InMarketing Today

Nearly a dozen industry bodies, including Innovate Finance, TheCityUK and techUK, have written to the Chancellor, John Healey, urging him not to drop the 12-month deadline for a government digital assets strategy, City AM reports today. The deadline sits in Clause 50 of the Financial Services and Markets Bill, added through a Conservative amendment, and the sector fears the government will use its Commons majority to remove or water it down at a vote next week.

The letter asks for a "north star", and the phrase does real work. A national strategy is a positioning statement: it tells investors, founders and talent where Britain intends to compete. The investment minister, Lord Pitt-Watson, argues the deadline risks "cutting across work currently under way" at regulators, which is the argument every marketing team hears just before a launch slips. The letter's sharpest line is aimed at complacency: the UK "should not assume that its existing strengths will automatically translate into long-term leadership." Without a published direction, other jurisdictions get to tell that story instead.

If you run a digital asset, payments or custody business, watch next week's vote and prepare two versions of your client and investor message: one for a UK with a dated plan, one without. Then apply the same test to your own firm. If your digital assets strategy has no published milestone, clients will assume it has no priority.

The Hong Kong Monetary Authority (HKMA) used Hong Kong FinTech Week today to announce four initiatives under its Fintech 2030 plan. Project SAGE will explore agentic artificial intelligence (AI) in banking alongside a governance framework; a post-quantum cryptography toolkit will help banks move to quantum-resistant encryption; Project Ensemble now supports round-the-clock settlement of tokenised assets; and Hong Kong's Faster Payment System will link with UnionPay International by year-end, opening QR code payments at more than 100 million merchants across Hong Kong, the Chinese Mainland and over 40 other markets.

Regulators rarely market themselves well. This is how it is done. Every initiative has a name, a pillar it sits under and, where possible, a date. The HKMA also published a discussion paper on banks' use of agentic AI and invited the industry to respond, which turns a regulator's homework into a conversation it leads. Chief executive Eddie Yue's line, that Hong Kong is "well placed" to capture fintech growth, is ordinary. The four named projects behind it are what make it credible.

If you run a bank or fintech with Asian clients, read the agentic AI discussion paper and decide whether to respond: firms that help shape governance early get to explain it to clients later. If you are deploying AI agents anywhere, write down now who approves what an agent may do on a client's behalf. Regulators are about to ask, and clients will ask first.

Generative artificial intelligence (AI) now lets marketers with no coding skills explore customer data, compare groups and write analytical code, but finding a pattern is not the same as explaining it, Kath Pay, chief executive of Holistic Email Marketing, argues in MarTech today. She cites an analysis of a UK retailer's transactions in which repeat customers made up 65.6% of customers and generated 92.9% of revenue: accurate figures that say nothing about why those customers returned or which retention tactic would bring more back.

This is where AI-assisted analysis goes wrong in a regulated firm. The figures can be right while the conclusion is wrong, and a correlation dressed as insight soon becomes a segment, a campaign and a board slide. Pay's sequence is the discipline: AI finds relationships worth investigating, behavioural science suggests explanations, and testing shows whether changing anything moves the result. Skip the middle steps and, in her words, you are "simply generating more sophisticated assumptions".

The next time your team brings you a pattern AI has found, ask for three possible explanations before anyone proposes a campaign. Then fund one test that measures the incremental effect on behaviour, not clicks. A pattern that cannot survive the question "why?" should not reach a client.