Client Website Feedback Analysis

Clients share a lot of feedback during website projects. But what do they actually ask for most? To find out, we analyzed 1,000,000+ client feedback items submitted through Feedbucket.

We anonymized every entry and used a local LLM to classify each item into one of eight categories, and for the three largest, into sub-categories as well. In total, 98.7% of all feedback fell cleanly into these eight buckets.

Here’s what we found:

Feedback Distribution in 8 Buckets

Content & Copy Feedback

Content specific feedback was the biggest bucket of them all and stands for 36.8% of all the feedback. We could sort the content feedback into the below 5 sub-buckets.

• Text Revisions & Rephrasing: Requests to rewrite, rephrase, or update existing text for clarity, tone, branding, or marketing impact (e.g., rewording taglines or descriptions).
• Content Additions: Requests to add new text, links, documents, videos, or other content elements, such as service descriptions, dates, or menu items.
• Content Removal: Requests to delete text, links, or elements, such as outdated terms, redundant phrases, or SEO titles.
• Text Corrections (Grammar, Spelling, Formatting): Fixes for typos, grammar, punctuation, or formatting issues, such as adding commas or standardizing quotation marks.
• Content Accuracy & Updates (Data-Specific): Updates to specific data like dates, prices, or factual details to ensure accuracy (e.g., correcting webinar dates or monetary values).

Visual Design Feedback

Visual design was the second largest bucket and stands for 31.5% of all feedback. We could sort the visual design feedback into the below 5 sub-buckets.

• Image & Media Updates: Requests to replace, add, resize, crop, optimize, or update images, photos, logos, videos, icons, or other media elements. This includes sourcing client-specific assets or fixing mismatches (e.g., wrong stock photo or blurry logo).
• Layout & Positioning Fixes: Adjustments to element placement, alignment, reordering, overlaps, or device responsiveness (e.g., mobile distortions, centering content, or rearranging sections for better flow).
• Color & Visual Effects: Changes to colors, transparency, hover states, gradients, backgrounds, or subtle effects to enhance readability, contrast, or branding alignment (e.g., fixing low-contrast text or updating hover colors).
• Typography & Text Styling: Refinements to font sizes, weights, alignments, formatting, or hierarchy for improved clarity and consistency (e.g., making text bolder or ensuring uniform heading sizes).
• Spacing & Padding Adjustments: Modifications to margins, gaps, padding, or whitespace between elements to achieve visual balance (e.g., reducing "dead space" or adding padding in containers).

Functionality/Bugs Feedback

The third largest bucket but a lot smaller than visual design and content. Stands for 11.6% of the feedback. We could sort the functionality and bugs feedback into the below 5 sub-buckets.

• Display & Rendering Issues: Problems with element rendering, such as incorrect image sizes, missing content, duplicate elements, or visibility issues (e.g., content cutoff or overflowing).
• Data & Logic Errors: Issues with incorrect data display or faulty logic, such as wrong sorting, numerical discrepancies, or conditional logic failures (e.g., required fields bypassed).
• Broken or Incorrect Links: Issues with hyperlinks that are broken (e.g., 404 errors), redirect incorrectly, or are missing (e.g., unlinked text that should be clickable).
• Interactive Element Malfunctions: Issues with buttons, menus, forms, or other interactive components that don’t work as intended (e.g., unclickable buttons, non-functional menus, or incorrect hover states).
• Mobile-Specific Bugs: Issues specific to mobile devices, such as interaction failures, display problems, or navigation issues on smaller screens.

Methodology

1. Select only client-submitted feedback: Using declared use cases and internal signals, we filtered the dataset to include only feedback originating from clients (excluding internal/testing entries).
2. Anonymize all data: Removed names, URLs, company/person identifiers, and any other metadata; retained only the feedback text to protect customer privacy.
3. Classify with a local LLM: Ran each item through Gemini-2.5-pro on our own servers to assign one of eight categories; for the three largest categories, also assigned a sub-category. We used purpose-built prompts for overall and category-specific classification.
   • Prompt for Top level categorization
   • Prompt for Content & Copy sub-bucket
   • Prompt for Visual Design sub-bucket
   • Prompt for Functionality & Bugs sub-bucket
4. Aggregate and quality-control: Removed uncategorized items and obvious test entries. In total, 98.7% of all feedback mapped cleanly to the eight buckets.

Category definitions:

1. Visual Design - Photos, images, colors, fonts, styling, layout, animations, logos, visual appearance
2. Content & Copy - Text updates, titles, copy changes, spelling corrections, translations, contact info
3. Functionality/Bugs - Things not working, broken features, errors, malfunctions, display issues
4. User Experience (UX) - Navigation, usability, user flow improvements, ease of use, interaction design
5. Performance - Speed, loading times, mobile responsiveness, optimization issues
6. Feature Requests - Adding new functionality, missing features, new integrations, enhancements
7. Technical/Development - Code changes, SEO, accessibility, browser compatibility, backend work
8. Project Management - Timeline discussions, communication, process feedback, project coordination