SEO for Startups: A 90-Day Plan

A startup can spend a year publishing pages for searches its customers never make.

SEO for startups works when you first confirm demand, make the important pages crawlable, and build a small set of pages around problems buyers already describe. A young domain with no authority and a product with no search demand may not deserve a content program yet.

That is a hard answer, but it is cheaper than producing a library nobody asked for.

The plan below gives you a 90-day test. It starts with the site and the buyer, not a tool score. By the end, you will know whether search is showing early evidence, what to fix next, and whether another channel deserves the hours instead.

Decide whether SEO belongs in this quarter

SEO is a distribution channel with a slow feedback loop. If you need qualified conversations this month, it is usually a poor solo bet. If customers already search for the problem you solve, and your product can answer it better than the current results, it can become a compounding channel worth building beside faster routes to market.

The plan's brief identifies two startup conditions that change the decision: a new domain with no authority, and a product with no proven search demand. Those are not the same problem. A technically clean site can fix the first site's visibility issue. It cannot manufacture a query people have no reason to type.

Use this decision test before you schedule content:

Open five recent sales calls, support threads, or customer interviews. Copy the exact language people use when they explain what prompted them to look. Then inspect the results for those phrases. If the pages answer a different question, the phrase is not a fit just because a tool assigns it a volume.

Check the product's go-to-market strategy against this decision. Search works best when it has a named customer, a specific problem, and a useful next step to support. A new page cannot rescue a blurry offer.

Make the choice on a calendar, not in a founder's mood. If you have one clear buyer problem and a way to help that buyer today, give SEO a bounded test. If those pieces are missing, spend the quarter learning the buyer language before publishing at scale.

For more of the site's SEO guides, keep the split clear: technical eligibility is something you can fix today; demand is something you have to establish with buyers.

Give Google a site it can understand

Before adding pages, make sure the pages you already need can be found, crawled, and understood. A new site can be invisible for a boring reason: a blocked route, a noindex directive left from staging, a broken canonical, or links hidden behind interactions a crawler does not follow.

Google describes Search as crawling, indexing, and serving. A page can be crawled without being indexed, and indexing is not guaranteed. That distinction matters: a URL showing in a sitemap is not proof that it appears in search.

Run this first-hour check:

1. Open Google Search Console and verify the correct property. If the domain uses both www and the root version, check which version is canonical and ensure your analytics and reporting use one consistent view.
2. Inspect the homepage and your main product or service page with URL Inspection. Check whether crawling is allowed, whether indexing is allowed, and which canonical Google selected.
3. Open robots.txt in a browser. Confirm it does not block pages you want indexed. Check the page source or your CMS settings for a stray noindex left from development.
4. Open your XML sitemap. Keep it to canonical URLs that return successfully. Submit it in Search Console if it is not already submitted.
5. Click every main navigation link yourself. A buyer and a crawler should be able to reach the important pages through ordinary links, not only through a form or a script that may fail.
6. Fix the first broken or blocked page, then inspect it again. Request indexing only after the live page is accessible and indexable.

Google's URL Inspection guidance, checked November 3, 2026, says a request places an eligible page in a queue, but does not guarantee inclusion. Use the tool to diagnose a real blocker, not to tap a button repeatedly and call that a strategy.

A simple example: your home page is indexed, but the page explaining the product is not. If the product page has a noindex setting, submitting another sitemap changes nothing. Remove the setting, confirm the page responds correctly, make sure the canonical points to itself or the intended equivalent, then inspect the live URL. If it becomes eligible, you have fixed the cause. If the report still shows an issue, work from that specific report instead of publishing more pages.

Use the Google SEO Starter Guide, checked November 3, 2026, as the baseline rather than a vendor's list of mysterious ranking switches. Google's guide says changes may take from hours to several months to show, and suggests waiting a few weeks before assessing likely effects. That is a reason to define what you will inspect, not a reason to wait passively.

Pick pages by buyer intent, not by keyword score

A startup usually needs fewer pages than its content tool suggests. Begin with the pages that help a likely buyer understand the problem, compare options, and decide what to do. The right number depends on the product and the queries you can prove, so do not turn a recommendation into a quota.

Build a small query map in a sheet. Give each row one intent and one page that should answer it. If two phrases mean the same thing to the same buyer, put them on one page. If they ask different questions, separate them only when each answer has enough substance to stand alone.

Take one query and inspect the results page manually. Write down the apparent intent of the first several useful results: tutorial, category page, local result, product comparison, or something else. Then ask whether your proposed page does the same job more usefully. If the results are all tutorials and your draft is a product pitch, the mismatch is the problem. Adding the query to the title will not repair it.

For a new domain, prioritize a small group of specific, commercially relevant problems over broad head terms. A narrow search can still be a bad target if it has no audience or if your product cannot satisfy the need. Difficulty scores are estimates from a tool, not a permission slip from Google.

Use small business SEO for the broader foundations, then choose one startup-specific page to test. If your business serves a defined local area, the Google Business Profile optimization guide covers the separate local listing work. Do not build a national content plan when the first sale depends on a service radius.

Here is a practical screen for each candidate page. Keep it only if you can fill in all four answers in plain language:

1. Who searches this, and what happened just before they did?
2. What decision or task are they trying to complete?
3. What can your business show that a generic page cannot?
4. What is the next step that makes sense even if they do not buy today?

If you cannot answer the third question, gather evidence before writing. Ask customers for an example, get permission to use it, or perform the workflow yourself and record the result. Do not invent a case study to make a page feel authoritative.

Write one page that earns its space

The first useful page is not a manifesto about the product. It is a direct answer to one buyer problem, with the product appearing only where it genuinely helps.

Draft it in this order:

1. State the reader's situation in the words they use. Do not lead with your company history.
2. Give the short answer near the top. A busy reader should not have to decode your positioning to see whether the page applies.
3. Show the method. Use the actual steps, inputs, trade-offs, or decision criteria.
4. Include proof you can verify: a screenshot of your own workflow, a documented test, a policy, a named expert, or an honest example with permission.
5. Explain the limit. Who should not use this approach? What breaks when the assumptions change?
6. Offer one next action: a calculator, a useful template, a trial, a contact route, or a related guide.

A software startup targeting “invoice approval workflow” should not publish five near-identical pages for finance teams, remote teams, small teams, growing teams, and modern teams. If the workflow is the same, make one substantive page and let its examples cover the audience variations. Build another page only when the actual task and answer differ.

Google's spam policies, checked November 3, 2026, identify scaled content abuse as producing many pages primarily to manipulate rankings rather than help people. The policy applies regardless of how the pages were created. A spreadsheet of keyword variants is not editorial judgment.

This is where a founder's product knowledge can beat a large content operation. Put the awkward caveat in the copy. State what your tool cannot do. Show the decision that made you choose one approach over another. Readers can spot a sales page wearing a tutorial costume.

Keep the page title and visible heading aligned with the page's actual promise. Google's documentation on title links, checked November 3, 2026, says Google can use multiple sources to construct a result title. A carefully written title helps readers understand the page, but it is not a guarantee that Google will display it exactly as written.

Earn the first relevant links without buying trouble

A new domain has little history for other sites to reference. The answer is not to buy a pile of links or to publish filler until something sticks. Give relevant people a reason to cite a useful page, and make it easy for them to find it.

Start with the people already closest to the work. List customers, integration partners, trade groups, suppliers, community organizers, and publications that actually serve your buyer. For each, write down a specific reason they might link to a resource: a documented integration, an original benchmark, a directory entry, a useful explainer, or a quote from a person with relevant expertise.

Then do a modest outreach pass:

1. Choose one resource that is worth showing to another person. A page that only repeats product claims is not ready.
2. Identify a short list of relevant organizations or writers. Visit their sites and confirm they cover the topic now.
3. Write an individual note explaining why that resource might help their audience. Link the exact page, not the homepage.
4. Ask for a factual correction or inclusion only when the page belongs there. Do not demand a followed link or offer payment for a ranking signal.
5. Record the date, recipient, resource, and response. Send one courteous follow-up if appropriate, then move on.

Your first links may come from genuine partnerships or a useful contribution to an existing publication. That is different from a vendor promising a fixed number of links. Google's link spam policy, checked November 3, 2026, describes link practices intended to manipulate rankings as spam. Do not pay a seller to manufacture the very signal search systems are built to evaluate.

A failure mode here is confusing a mention with a lead. A relevant link can help someone discover the page, but it does not establish that the visitor is your customer. Add a tagged URL when you share the resource in your own outreach, and measure the resulting visits and meaningful actions. Do not claim that a link guarantees rankings or revenue.

If you want the deeper mechanics of evaluating incoming links and deciding what to do with them, use link building for startups. For broader distribution beyond search, how to promote a new website lays out other routes. A young company should not wait for Google to discover it before speaking to the people who can use it.

Measure a test without pretending it is a forecast

The aim of the first quarter is not to prove that SEO always works. It is to collect enough evidence to decide whether your specific product, page, and market deserve more effort.

Set up Search Console and an analytics tool before you publish or revise the test page. In Search Console, record the page's impressions, clicks, click-through rate, and average position. In analytics, record organic sessions and the action that matters to the business: a qualified signup, booked call, trial activation, or purchase. Search Console shows activity before a visitor reaches your site; analytics helps you inspect what happens afterward.

Use a simple weekly log:

Treat those rows as a review framework, not a universal ranking schedule. Google notes that changes can take weeks or months to show, and Search Console data can be sparse on a new site. A blank chart after a short interval is not proof that the market has no demand. It may mean the page is not indexed, the query volume is low, the result is not compelling, or the measurement window is too short.

Use this diagnostic sequence when the page has impressions but few clicks:

1. Compare the query text with the page's actual answer. If the page is answering a different question, revise the page or choose a better query.
2. Read the title and snippet as a stranger. Make the promise specific and accurate. Do not stuff variants into the title.
3. Check whether the result format makes a click less likely. A searcher who gets a direct answer on the results page may not need to visit.
4. Review clicks and qualified actions together. A page with fewer visits but better-fit leads may deserve more attention than a high-volume page that brings the wrong people.

When the page gets clicks but no qualified actions, inspect the path after arrival. Does the first screen confirm that the visitor is in the right place? Is the next action clear? Does the page promise something the product cannot deliver? The conversion rate optimization guide helps you diagnose that part without blaming search for a landing-page problem.

Keep the numbers honest. A position average blends queries and contexts. It is not a precise rank for every searcher. Google's Search Console traffic guidance, checked November 3, 2026, says impressions and clicks are the success measures to prioritize, while position is a signal to investigate rather than a verdict. A single position jump is not a business result.

A 90-day sequence for a startup team

Use the schedule as a constrained experiment. One founder or marketer should own the sheet. A technical colleague can handle specific access problems, but nobody needs a six-month content calendar before the first page has passed a basic check.

Days 1 to 14: prove the page can be found

1. Verify Search Console, analytics, the sitemap, robots rules, canonical URLs, and the indexability of the home page and main product page.
2. Speak with five customers or prospects. Capture the phrases they use for the problem and the alternatives they considered.
3. Choose one query group with a clear buyer task. Check the current search results and record what page format they reward.
4. Write a one-page brief: reader, situation, answer, proof, limitation, next action.
5. Set one business metric for the page. Choose a qualified action, not raw traffic.

The milestone is not “ranked.” The milestone is a technically eligible page that a buyer can use and a measurement setup that records the next step.

Days 15 to 45: publish one useful answer and show it to people

1. Publish or revise the single page. Make the answer clear near the beginning and include original product knowledge, a real procedure, or evidence.
2. Link to it from relevant pages on your own site so visitors and crawlers can find it.
3. Share it directly with a small set of customers, partners, or communities where the subject is relevant. Ask what was unclear, not whether they will promote it.
4. Contact relevant organizations with a specific reason the resource helps their readers. Track who you contacted and what happened.
5. Review impressions, queries, clicks, and qualified actions each week. Fix a visible technical or messaging defect when you find one.

Do not publish a second page just because the first has not climbed in a few days. Google says some changes take longer to surface. Use the early period to test the page's usefulness and whether your audience recognizes the problem.

Days 46 to 90: make one evidence-based decision

1. Compare the current query set with the brief. If the page appears for irrelevant searches, tighten its focus or correct the mismatch.
2. If relevant impressions exist but clicks are weak, test a clearer title and result promise. Change one major element at a time and keep a dated note.
3. If clicks arrive but the page produces no qualified actions, improve the page's fit and next step before making more content.
4. If the page is eligible but no relevant impressions appear, revisit demand. Ask more buyers, test another query source, and compare the terms with the actual category language.
5. Decide whether to continue, revise the target, or pause SEO for now. Write the reason down so the next quarter does not restart the same argument.

A startup with no search demand may get more value from interviews, outbound, referrals, or a launch community while the category forms. If people consistently describe a problem but no search audience is visible yet, keep learning and revisit later. Do not call the channel broken because the product is too new for search to be the place customers begin.

If the category already has demand and your page earns relevant impressions, give the work another bounded period. If the page remains unseen, verify technical access and whether the query has real intent before increasing publishing volume. The trade-off is straightforward: keep investing only where evidence is accumulating.

The Board is one paid placement on a public leaderboard of brands, where rank follows money paid. A startup can point its listing at a page it wants people to visit, but that is another distribution test, not a substitute for SEO or a promise of search rankings. Use it only if the audience and the destination make sense for your business.

The test to keep

A search plan should leave you with a page customers can use even if rankings take time. That is the useful floor: clearer language, a better answer, a working path to the product, and evidence about whether the problem has search demand.

Open Search Console today. Inspect the home page and your core offer page. Then ask one customer how they described the problem before they found you. Put that phrase beside the actual results and decide whether one page can answer it better.

If it can, make that page the experiment. If it cannot, spend the next hour learning what the customer needs before you ask a search engine to send them.