How to Build Your Own Management System

Career tests at his 1990s school kept calling Mikhail a technical type, though he was set on law. He landed by chance on a preparatory course at the Moscow Institute of Electronics and Mathematics, picked materials science there, then fell short of the marks for Bauman and stayed; he also applied to a flying school and is glad he was not accepted. In postgraduate study he passed the candidate minimum, wrote half a dissertation on superconductivity in thin films and published several papers — until he brought a professor a strange effect. The professor told him to raise the temperature and pressure, the effect dissolved: it was dirt on the sample: a silicon wafer is first studied in a sterile room in the States, then handed down through institutes and finally given to students free. Eighteen months to two years had gone into researching dirt. Mikhail quit science, drew an animation in Flash, was hired to lay out e-learning courses, became a tester, tweaked the LMS engine with a book on ActionScript and certified free as a Lotus developer and administrator.

He then sold licences from IBM's leads, wrote requirements, ran projects and grew into chief executive of his own company, with offices in Belarus and St Petersburg. Their anchor client was a large bank Mikhail declines to name: it had poached a manager out of Alfa-Bank, who arrived with a requirements document marked Lotus Notes at the bottom. Ever since, "you are our favourite client" has been a red flag to him — it means the price goes up. The Lotus document database held about a million records while that client generated more complaints than that, so the team lived on workarounds; in interviews Mikhail hired those who invented one rather than those who said it was impossible. The hardest stretch was the cash gap: up to eighteen months between promised money and payroll, an owner who opened a notebook of debts and refused, half a year on cut salaries. Mikhail gave three of his best people to an acquaintance from Alfa-Bank — and eighteen months later that recommendation came back as a job running an analytics group.

The loyalty project he had joined on arrival was closed three weeks later, and Mikhail was moved onto an operational CRM rollout. There he saw enterprise discipline for the first time: some sixty processes, daily meetings and a workaround prepared in advance for every process in case it failed after launch. His own normal until then had been a developer opening a console straight on production, nothing working for a day after a delivery, some five hundred people sitting idle. When former colleagues rang offering more money and a bigger share to come back, he refused: returning meant returning to his own cage, however golden. From then on he was quietly handed everything new, the segment strategy among it, and directions grew around them. Hence his rule: you take something on out of interest and show a result; the resource arrives afterwards. Alexander adds that this is how the generative culture of growing companies works, while stalled ones fill with committees and haggling over people before commitments.

Mikhail led the separate loyalty stream himself and decided to buy a product on the market: a newly arrived manager brought in a vendor he had launched loyalty with at another bank, whose cloud solution now had to run inside the bank. Half a year of promises and demos later, the delivered solution turned out to be built on Microsoft Dynamics, with four hundred fields and no customer-facing interface; the demo had been a picture. Deployment took three months and would not start on the bank's infrastructure; three weeks went into hunting one bug, and the answer was that the field had to be set to three, nobody knew why. When an agreed fifty-fifty split came back in the contract as a different figure, Mikhail wrote a resignation letter. Martin Peletsky, then a top manager, talked him out of it: the bank had just spent a sum with many zeros on his education, no one else would invest that much, so stay and I will mentor you. The vendor system ran for about a week and was replaced by the team's own parallel build.

The product transformation revealed that one retail product required changes in twenty-five to thirty systems. An autonomous team assembled to cover them all swelled to fifty people: the product owner could load two or three systems, and the rest took any work rather than sit idle. Mikhail, by then account manager for the retail business, put direction leaders with real coordination skills above the Scrum teams — that structure was built in 2017–2018 — and took over a directorate in 2018. The next role, the digital partner, was meant as an interface between business and IT, yet it almost always sided with the business: instead of one customer there were two, and the second understood how development worked. Competence centres closed their quarterly goals, but during an incident each reported that all was well on its side, and nobody owned the mobile bank as a product. From 2019 domains began to be handed to CTOs — a change that, by Mikhail's account, came bottom-up and turned IT from reactive to proactive. Alexander's verticals were getting their own CTOs too: Pasha Permyakov ran investments.

Mikhail first treated headcount as the measure — five hundred employees delighted him and a thousand made him jump for joy — before changing the criterion to the volume he influenced: an investment product was built by a hundred people where a comparable player spent several hundred. It became frightening when every change demanded a new team: inefficiency was plugged with people, hiring quality slipped, and the system was held together by tech leads. Hence the bet on automation and a platform — and the conflict he calls "I am the platform, you are the fool". That is what he had tried to remove with a coordination centre and a stack of resignation letters at a strategy session: nobody signed, everybody took offence, and he concluded that cooperation cannot be ordered, only agreed as mutual value. He has no working answer yet; his benchmark is Apple, where teams of about ten build products on a strong platform, and his second goal is end-to-end CTO ownership including support. Banking is turning into a commodity, the fastest experimenter wins, and he set up the closed CTO Dialogues club so peers could compare reality without the marketing facade.