How UpSearch Scoring Works

Why transparency matters

Every score in UpSearch is derived from real data. There are no black-box numbers, no proprietary "SEO scores," and no vanity metrics. When UpSearch assigns a score or confidence level, you can trace it back to the specific data points that produced it.

This page explains the logic behind UpSearch's scoring so you can evaluate whether a score is meaningful for your situation.

The evidence hierarchy

All UpSearch scoring follows a strict evidence hierarchy:

1. Google Search Console data is treated as ground truth. Impressions, clicks, position, CTR, and indexing status come directly from Google. These are measurements, not estimates.

2. Google Analytics data provides behavioral context. Sessions, engagement, and conversion data add depth but are not search-specific. GA data is used to enrich findings, not to drive them.

3. Live crawl data shows what UpSearch can observe when it visits your site. Page titles, meta descriptions, heading structure, schema markup, response times, internal links, and content structure. This is direct observation, not inference.

4. SERP data shows what actually ranks for your target queries. Competitor identification, result types, and SERP features come from real search results.

When data sources conflict, UpSearch trusts them in this order. When a data source is missing, UpSearch says so explicitly rather than filling in gaps with assumptions.

Confidence levels

UpSearch assigns confidence levels to findings based on the strength of the underlying evidence:

High confidence. Multiple data sources agree. The finding is supported by clear, quantitative evidence. Example: a page losing impressions and clicks across multiple queries over a sustained period, confirmed by GSC data.

Medium confidence. The finding is supported by data but with some ambiguity. Example: a potential cannibalization issue where two pages share some queries but the overlap is not severe.

Low confidence. The finding is based on limited data or indirect signals. Example: a content quality concern based on crawl data alone, without GSC performance data to confirm the impact.

UpSearch does not present low-confidence findings as facts. It labels them clearly so you can decide how much weight to give them.

What UpSearch does not score

UpSearch deliberately avoids certain types of scores:

No overall SEO score. A single number cannot meaningfully represent the complex state of a site's search performance. UpSearch provides specific findings in specific areas instead.

No keyword difficulty scores. These are unreliable estimates that vary wildly between tools. UpSearch shows you what actually ranks for a query (SERP data) and lets you assess competitiveness from reality.

No domain authority equivalent. Domain-level authority scores are proprietary metrics that Google does not use. UpSearch evaluates authority through actual ranking performance and backlink data for specific pages and queries.

No health percentage. A site "health score" of 85% is meaningless without context. UpSearch identifies specific issues with specific severity levels instead.

How specific scores are calculated

Traffic decline severity. Calculated from the delta between two time windows in GSC. The magnitude of impression and click loss, the number of affected pages, and the number of affected queries determine severity. A decline affecting your top 5 pages by traffic is more severe than a decline affecting 5 low-traffic pages.

Content quality signals. Derived from crawl data: presence and quality of title tags, meta descriptions, heading structure, content length relative to topic complexity, schema markup, and internal link density. These are proxy signals, not direct quality measurements, and are labeled accordingly.

Technical health indicators. Based on crawl results: server response times, HTTP status codes, redirect chains, canonical configuration, robots.txt rules, and Core Web Vitals data from GSC. Each issue is categorized by severity based on its potential impact on crawling, indexing, or ranking.

SERP confidence. When UpSearch analyzes competitors from SERP data, it assigns a confidence level based on the number of unique domains found, the number of results returned, and the consistency of the data. Thin SERP data (few results, few unique domains) produces low confidence, and UpSearch says so.

When scores are missing

If UpSearch cannot calculate a score because required data is missing, it does not guess. It returns a clear statement that the data is unavailable. This is a feature, not a limitation. A missing score is more honest than a fabricated one.

Common reasons for missing scores:

• Google Search Console is not connected
• The site has not been crawled yet
• SERP data returned insufficient results
• The time period is too short for meaningful comparison

Practical takeaway

Treat UpSearch scores as what they are: evidence-based assessments with explicit confidence levels and transparent methodology. Use them to inform your decisions, not to replace your judgment. When a score seems surprising, check the underlying data. The data is always available.