A free site evaluation should produce evidence, not anxiety
Most free website evaluations are designed to begin a sales conversation. That is not automatically a problem. The problem begins when the report uses an unexplained score, declares the whole website broken and offers a rebuild before anyone has tested what the site is meant to achieve.
A useful free evaluation should do three things:
- identify repeatable technical or usability signals;
- show enough evidence for you to reproduce the important findings; and
- distinguish urgent failures from improvements that can wait.
It should also be honest about its limits. An automated grader can inspect code and measurable page behaviour quickly. It cannot understand your sales process, judge whether a claim is persuasive or know why a customer chose not to enquire.
This guide explains what a credible free site evaluation should include, what it cannot prove and how to decide whether the resulting recommendations deserve your time or budget.
If you already have a numeric result and want help interpreting it, use our separate guide to website grader scores and what to fix first. This article deals with the evaluation itself: what should be tested before the score is trusted.
The three common types of website evaluation
The phrase "free website audit" is used for very different services. Establish which one you are receiving before comparing results.
Automated website grader
You enter a URL and software checks the page against a set of rules. The result may cover performance, mobile rendering, metadata, security basics and accessibility signals.
The advantages are speed and consistency. The same checks can be repeated after changes, and a technical issue can be found without waiting for a consultant.
The limitation is context. The grader sees a page and its resources; it does not see your commercial priorities, internal processes or the conversations your team has with customers.
Short expert review
A designer, developer or marketer reviews several representative pages and sends observations or records a video. This can identify unclear messaging, weak calls to action and awkward journeys that automated rules miss.
Quality varies substantially. A strong reviewer explains the route tested and the consequence of each issue. A weak one lists design preferences as universal rules.
Full audit or discovery
A proper audit combines technical testing, analytics, search data, real user behaviour, content, accessibility and business context. It may include stakeholder interviews, conversion journeys and implementation planning.
That amount of work is rarely free. A no-cost evaluation should not pretend to be a full audit; it should tell you whether deeper work is justified and where it should begin.
What an automated grader should test
An automated evaluation is most valuable when it focuses on questions that code can answer reliably.
Page availability and security basics
The tool should confirm that the URL resolves successfully, uses HTTPS and does not create obvious redirect or server-error loops.
For important pages, availability to a normal browser is not the same as availability to search engines. Google recommends using Search Console's URL Inspection workflow to see what Google knows about a specific page and test a live version when diagnosing indexing.
A free grader may detect an index directive or canonical tag, but it cannot access your private Search Console coverage history unless you explicitly connect that account.
Mobile rendering
The evaluation should use a genuine mobile viewport rather than simply shrinking a desktop screenshot. At minimum it should detect:
- horizontal overflow or content wider than the screen;
- text that is clipped, hidden or too small to read comfortably;
- controls that cannot be tapped reliably;
- forms that do not fit or expose usable error messages; and
- layouts that depend on hover.
The homepage alone is not enough. A site can have a polished hero and a broken enquiry form, article template or booking journey.
Performance
Google's open-source Lighthouse tool audits performance, accessibility, SEO and other quality signals. A responsible evaluation should name its test method, device profile and date because results change with network conditions, caching and third-party services.
Useful performance evidence includes the individual metrics and the element responsible for the delay, not only a score. A low Largest Contentful Paint result is actionable when the report identifies whether the delay comes from an image, server response, font or deliberately hidden text.
Run a technical test more than once before commissioning work. One slow external response should not become a proposal to rebuild an otherwise healthy site.
Search foundations
An automated evaluation can check for:
- a descriptive page title and meta description;
- one clear primary heading;
- canonical and index directives;
- sitemap and robots availability;
- crawlable internal links;
- structured-data syntax; and
- broken or redirected destinations.
These are foundations, not a ranking forecast. A technically valid page may still fail because it does not answer a real search intent, has no authority or competes with another page on the same site.
Accessibility signals
Automated checks should cover detectable issues such as missing alternative text, form labels, document language, colour contrast and some keyboard problems.
The WCAG 2.2 Quick Reference groups accessibility requirements under perceivable, operable, understandable and robust principles. Automation can support that work, but it cannot replace keyboard testing, screen-reader judgement or a review of whether instructions and errors make sense.
A credible report says "automated accessibility checks passed" rather than "the site is accessible".
What automation cannot tell you
The most commercially important questions often need people, data or both.
Whether the offer is clear
A grader can detect a heading. It cannot decide whether the heading tells the right visitor what you do, why it matters and who it is for.
If people arrive but do not enquire, the issue may be positioning, proof, risk or the next step rather than page speed. Work through the evidence in why a website is not generating leads before treating every weak conversion rate as a technical fault.
Whether the journey matches customer intent
A button labelled "Contact" can pass every automated test while sending a ready-to-buy visitor to a long generic form. A manual review should test a named task: request a quote, book a table, check availability, compare services or find a specific answer.
The evaluator should record where the journey becomes ambiguous, not merely count the clicks.
Whether traffic is genuine or valuable
Analytics totals can be dominated by bots, internal testing, legacy URLs or misclassified campaigns. A large rise in Direct traffic is not automatically growth.
A deeper evaluation should separate acquisition channels, landing pages, engagement and real conversion events. Session recordings can reveal friction, but only after internal verification and automation sessions are excluded.
Whether a redesign is necessary
Many issues belong to one component, template or integration. A global navigation problem can affect every page and still need one focused repair.
A free evaluation should not jump from "we found twelve warnings" to "replace the website". It should first identify whether the root cause can be fixed in the current system.
What a useful short human review adds
Even a limited expert review should add context that automation cannot.
A representative route
The reviewer should state the pages and task examined. For example:
- Arrive from a service-related search.
- Understand the offer and relevant proof.
- Compare the available route or engagement.
- Reach and complete the enquiry form.
- Confirm what happens after submission.
That route is more informative than scrolling the homepage and commenting on taste.
Mobile-first reading and interaction
The reviewer should use a real phone-sized layout from top to bottom. They should look for random type scales, awkward line breaks, repeated decoration, controls hidden below oversized sections and content that assumes desktop space.
This is also where real content matters. Placeholder copy rarely exposes the line wrapping and hierarchy problems that users encounter.
Trust and decision support
A human can ask:
- Is it clear who operates the business?
- Is the evidence specific and relevant to this service?
- Are claims supported rather than generic?
- Can the visitor understand the process before sharing details?
- Does the call to action match their readiness?
These questions cannot be reduced to the presence of a testimonial component.
A realistic priority order
The final value of a human review is judgement. The evaluator should explain why one issue matters more than another and what evidence would prove it fixed.
The evidence you should receive
A finding is useful when another person can reproduce it.
For every important issue, ask for:
- the exact URL or shared template;
- device and viewport;
- reproduction steps;
- screenshot, recording or measured audit evidence;
- the affected user or search journey;
- the likely root cause, clearly labelled if still uncertain;
- the smallest sensible correction; and
- the check that will confirm completion.
"The website feels dated" is an opinion. "The service heading extends 48 pixels outside a 390-pixel viewport and hides the final word" is a finding.
Red flags in a free site evaluation
Be cautious when the evaluation:
- produces a score but hides the underlying checks;
- claims one automated run proves lost revenue or rankings;
- uses urgent language without showing a broken journey;
- recommends a complete rebuild before asking what the site does;
- treats every warning as equally important;
- guarantees rankings or conversion improvements;
- reports accessibility as a single automated pass;
- asks for sensitive account access before explaining why it is needed; or
- compares your result with unnamed "industry standards" that cannot be inspected.
A free evaluation can lead to paid work and still be honest. Transparency is the dividing line.
How to prepare before running an evaluation
Give the reviewer or tool the right target.
Choose the important URL
Do not automatically test only the homepage. Use the page that receives valuable traffic or starts the main conversion journey: a service, booking, category, product or landing page.
State the primary task
Write one sentence describing what the intended visitor should achieve. This lets a human reviewer distinguish visual preference from functional friction.
Record the current baseline
Preserve the date, analytics period, relevant search queries and conversion event definition. Without a baseline, later improvement claims become guesswork.
Avoid changing five things at once
If you plan to act on the findings, group them by root cause and test after each material change. One template correction may resolve dozens of reported page issues.
How to prioritise the findings
Use this order:
- Access and operational failure: server errors, broken forms, accidental noindex, unsafe behaviour or essential controls missing on mobile.
- Primary conversion journey: unclear routes, validation failures, missing confirmation, weak trust or severe delay.
- High-traffic template quality: performance, accessibility, layout and search foundations on pages people actually reach.
- Content and intent: pages that rank or receive impressions without answering the searcher's question convincingly.
- Maintenance improvements: lower-impact warnings, consistency and technical debt.
This prevents a team polishing minor audit warnings while an enquiry form silently fails.
What a good free evaluation should not promise
A no-cost report cannot establish a reliable traffic forecast, conversion uplift or return on investment without your data. It also cannot prove why an individual visitor left.
The right conclusion may be that the site is technically healthy and needs better distribution, a clearer offer or more qualified traffic. An evaluator willing to say "do not rebuild this" is more useful than one with a fixed service to sell.
The bottom line
A good free site evaluation is a disciplined first pass. It tests repeatable technical signals, reviews one meaningful journey, shows reproducible evidence and gives you a sensible order of work. It does not disguise a sales script as a diagnosis.
You can run the LogicLeap free website grader for an immediate technical baseline, then use the score interpretation guide to prioritise the output. For a broader review of conversion journeys, search visibility and implementation risk, start with our digital audit or talk to LogicLeap. We will tell you what is genuinely broken, what can wait and when the current site is already good enough.