Sam needed to change a benefit payment date. Not cancel it, not query it — just shift it by four days so it would land after the rent went out rather than before. A small thing. The kind of thing that, in a different world, would take sixty seconds on the phone.
The phone line no longer existed. It had been replaced, eighteen months earlier, by a digital self-service portal. The announcement at the time had used phrases like improved accessibility and streamlined experience. What it meant in practice was this: the only way to change a payment date was through a website that required a verified digital identity account, a mobile phone capable of receiving authentication codes, and a browser that supported a security certificate the local library’s computers did not.
Sam had none of these things. Sam had a landline.
What the metrics were measuring
The organisation that ran this service had not made a reckless decision. The move to digital self-service had been carefully planned. User research had been conducted, pilot groups consulted. The numbers were compelling: a significant reduction in call volumes, faster query resolution for the majority of users, lower operating costs. By every metric they were measuring, it was a success.
The chatbot that handled the new service answered thousands of queries a week. It was available around the clock. It never put anyone on hold. For most people who used it, it worked exactly as promised. The efficiency gains were real.
What the metrics did not capture was Sam.
The system was a success. It just couldn’t help the person who needed it most.
The system worked exactly as designed
There is a version of this story that gets told as a technology failure — the chatbot was badly designed, or the digital channel was too complicated, or the organisation didn’t think hard enough about edge cases. That version is easier to fix, and easier to forgive.
The harder version is this: the system worked exactly as it was designed to work. The problem is what it was designed for.
When an institution decides to move a service to a digital channel, it makes a calculation. The calculation is usually honest about what it is optimising for — cost, speed, scale. What it is less honest about is who gets pushed out of it. The people who cannot reach the digital channel are the ones pushed out. And because those people are typically the same people who already have the least power, the least resource, and the least capacity to absorb disruption, the cost of being pushed out lands exactly where it can least be afforded.
This is not an oversight. It is a structural choice. Calling it progress obscures what it actually is: the transfer of cost from the institution to the individual.
What the efficiency report doesn’t show
Sam spent three hours on the day the payment was due trying to find another route. The website’s help section linked to a phone number. The phone number explained, in an automated message, that calls were no longer being taken and directed callers back to the website. Sam wrote a letter. The letter would take ten working days to process.
The rent went out first. The payment arrived four days late. There was a charge — small, but it meant the following week was tighter, which meant a decision about food, which meant the school lunch that didn’t happen, which meant a conversation with a teacher that Sam did not want to have.
None of this appears in the organisation’s efficiency report. There is no metric for it. The chatbot’s performance dashboard shows response times, resolution rates, and user satisfaction scores from people who were able to use it. It does not show Sam. Sam is the externality.
What the design is actually doing
The question here is not whether digital self-service is a good idea. For most people, in most circumstances, it is. The question is whether it meets the standard it should be held to.
For the majority of users, the service made things easier and gave them access at any hour. For Sam, it removed the one channel that had previously worked. That is not assistance. That is substitution — and substitution is only equivalent if what replaces the original is equally reachable.
For the organisation, the technology added real capability: speed, scale, availability around the clock. For the individual excluded from it, nothing was added. The service simply disappeared.
Where the system fails most completely is in responding to a person’s actual circumstances — no smartphone, no broadband, no compatible device. A system that cannot bend to meet those circumstances is not responsive. It is uniform. And uniformity, applied to an unequal starting point, produces unequal outcomes.
A technology that helps most people and harms the most vulnerable is not a success by any measure that takes people seriously. The question of who bears the cost of efficiency is not a detail to be addressed in version two. It is the central design question, and it belongs at the start of the process, not appended to it.
My opinion
I don’t think anyone at this organisation set out to fail Sam. That’s what makes it hard to fix. A project can hit every target it was given and still leave someone worse off than before, because Sam was never part of what the target was measuring. Success defined only by what most people experienced was always going to look this clean. Working and helping are not the same thing, and until an organisation measures the second, it will keep mistaking one for the other.
The questions the report doesn’t ask
Has a service you rely on moved to digital-only in the past few years? If so, did you notice what it stopped offering — and whether you noticed because it affected you, or because it affected someone else?
If you have had an experience like Sam’s — or supported someone through one — what actually happened in the end? How did the system eventually respond, and what did that cost?
If you were responsible for this service, what would you have done differently? Not in hindsight, but at the point when the efficiency case was being made — what would it have taken to hold the equity question alongside the cost question from the beginning?
Is digital-first, as it is currently practised, genuinely a more accessible model — or is it a more accessible model for people who were already well served, and a worse one for everyone else?
If this rings true, follow The Next Evolution at neilcatton.substack.com. New essays every week, on the technology that shapes us and the choices still available to us.
Authors Note
Sam is a fictional character. Their story is drawn from a combination of professional observation and personal proximity to real events. The experiences described are real. The person is not.
You’re reading The Next Evolution by Neil Catton, articles that explore the human world and the intersection of technology, they try and ask difficult questions - not to scare - but to inform. If someone forwarded this to you, you can subscribe free at neilcatton.substack.com.
Neil Catton is the author of The Next Evolution, The Cognitive Crucible and The Shadow System - available on Amazon, and writes at the intersection of technology, ethics, and human purpose.