Why generic AI content is making your charity’s web traffic disappear
The Charity Digital Skills Report 2026 reveals that 81% of UK charities have advanced digitally and 79% now use AI. But look closer at the numbers.
Trust in AI has plummeted across the sector. More than 35% of charities do not trust the digital tools they use. Squeezed finances are the biggest barrier to progress for 63% of organisations. Only 17% of charities secured digital grants this year, down from 30% last year.
This is hurting the charity sector. Relying on public AI tools like ChatGPT to write bids or website articles produces generic content that funders ignore. AI search engines now answer donor and beneficiary questions directly, so your website is often left out. That is why many charities are losing organic website traffic.
To stay visible and credible, you need to move past generic AI content and use secure, evidence-based workflows and AI search optimisation.
AI slop in grant writing
Charities are rushing to use AI for grant fundraising. 45% now use AI to write bids, and for small charities in London, it is 76%.
But sending more bids with AI has backfired. Fifty-eight per cent of grantmakers say AI is changing the bids they receive. They report a huge increase in volume and repetitive phrasing. Yet 55% of funders admit they do not know which applications are AI-assisted!
Using public AI models to send more bids is a race to the bottom that leads to more rejections and puts extra pressure on your funding team.
The problem is not understanding how these AI tools work. Public models generate text based on patterns, not evidence. If you use a generic prompt, you get a generic answer. The result may sound polished, but it lacks detail. Funders reject bids that lack clear evidence, do not meet their criteria, or raise doubts about your ability to deliver. Generic AI text makes this worse by making bids look finished when they are not.
To win, your AI tools should connect directly to your data, such as outcome metrics, local demographics and past case studies, so every bid is based on facts.
This is especially important for underfunded groups. Many majority-led and neurodivergent-led charities use AI for grant bids, but free public models just produce more of the same, leading to more rejections.
When you use your own Charity Commission data and internal outcomes, make your bids stand out and reflect your real voice.
Our AI Engines and Workflows build evidence-grounded drafts directly from your secure database.
| Dimension | Evidence-grounded agentic drafting | Traditional manual drafting | Generic ChatGPT prompting |
|---|---|---|---|
| Time efficiency | Fast Minutes with source validation |
Slow Hours to days per draft |
Fast Minutes per draft without proofing |
| Accuracy | High Sources cited, traceable outputs |
Reliable High accuracy, but subject to human error |
Variable May hallucinate or omit critical facts |
| Consistency | Consistent Strictly uses pre-defined templates and system rules |
Variable Highly dependent on author skill & attention |
Inconsistent Lacks standardized, deterministic structures |
| Source traceability | Automated Every claim generated is linked directly to evidence |
Manual Manual citation, highly prone to accidental omission |
Lacking Often lacks explicit or verifiable sources |
| Scalability | Highly Scalable Seamless scaling backed by quality control mechanisms |
Hard to scale Strictly limited by available staff bandwidth |
Easy scaling Scales instantly, but quality and safety vary wildly |
| Compliance | Built-in Continuous compliance checks and clear audit trails |
Manual check Relies completely on individual author's diligence |
High Risk Risk of compliance breaches or copyright issues |
Where has charity website traffic gone?
Charities keep spending their budget on old-school SEO and website redesigns, but user behaviour has changed for good. ChatGPT, Gemini, and Google AI summaries now replace the first page of search results. If your site is not set up to be cited by these tools, your audience will not find you.
What’s a zero-click search?
Users get summaries on the search page and never visit your site. Bain and Company found that 80% of search users rely on AI summaries at least 40% of the time. 60% of searches now get no clicks. The Patient Information Forum found that only one in ten people click through to a charity website after reading an AI summary.
Vulnerable people are left without your guidance and may be exposed to misinformation due to limited human oversight. As a result, you miss out on valuable donor data and potential volunteer registrations. Meanwhile, your expert content is used to train commercial AI models. Which then fail to promote your brand, further diminishing your organisation's visibility and impact.
To stay visible, your digital team must focus on AI search optimisation. This is called Generative Engine Optimisation, or GEO. GEO means structuring your website so AI search engines can find, trust and cite your brand. It is not just about keyword rankings. You need to answer real questions and make your site’s code easy for machines to read. Talk with me about adjusting your site structure and organising your content so AI tools recognise your authority right away.
| Dimension | Generative Engine Optimization (GEO) | Traditional SEO |
|---|---|---|
| Primary goal | Direct Citations Achieve direct citations in generative AI engines (ChatGPT, Perplexity, Gemini) |
Search Rankings Rank highly in traditional search engines (Google, Bing) |
| Content structure | AI Parsing Structured heavily for AI models: FAQs, dynamic tables, structured schemas |
Crawler Friendly Written primarily for human readers and search engine crawlers |
| Optimization tactics | Conversational Context Direct answers, structured data, high semantic conversational context |
Keywords & Metadata Keywords, backlinks, meta tags, and site technical speed |
| Result visibility | AI Synthesis Integrated AI-generated responses with cited interactive sources |
SERP Listings Standard organic search result listings and blue links |
| Measurement | Citation Analytics Source citations in AI outputs, brand share of voice, API analytics |
Clicks & Impressions Clicks, impressions, CTR, and SERP keyword positions |
| Adaptation speed | Real-time / Rapid Fast—AI engines update frequently, requiring continuous tuning |
Delayed Slow—highly dependent on core search algorithm rollouts |
The leadership gap and the Fractional Digital subscription model
Implementing a GEO strategy can be difficult due to a knowledge gap in leadership.
One third of boards and a quarter of CEOs lack AI skills. Half of charities do not have a trustee with digital experience.
Tight budgets are the main barrier to digital progress for most charities. Hiring a senior Head of Digital at £60,000-£80,000 plus overheads is not realistic.
Digital tasks get handed to admin or programme staff who are not trained for them. Or you pay high project fees to a firm. They deliver a project but can leave your team with disconnected systems and no ongoing strategy.
I take a different approach. For a flat monthly fee, you get a Head of Digital who runs your marketing and systems. You get senior leadership and delivery, without the cost or risk of a full-time hire.
This subscription model is really important when only 10% of small charities in the South West accessed digital funding this year.
My Fractional Head of Digital service removes these barriers. You get senior guidance from the start and can cancel with one month’s notice.
| Operational vector | Fractional subscription model | Full-time hire | Legacy firm |
|---|---|---|---|
| Cost | Optimized Lower—pay for targeted outcomes with maximum flexibility |
High cost High—heavy burden of salary, benefits, and physical overhead |
Premium High—burdened by massive client retainers and firm overhead |
| Expertise depth | Evolving specialists Deep specialists embedded directly, leveraging up-to-date modern skills |
Isolated Deep in one organization, but skillsets are prone to stagnation |
Diluted Broad structural footprint, but often generic and slow to adapt |
| Responsiveness | Always-on Always-on support structure paired with a rapid ramp-up window |
Siloed Availability completely rests on and is limited to a single person |
Transactional Billed strictly by the hour or project; notoriously slow turnaround |
| Continuity | Institutional memory Shared knowledge and institutional memory preserved across clients |
High risk Severe risk of turnover, resulting in immediate organizational knowledge loss |
Unstable Highly prone to sudden staff changes and constantly shifting account teams |
| Scalability | Elastic Flex up or down instantly based on needs with entirely predictable costs |
Friction-heavy Hard to scale quickly; slow onboarding and operational overhead |
Contract-locked Requires entirely new contracts and steep operational cost increases |
| Innovation | Cutting edge Constantly updated; actively leverages the latest tech and systems |
Reactive Likely to resist rapid change due to internal politics and habits |
Lagging Slow adoption rate hampered by a conservative structural approach |
Rebuilding trust and security with dedicated AI
Most charities use AI without clear guidelines. More than half lack an AI policy, and almost half are not managing AI risks. This creates serious legal risks. Staff may paste sensitive beneficiary notes and donor lists into public AI tools, violating GDPR and putting vulnerable people at risk of data breaches.
We solve this by creating secure private AI search engines that connect directly to your systems. Staff can ask questions and get accurate, cited answers without moving your data outside your organisation. Your data stays secure in Microsoft Azure UK or EU data centres.
Vulnerable and marginalised-led charities face the biggest safety barriers here. Among disabled and deaf-led charities, 46% struggle to cover digital costs. Yet they support 73% of their communities with digital inclusion.
Do not rely on insecure, piecemeal integrations. Build security into your systems from the start by using certified frameworks such as Cyber Essentials.
Explore how we handle Trust, Compliance, and Security to protect your organisation.
| Risk vector | Secure private AI engines | Public AI tools |
|---|---|---|
| Data leakage | Isolated Minimal—data is completely isolated in private environments and never used for training |
High risk High—user content may be used for model training or exposed to third parties |
| Compliance risk | Low Risk Low—customizable controls, full audit trails, and strict jurisdictional choice |
High Risk High—often unclear, non-compliant GDPR policies and sector compliance gaps |
| Brand trust | Maintained Maintained with strict oversight, structural transparency, and human-in-the-loop validation |
Erosion risk Erodes quickly if sensitive data is mishandled or AI hallucinations occur |
| Security of integrations | Strong Strong—controlled access, thoroughly vetted integrations, and organization-wide policies |
Weak Weak—limited, fragmented, or highly unreliable API governance |
| Customization | Extensive Extensive—fully tailored models and integrated workflows for specific needs |
Limited Limited—rigid, one-size-fits-all model structures |
| Operational control | High Control High—organization controls version updates, user access, and fallback procedures |
Low Control Low—unannounced service disruptions or model updates outside your control |
Reclaiming time for human connection with AI Engines and agentic workflows
Traditional automation just moves data from one place to another. Agentic AI goes further. It uses specialised AI agents to handle complex workflows such as interpreting grant guidelines, pulling metrics from your CRM, and logging activity. A human always approves the final work. This approach replaces manual admin and tackles your team’s biggest bottlenecks.
Grant reporting is a great example. Plinth’s research shows that UK charities spend 15.8 million staff hours every year on manual grant monitoring. This costs the sector £204 million in lost staff time. Our custom AI Engines and Workflows connect directly to your systems such as SharePoint. They compile reports automatically using real outcome data.
In our workflows, this means a 90% reduction in manual data processing time.
Neurodivergent-led charities benefit most from this. Staff who are autistic, dyslexic, or have ADHD say that programmatic transcriptions and secure AI synthesis reduce cognitive load and admin stress. But this work needs careful implementation. About 45% of neurodivergent-led charities say their colleagues are hesitant about AI.
Midday builds custom workflows that keep your team in control of every decision. We handle the background admin, so your staff can focus on fundraising, helping clients, or building real relationships with donors.
| Process bottleneck | Automated agentic pipeline | Manual administrative path |
|---|---|---|
| Donor data entry | Real-Time Sync AI agent automatically pulls transaction logs directly from your payment processor and updates the CRM instantly. |
Manual input Administrative staff must manually type, verify, and input donation details into the CRM. |
| Grant deadline tracking | Auto-monitoring AI agent monitors external grant portals, calendars, updates deadlines, and triggers automated push reminders. |
Spreadsheet log Staff must manually maintain static spreadsheets and set personal calendar alerts. |
| Report generation | On-Demand Drafts AI agent dynamically gathers system metrics and generates beautifully formatted reports for stakeholder review. |
Manual compile Staff manually extract data points across systems, format charts, and build reports by hand. |
| Volunteer scheduling | Automated match AI agent instantly parses incoming availability, maps ideal schedules, and sends dynamic confirmation emails. |
Back-and-forth The volunteer coordinator spends hours sorting email threads, confirming dates, and updating calendar invites. |
| Gift aid claims | Pre-checked / Auto AI agent systematically checks criteria eligibility, pre-fills government forms, and submits claims with minimal human touch. |
Manual audit Finance team manually reviews transactional logs, flags eligible records, and submits individual claim paperwork. |
Phased transition plan to make your charity AI-ready
This step-by-step plan gets your site ready for AI search and secures your workflows.
| Stage | Focus area | Core activities | Metrics |
|---|---|---|---|
| Stage 1 | Infrastructure Technical audit |
Audit robots.txt rules and configure Cloudflare AI bot blocks to control model access. | Secure crawler access |
| Stage 2 | Accessibility Code structuring |
Implement highly semantic HTML5 structures and structured schema markups. | Clear index eligibility |
| Stage 3 | Visibility Content optimization |
Rewrite core landing pages into highly scannable, dense FAQ blocks and key data tables. | Direct citations in ChatGPT |
| Stage 4 | Privacy First Data security |
Transition internal data operations off public tools into private Azure OpenAI RAG setups. | Zero data leakage risk |
| Stage 5 | Automation AI Engine setup |
Connect live business CRM APIs and construct highly customized action-agent pipelines. | 90% manual time saved |
| Stage 6 | Oversight Strategic governance |
Draft comprehensive corporate AI use policies and schedule recurring board performance audits. | Fully compliant operations |
Bring in a Head of Digital to get your digital projects moving
At Midday, I work directly with your team as a senior advisor. I turn your data into board-ready metrics so you can make confident decisions. Every partnership starts with a flat monthly fee, a clear four-month trial, and the option to cancel at any time.
Check our Midday Pricing options to see what fits your organisation. Book a discovery call with Adam Weston to find out what is holding back your digital performance.