Understanding the need for #AltAtSource through six real-world scenarios

Digital accessibility is often framed as a technical requirement, but in reality, it is a human challenge. When we ask content creators to manually add alt text every time they upload an image, we are relying on an unrealistic alignment of training, time, and motivation. We are essentially asking humans to be perfect database entry clerks — a task we’ve known humans are historically terrible at since the invention of the spreadsheet.

The #AltAtSource campaign advocates for a fundamental shift: embedding accessibility metadata — such as IPTC AltTextAccessibility and XMP dc:description — directly into the image file so it becomes portable. By combining this metadata with the power of AI, we can move away from a model of manual labour and toward a model of automated accuracy.

Let’s look at how this works in the real world.

Scenario 1: The boutique owner and the path of least resistance

Imagine the owner of a small online stationery boutique uploading hundreds of pen variants. Faced with a deadline and zero caffeine, the owner takes the path of least resistance, typing the word pen into every mandatory alt-text field just to bypass the system.

In this scenario, AI is a powerful ally. Unlike a tired human, AI doesn‘t get bored or frustrated by the 400th ballpoint. It can be trained on specific rules — such as “be succinct” and “avoid starting with image of” — that it implements reliably. If metadata isn‘t present, AI provides a consistent baseline description that the owner simply verifies, turning a daunting creative task into a quick quality-control check.

Scenario 2: The local venue and the fragmented identity

A local venue uses the same image of a concert hall for PR, event listings, and social media. Currently, this results in three different people writing three different descriptions. It’s a digital game of “Chinese Whispers” where the user loses every time.

#AltAtSource ensures the definitive description is baked into the file. If that image needs to be adapted for different contexts, AI can use that original metadata as a source of truth to generate variations. This maintains a golden thread of consistency across platforms that is impossible to achieve through manual entry alone.

Scenario 3: The public sector and the diligent dead end

In the UK, government departments and charities work tirelessly to conform to PSBAR (Public Sector Bodies Accessibility Regulations). They hire specialists to audit resource libraries and write high-quality alt text. It’s diligent, expensive, and noble work.

The tragedy occurs the moment a staff member downloads an accessible image from the central library and uploads it to a new landing page or a partner portal. Most current platforms strip out metadata or simply ignore it, meaning that expensive, expert-written alt text vanishes into the digital ether. #AltAtSource ensures that accessibility is a one-time investment that stays with the asset, rather than a recurring tax paid every time a file is moved.

Scenario 4: The professional photographer and the authoritative source

A professional wildlife photographer adds rich metadata to a photo — species, location, and lighting. While this data is invaluable, it is often technical. However, the photographer remains the authoritative source. They know the exact energy of the moment; they know if that “brown bird” is actually a rare subspecies or just a very muddy pigeon.

AI can bridge this gap. By using the photographer’s metadata as a foundation, AI can translate those facts into descriptive alt text. The photographer then reviews the draft, ensuring their expert knowledge is preserved rather than replaced by a generic guess from a journalist in a distant newsroom.

Scenario 5: The PR powerhouse and the broken press pack

A global PR agency spends thousands of hours (and even more of the client’s money) perfecting a press pack for a product launch. They obsess over the exact shade of “brand-approved teal” and the perfect lighting. They package these high-res assets into a ZIP file and send it to the media.

The news outlet, working against a ticking clock, unpacks the ZIP and hits upload. Because the alt text wasn’t embedded, the product is launched with a digital blindfold for millions of assistive technology users. The agency spent a fortune on visual branding only to make the product invisible to the Purple Pound. With #AltAtSource, the accessibility travels with the ZIP, ensuring the brand reaches every single potential customer.

Scenario 6: The hidden barriers of creation

We must consider the capacity of the person writing the description. A content author with a colour vision deficiency (CVD) may struggle to describe a colour-coded chart, while others may struggle with spelling or cognitive impairments that make descriptive writing a source of anxiety.

AI doesn‘t have off days. By leveraging tools to handle the heavy lifting of visual analysis, we remove the burden from authors who may have their own accessibility challenges. It allows the human to be the editor of an intelligent draft, rather than the struggling architect of a blank box.

Action, not invention

We cannot solve the accessibility gap by simply telling people to do better. For too long, the industry has relied on a cycle of chastising humans for their failures and then attempting to train them out of their natural tendency to find the path of least resistance.

It is time to change the model.

We need to stop treating accessibility as a manual tax on content creation and start treating it as a native property of digital media. The purpose of the #AltAtSource campaign is to remove the friction in the system by leveraging technology that already exists.

We do not need to wait for a new invention. We do not need to wait for every human on earth to become an accessibility expert. We simply need platforms to act on the tools at their disposal — to read the metadata we provide and use AI to fill the gaps. We have the solution; we just need the industry to hit import.