The black box problem is not abstract Food delivery riders are currently working with academics to reverse-engineer the algorithms that determine their pay. They say the systems have cut their earnings, but they cannot prove it because they cannot see how decisions are made. The platforms call it proprietary. The riders call it a black box. This is not a niche labour dispute. It is the shape of work now. If you run a small business that depends on a platform, you are in a version of this situation. Maybe you sell through Amazon or Etsy. Maybe your clients find you through Google or Instagram. Maybe you drive for a ride-share app or list properties on Airbnb. The algorithm decides who sees you, when, and at what price. You do not get to see the formula. I have worked with people in London and beyond who built real businesses on these platforms, then watched their income drop 30% in a quarter with no explanation. No policy change was announced. No warning was given. The algorithm just shifted. What you actually cannot control Let me be direct about this: you cannot control the algorithm. You cannot see it. You cannot appeal to it. You cannot negotiate with it. Platforms are under no obligation to explain their ranking systems. Even when they publish guidelines, those guidelines describe inputs, not weights. Knowing that "customer reviews matter" tells you nothing about whether a 4.7 rating puts you on page one or page five. You also cannot control when the rules change. Instagram's algorithm shifted in 2023 to favour Reels over static posts. Creators who had built audiences around photography saw their reach collapse. Some adapted. Some did not. None were consulted. This is the deal you made when you built on rented land. I am not saying it is fair. I am saying it is the deal. What you can control The list is shorter than you want, but it is real. First, you control your customer list. If someone buys from you through a platform and you have no way to contact them directly, you have a transaction, not a relationship. Every platform makes it hard to export customer data. Do it anyway. Build an email list. Collect phone numbers. Create a reason for people to come to you directly, even if they found you somewhere else. Second, you control your margins. When a platform takes 15-30% of every sale, your pricing has to account for that from the start. I have seen people price for the platform, then panic when fees rise or visibility drops. Price for independence. If the platform becomes unworkable, you need enough margin to survive the transition. Third, you control your data. Not the platform's data about you, but your own records of what sells, when, to whom, and at what cost. One person I worked with last year discovered that 40% of her revenue came from repeat customers who always ordered the same three products. She had never noticed because the platform's dashboard did not surface that pattern. Her own spreadsheet did. Fourth, you control your dependencies. If 90% of your income comes from one platform, you are not running a business. You are a contractor with one client. Diversification is not a growth strategy here. It is a survival strategy. The real question underneath The delivery riders are not just asking to see the algorithm. They are asking a bigger question: who decides what your work is worth, and on what basis? Platforms answer this question with data. They measure speed, ratings, acceptance rates, cancellation rates. They weight these factors in ways they do not disclose. The result is a number that determines your income. But data is not neutral. The choice of what to measure, and how to weight it, is a decision made by people with interests that are not yours. The algorithm is not a referee. It is an employee of the platform. This does not mean platforms are evil. It means their incentives are not aligned with yours, and you should plan accordingly. Building off the grid I am not suggesting you abandon platforms entirely. For most small businesses, that is not realistic. Platforms offer distribution you cannot replicate on your own. But I am suggesting you build something parallel. A channel that belongs to you. A way to reach customers that does not depend on an algorithm you cannot see. This takes time. It is slower than the platform. It does not scale as fast. But it is yours. One business owner I worked with through alira.london spent six months building an email list of 800 people. When her primary platform changed its fee structure, she moved 60% of her sales to direct orders within eight weeks. The email list was not a backup plan. It was the real business. The platform was just the acquisition channel. What to do this week Audit your dependencies. Write down every platform that affects your income. Next to each one, write the percentage of your revenue that would disappear if you lost access tomorrow. If any number is above 50%, that is your priority. Start the list. If you do not have a way to contact customers directly, fix that this week. Add an email signup to your website. Offer something small in exchange. The goal is not to build a marketing machine. The goal is to own the relationship. Export your data. Pull whatever customer and sales data you can from every platform you use. Put it in a spreadsheet you control. Look for patterns the platform dashboard does not show you. This takes an hour. Do it before the algorithm changes again.