How GiveBackIQ works
Every score and recommendation can be traced back to specific evidence. This page explains the rules, so you can judge the results for yourself.
Where the evidence comes from
- A safe website check. GiveBackIQ reads your public homepage and a small sample of linked pages (62 automated checks). It never logs in, fills in forms, or runs scripts.
- Independent sources. Google PageSpeed Insights, the Chrome UX Report, Google Places, Google's Knowledge Graph, YouTube, and IRS records from ProPublica Nonprofit Explorer.
- A GiveBackIQ review. We ask an AI assistant (an OpenAI model with live web search) two questions a donor might ask about your charity, and we review your homepage against a fixed checklist: is the mission clear, is Donate easy to see on a phone, is impact shown, is the language plain, and are real people shown. We also review your donation and volunteer pages as a phone shows them: can donors see the form right away, pick an amount, pay quickly, and finish in a few fields, and can volunteers sign up, see current roles, and know what happens next.
- A short set of questions. About 18 questions on things software can't see, such as how quickly you thank donors or reply to volunteers.
How we use AI
The homepage, donation page, and volunteer page reviews and the assistant check use AI models from OpenAI. The checklist, the questions, and how results are scored are ours, and every result shows its reasons so you can judge it. AI can be wrong and answers vary between runs, so each result is labeled with where it came from. The public text and phone screenshots of those pages are sent to OpenAI for the reviews; nothing you type into the questions is.
How scores are calculated
Each answer or check becomes a value from 0 to 1. A pillar's score is the weighted average of the items we know, multiplied by 100. Unknown items, including Not sure answers and checks that couldn't run, are left out of the score and lower its coverage instead.
A pillar is only scored when at least 25% of its weighted information is known. The overall score is a weighted average of the pillars that could be scored.
| Pillar | What it measures | Share of overall score |
|---|---|---|
| Search and AI Visibility | Whether people, Google, and AI assistants can find you and describe you correctly. | 30% |
| Presentation and Trust | Whether a first-time visitor quickly understands what you do and trusts you. | 25% |
| Member Experience | How easy it is to donate, volunteer, join, or attend, especially on a phone. | 30% |
| Engagement | Whether you keep donors, volunteers, and members once they act. | 15% |
Score ranges
- Critical0 to 39. Urgent gaps are likely costing you donors, volunteers, or trust.
- Needs Attention40 to 59. Important basics are missing or inconsistent.
- Developing60 to 74. A workable foundation with clear room to improve.
- Strong75 to 89. Well managed, with a few targeted opportunities.
- Leading90 to 100. Among the best practices we see.
How recommendations are prioritized
Recommendations come from documented rules tied to specific answers or checks. Each is rated for impact and effort. High impact, low effort work becomes an immediate quick win; larger projects are scheduled into days 31 to 90 or later. Urgent problems move earlier, and recommendations that match the goals you choose move up.
Website check safety and limits
- Only public http and https addresses on standard ports are inspected. Local, private, and cloud metadata addresses are always blocked.
- Every address, including each redirect, is checked before it is visited. Requests time out quickly and downloads are size limited.
- robots.txt directives are respected, and only a small number of links are checked.
- Only the homepage, a few key pages, and a small sample of internal links were read. Other pages may differ.
- GiveBackIQ's own page reading does not run JavaScript, so content or donation forms that load through scripts may not be visible to it. Google PageSpeed Insights does run scripts when it measures speed.
- The donation and volunteer page reviews see only the first step of a form. GiveBackIQ never fills in or submits a form.
- Automated checks cannot judge design quality, readability, or whether content is accurate.
- Checks that could not run are shown as Not Available and do not lower any score.