Cognitive behavioural optimisation: optimise for minds, not just metrics
Optimisation has got good at persuading people. The next leap is designing around how people think and the capacity they have in the moment. Eight practical steps, from pressure tactics to AI agents, to stop digital exclusion growing.
The technology has changed beyond recognition since I started out. Human behaviour hasn't. We still hesitate when we're unsure, stall when we're overloaded and walk away when something feels like hard work, so the core challenges of conversion rate optimisation (CRO) look remarkably familiar. CRO is the practice of testing changes to a website or service so more people complete an action, such as buying or signing up. What has changed is the stage. Journeys now span social, web, app, store, chat and phone, and too often they don't join up. Many organisations are also moving customers from phone lines to chat and AI agents. My work now spans special educational needs and disabilities (SEND) and inclusion with school leaders and the Department for Education, and mental health as a trustee at Mind in Haringey. That lens has crystallised an approach I call cognitive behavioural optimisation: optimising around how people think and the capacity they have, rather than pushing harder for the click. It has no connection to cognitive behavioural therapy. Here are eight ways to put it to work.
1. Treat every conversion as a cognitive event
Simply put, conversion happens in the mind before it happens on the page. In my 2017 CRO playbook I wrote: "Your brand experience is way more than a linear sales funnel. Each journey consists of many micro-events each with intent and purpose." I backed it with a formula, because I'm a data person and I can't help myself. It's my own working model, not a scientific law: C = 4m + 3v + 2(i − f) − 2a. Conversion rises with motivation, a clear value proposition and an incentive to act. It falls with friction and anxiety about committing. Three of those five factors live in the mind. Clarity is understanding, friction is mostly mental effort, and anxiety is uncertainty and stress.
2. Design for the hardest day
Friction and anxiety depend on the person. The same form can take one person thirty seconds and another their whole evening. Working memory, attention, reading ability, tiredness and stress all change the maths. Think of a parent filling in a school form at 11pm, or someone reading in their second language. Omnichannel journeys make this harder. Every handover between social, app, store and phone asks people to remember what they did and what was promised. Carry the context across channels and never make someone start again. Ask one question every time: how much capacity does this person have right now, and how much are we asking them to spend? Design for the person having the hardest day, and you've designed for everyone.
3. Double down on what already works
In my experience, many techniques that reliably win tests are cognitive accessibility patterns in a commercial suit. Plain English beats jargon. Short sections ease the load on memory. One clear recommendation beats six equal options. Letting people recognise rather than remember is one of Jakob Nielsen's usability heuristics. And icons work brilliantly, as long as they bring a word with them. The accessibility community got here too. The World Wide Web Consortium (W3C) guidance on content for people with cognitive and learning disabilities covers the same ground, and the Web Content Accessibility Guidelines (WCAG) 2.2 include criteria such as Redundant Entry (3.3.7). Two communities, one insight, different vocabulary. That's a huge opportunity.
4. Put pressure tactics through a cognitive check
Some techniques add pressure rather than reduce effort. In my view they need a cognitive check before going live. Countdown timers and "only 2 left" messages can make it harder to think clearly, so use them only when the urgency is real. The Competition and Markets Authority has acted on misleading urgency claims. Loss framing, rooted in Kahneman and Tversky's research, can tip into anxiety. Sticky buttons and overlays shouldn't cover what people are trying to read. Extras at checkout add decisions at the most demanding moment. And clever wordplay can confuse people who read language literally. The test is simple. Does this win by making the decision clearer, or by making it harder to think?
5. Don't let deflection become exclusion
Call deflection moves customers from phone lines to chat or AI agents, some of which can now take actions on someone's behalf. The commercial logic is clear, and done well these tools help. My concern is this. Chat relies on reading, typing and holding a conversation in your head. For many people with cognitive disabilities, anxiety or low digital confidence, the phone was the accessible route. Close that door without designing for cognition and I believe digital exclusion will grow. Keep a human within easy reach. Write agent replies in short, plain sentences, one question at a time. Let people pause and pick up later. Confirm before an agent acts. And be honest about containment: a chat that ends because someone gave up isn't resolved, it's lost.
6. Measure effort, not just outcomes
It baffles me that so many programmes still judge success on conversion rate alone. Form errors, re-entered information, back steps, time taken and people bouncing between channels all show where people struggle. Put guardrail measures such as returns, early cancellations, complaints and repeat contact on the same dashboard as conversion.
7. Bring lived experience into the room
Test with disabled, neurodivergent and less digitally confident people as standard, across every channel, and pay them properly. Bring accessibility specialists in at the design stage, not just after launch. The best ideas I've seen come from teams where data, design and lived experience sit side by side.
8. And finally, make it commercial by design
In my view, a conversion won by overwhelming someone is a loan, not a sale. It gets repaid through refunds, cancellations, complaints and regulatory risk, long after the test has closed. Across retail, software and education, the changes I've seen last are the ones that reduced effort for people with the least capacity. Agree up front which techniques are off the table, whatever uplift they promise.
A clearer future
A conversion only counts if the person understood what they were choosing, and had the headspace to choose it. The technology will keep changing. People won't. The future is bright if we design for that, and make sure the shift to chat and AI widens access rather than narrowing it. Time to stop chasing clicks and start earning them. Sources mentioned: W3C, Making Content Usable for People with Cognitive and Learning Disabilities (2021); W3C, WCAG 2.2 (2023); Jakob Nielsen, 10 Usability Heuristics, Nielsen Norman Group; Kahneman and Tversky, Prospect Theory, Econometrica (1979); Digital Markets, Competition and Consumers Act 2024.