Business & SEO / essay

What Can a New Website Prove About SEO in Ninety Days?

Plan a transparent search experiment that separates technical changes, observations and unsupported ranking claims.

In this note

Launching a personal website creates a useful opportunity to study search visibility in public. It does not create an instant case study of ranking success. A new domain may attract little search traffic at first, and even a technically correct page may not appear for a competitive query. The valuable experiment is to document what was published, what was measured, and which conclusions the evidence can support.

This guide proposes a careful ninety-day search experiment for a professional portfolio and technical blog. It avoids the common mistake of equating publication with indexing, indexing with ranking, or ranking with client-ready commercial results.

A technical baseline you can inspect today

For this website, a useful first check is whether the homepage exposes a title, a description, one H1 and the intended canonical URL. The following Python 3 script reads only the public homepage. It uses the standard library and does not log visitors or access Search Console.

from html.parser import HTMLParser
from urllib.request import urlopen

class PageChecks(HTMLParser):
    def __init__(self):
        super().__init__()
        self.h1_count = 0
        self.title_count = 0
        self.canonical = None
        self.description = None

    def handle_starttag(self, tag, attrs):
        attrs = dict(attrs)
        self.h1_count += tag == "h1"
        self.title_count += tag == "title"
        if tag == "link" and attrs.get("rel") == "canonical":
            self.canonical = attrs.get("href")
        if tag == "meta" and attrs.get("name") == "description":
            self.description = attrs.get("content")

with urlopen("https://itsekramah.github.io/", timeout=20) as response:
    html = response.read().decode("utf-8")
checks = PageChecks()
checks.feed(html)
assert checks.h1_count == 1
assert checks.title_count == 1
assert checks.canonical == "https://itsekramah.github.io/"
assert checks.description

Passing those assertions demonstrates those specific HTML properties at the time of the request. It says nothing about impressions, clicks, ranking or conversions. The Site Audit Kit demonstration similarly distinguishes inspectable HTML findings from search-performance claims. No Search Console export or commercial SEO outcome is presented here.

Update, 16 November 2026: added and executed the public-page baseline example, with no change to the original publication date.

Define the experiment before the outcome

Start with a clear question: can a new technical website gain discoverability for specific, useful queries without buying ads or backlinks? That is different from asking whether every post will reach Google’s first page. The former can be evaluated over time using observable search metrics; the latter is an unrealistic guarantee.

Record the launch date, domain, hosting setup, publication cadence, and baseline search visibility. Keep a simple change log for major edits. If the domain changes during the study, document the migration because the change can affect the interpretation of impressions and indexing behavior.

Choose topics based on real expertise

A technically skilled writer has an advantage when they can explain problems that they have actually investigated. Instead of targeting an enormous keyword such as “data science,” consider a specific tutorial about validating a data-cleaning step or diagnosing an indexing issue. A narrow search query is not automatically easy to rank for, but it often has clearer intent and gives the writer a more defined task.

Use search results to inspect whether existing pages answer the question well. Look for missing explanations, outdated steps, untested code, or failure to discuss limitations. The goal is to improve usefulness, not merely to repeat a competitor’s headings with different wording.

Make every article worth reading without search traffic

Search engines are not the only audience. Articles should be useful to a prospective employer, collaborator, or client who reaches the site through GitHub or LinkedIn. Original diagrams, validated code, careful examples, and transparent methodology provide lasting value even when a page attracts few clicks.

Avoid publishing large numbers of generic articles solely to populate categories. Five well-researched posts can be a stronger beginning than fifty interchangeable summaries. The website’s editorial standards should remain stable regardless of ranking fluctuations.

Establish the technical foundation

Each article should have a unique, descriptive title and URL, an indexable canonical page, sensible headings, and internal links where they genuinely help readers. Make sure pages are not accidentally excluded by robots directives or noindex metadata. Generate a sitemap and submit it through Google Search Console once the site is live.

These measures support discovery and interpretation, but none guarantees indexing. Also check that pages load correctly, are usable on mobile, and do not rely on client-side JavaScript merely to reveal their core content. A static site can make this easier, provided it is configured properly.

Measure the right things

Search Console can show impressions, clicks, queries, pages, and average position in Google Search, subject to its reporting limitations. An impression means a result was shown under applicable reporting rules, not that a human read the article. A click is stronger evidence of search engagement but does not establish satisfaction or conversion.

Average position can vary by query, location, device, and period. Report the exact time range and the set of queries being compared. Avoid claiming that a page “ranks #1 on Google” based on a single personalized search or a screenshot without context.

Keep a transparent tracking sheet

For each post, record category, target question, publication date, updates, indexing status, observed queries, impressions, clicks, and any relevant technical changes. Do not fill empty cells with fabricated progress. If a post receives no impressions after several weeks, that is an observation worth recording, not a reason to rewrite the result into a success story.

A useful experiment log also notes confounding factors: outside links, a domain migration, major site redesign, social sharing, or changes in the search results. Since many variables shift together, a before-and-after observation rarely proves that one edit caused an improvement.

Distinguish organic performance from paid promotion

A page can attract visitors without advertisements through search, social sharing, direct links, newsletters, or referrals. These are different acquisition channels. If the claim is about organic Google Search, use Search Console search-performance data rather than total site visits from an analytics dashboard.

Likewise, “without SEO” is often a misleading phrase. Useful titles, sensible internal linking, structured pages, and clear content already overlap with search optimization. A more defensible statement is that a site grew without paid advertisements or purchased link placements, if that is what the evidence shows.

Understand what ninety days can and cannot show

In the first month, focus on a functioning site, a small set of high-quality articles, indexing checks, and a consistent publishing process. In the second month, review queries and technical issues, then update content only where the changes genuinely improve the answer. By the third month, inspect trends and document what happened.

Ninety days may be enough to observe initial discovery or early query-level impressions, but it may not be enough to draw strong conclusions about long-term ranking potential. Competitive topics, low search volume, and a new site’s limited recognition can all influence outcomes. An honest report may conclude that the experiment produced insufficient evidence for certain claims.

Turn results into a credible portfolio case study

A professional case study should explain the initial problem, technical setup, content strategy, measurement method, actual observations, and limitations. Use screenshots or exported reports with dates, and make clear which figures refer to the site as a whole versus individual articles. If you report an improvement, include the baseline and comparison period.

Do not invent client testimonials or imply commercial success from academic experimentation. A self-directed SEO experiment can demonstrate technical understanding and disciplined measurement. It is not equivalent to proving that an intervention will work for every business.

Respect readers and privacy

Traffic measurement does not justify collecting unnecessary personal data. Prefer privacy-conscious analytics and avoid adding trackers just because other websites use them. State whether third-party services are involved. For a small professional site, Search Console may already answer many of the questions about Google discovery without needing intrusive analytics.

Be careful when publishing example client situations. Remove identifiable details and secure permission before using proprietary screenshots or metrics. A case study can communicate a problem and approach without exposing confidential information.

The most useful result is an honest one

A successful search experiment is not necessarily a first-page ranking. It is a repeatable process that produces evidence: which pages were discoverable, what queries surfaced, what technical issues occurred, how content changed, and what remains uncertain.

That process is relevant to freelancing because clients need someone who can distinguish actions from outcomes and outcomes from causation. Publish carefully, measure consistently, and let the results—not an impressive-sounding promise—shape the story.

Sources & further inspection

Related work

  • Technical SEO
  • Measurement
  • Content strategy

More from the library

← Back to writing