How to measure if your website is actually working

Monthly unique visitors. The headline number. Without traffic, nothing else matters.

Source breakdown. Organic search, direct, paid, referral, social. Each source behaves differently and tells you what marketing is working.

Top landing pages. Which pages bring people in first. These are your real front doors and deserve the most attention.

What to ignore: total pageviews. A high number can mean engaged visitors or just lost ones clicking around. Focus on sessions and conversions instead.

Goal completions. Forms submitted, calls made, products purchased, bookings booked. Define a goal per page that has one.

Conversion rate by source. Organic traffic converting at 4 percent and paid at 1 percent tells you exactly where your marketing budget should go.

Drop-off points. Where in a multi-step process people leave. Often a checkout, a long form, or a page that does not answer the obvious question.

Impressions and clicks in Search Console. Impressions tell you whether you appear, clicks whether the snippet earns attention.

Average position for your key terms. Position 3 and position 9 are different worlds: the click-through rate drops sharply after the top 3.

Pages indexed vs pages published. A gap here means Google is rejecting some of your content. Worth investigating.

Core Web Vitals: LCP, INP, CLS. Search Console reports them in the Page Experience section. Anything outside the good range is a real ranking risk.

Uptime. A site that goes down even a few hours a month loses sales and trust. Free monitors like UptimeRobot give you alerts.

Mobile usability. Search Console flags issues here separately. Fix any flagged page promptly because mobile traffic dominates.

Revenue or pipeline attributable to the site. For e-commerce this is direct. For services, tag leads in your CRM with their source.

Cost per lead by channel. SEO has zero variable cost but takes months to build. Paid ads are instant but cost per lead matters every day.

Lifetime value of a website-acquired customer. Often higher than other channels because the visitor came actively looking, not interrupted by an ad.

Pull GA4 (or Plausible) for last month vs the month before. Note traffic and conversion changes.

Open Search Console. Check impressions, top queries, and any new errors.

Run a PageSpeed Insights test on the homepage and one key inner page.

Write three sentences: what improved, what got worse, what you will change next month. That is the whole review.

A site in Estonian, Russian and English is three audiences behind one domain, and a blended average hides what each is doing. If the languages live in their own folders, the split is easy.

• In Search Console, filter pages by the language folder to see impressions and clicks per language, and use the country filter to separate Estonia from the rest of the world.
• In your analytics tool, compare inquiries or orders per language version, not only visits.
• Check indexing per language. It is common for one language to be fully indexed while another sits in the crawled but not indexed list, which usually points to thin or machine-translated pages.

For an e-resident company the country view is the important one. If the customers are in Germany and nearly all impressions come from Estonia, the site is being found by the wrong market and the content, not the design, needs to change.

With a proper consent banner, visitors who decline are missing from cookie-based analytics. The tool then shows fewer visits and fewer conversions than really happened. That is the banner working as intended, not a fault.

So anchor your review in numbers that do not depend on consent: orders in the shop admin, payments reported by Montonio, Maksekeskus or Stripe, form emails received, calls logged. Use analytics for proportions and trends, such as which source or page performs better, and your own records for totals.

For a service business in Estonia a share of inquiries comes by phone. If calls are not written down with a note on how the caller found you, the website will look weaker than it is.

Estonia is a small market, and a local niche site may have a few hundred visits a month. At that size, monthly percentages mislead.

Take a hypothetical window installer in Tartu with 400 visits and 6 inquiries in March, then 380 visits and 9 inquiries in April. The conversion rate appears to jump by half, yet the difference is three people. No conclusion can be drawn from it.

With low volumes, compare quarters instead of months, count inquiries as absolute numbers, and read every inquiry: which service, which language, which page it came from. Ten inquiries read closely tell a small business more than a dashboard of percentages.