mid-sprint as planned: the hub, "The complete guide to job scheduling for trades businesses", linking everything.

Three spare slots stay unassigned on purpose, filled during the sprint by whatever the early Search Console data and sales conversations reveal, the first taste of the steer-by-data habit the sprint is really building. Swap the nouns for your market and this is the template: specific, buyer-first, winnable, and bounded.

Days 76-90: distribution and the decision gate

The final fortnight ships less and connects more.

Close the mesh. An hour auditing internal links across all ~18 new pages, both directions. Confirm everything's indexed; chase down anything that isn't.

Earn the first outside signals. Not a link-building campaign, just the startup-sized version: your product in the relevant directories, a founder answer where your buyers ask questions, one or two genuine mentions from adjacent newsletters or communities. A young domain needs *some* external evidence it exists.

Then the day-90 review, the sprint's actual deliverable. Pull Search Console and answer four questions. Are all pages indexed? Are impressions climbing week over week, and spreading across more distinct queries? Are the money pages entering the rankings anywhere, positions 15 to 40 counts as "yes", that's the on-ramp? Did any conversion, even one, touch a content page on its path? At day 90 those are the honest instruments: meaningful clicks generally arrive in months four through eight, so traffic is not the test yet, and anyone who promised otherwise was selling something. Ahrefs' sobering stat that only 1.74% of new pages crack the top 10 within a year describes the internet's ocean of abandoned content, and the entire design of this sprint, buyer-intent targets, winnable queries, real substance, consistent cadence, exists to put you in the other distribution.

The gate itself: climbing impressions, spreading queries, and drifting positions mean the machine works, continue into quarter two at the same cadence, extending the cluster where the query data points. Flat everything despite indexed pages means the thesis needs revisiting, usually the topics were wrong or the market searches differently than assumed, and you adjust with 90 days of cheap learning instead of a year of expensive faith. Either way, you decided on instruments, which is the whole startup way of doing things applied to a channel that usually escapes it.

After the sprint

Quarter two runs the same loop at sustainable cadence, one to two pages weekly, extending toward the second cluster once the first genuinely covers its ground. Somewhere in months four through eight, the compounding becomes visible in clicks and first content-sourced signups, and the channel starts arguing for itself in the payback terms a board understands. The sprint's real product turns out to be the operating habit: by the time SEO produces revenue, you're a team that ships, measures, and steers content weekly, which is the capability competitors can't copy by increasing a budget.

Quick answers for founders

Should the founder write this or should we hire? For the sprint, keep it close: the comparison and use-case pages depend on product knowledge and honest opinions that only insiders hold, and the editor-in-the-loop mechanics make insider-led production cheap. Hire or outsource at quarter two or three, once the working system exists for someone to slot into.

Doesn't a brand-new domain make this pointless? It makes it slower at the head terms and barely matters at the specific ones, which is precisely why the sprint targets specific ones. New domains win narrow contests constantly; what they don't do is leapfrog established sites on broad queries, and the plan never asks them to.

How does this interact with our paid spend? Complementarily and immediately: the comparison and use-case pages built for search double as the best landing pages your ads have had, and paid clicks onto them generate conversion data months before organic traffic arrives, an early read on whether the pages themselves work.

What if we pivot in six months? Then you'd lose some pages, which is why the sprint deliberately builds bottom-funnel content tied to the problem you solve rather than a thought-leadership empire tied to a positioning. Problem-shaped content survives most pivots; brand-essay content survives none. Scope the investment to your confidence, and remember the other deliverable, the operating habit and instrumentation, transfers to whatever you build next.

The one-line version

A startup's first 90 days of SEO shouldn't chase traffic it can't have yet: spend two weeks instrumenting and mapping the queries closest to revenue, eight weeks publishing a single bottom-funnel cluster at two edited pages a week, and two weeks connecting and reviewing, then decide quarter two on the leading indicators. You'll have claimed the ground nearest the money, built the machine, and turned "should we do SEO?" from a faith debate into a data review.