Q: What is best practice now?
A: Data transformation is one of those phrases that can mean everything and nothing, so it is worth being clear about it. At its simplest, it is about changing how a business collects, works with and uses data, so that data, and increasingly the AI built on top of it, becomes central to the value you create rather than a by-product of doing business. For publishers, that is a tantalising prospect. Every piece of journalism, every event, every subscriber interaction generates data, and most of us are putting a fraction of it to work.
It helps to think of the work in phases. The first is to settle on a thesis: not a technology, not a tool, but a clear statement of the value you want to unlock. It might be extracting more from the data your core journalism already produces, or turning a few days of event attendance into a year-round relationship. Whatever it is, that thesis becomes the thing every later decision is measured against.
The second phase is mapping the current state honestly: where data comes from, how it flows, how it is captured, stored and used, who works with it and what they are trying to achieve, and where the painful bottlenecks sit. You weigh the skills you have against the ones you will need. Only then comes the third phase, designing the future state. The decisive work here is agreeing what your data actually means. A shared data dictionary and a sensible ontology sound dry, but without them, teams talk past each other and build reports nobody quite trusts.
The fourth phase, implementation and migration, is where most of the difficulty lives, and most of it is human. Getting colleagues to work differently, untangling processes that have calcified over years, and staying focused when distractions present themselves is what really tests a programme. Success tends to look iterative rather than monumental, with frequent, visible releases that prove the direction is right and keep everyone bought in.
The prize is commercial as much as operational. Done well, data stops being something you wrangle with and becomes something that earns its keep: sharper reporting, automation of low-value tasks, and new products built on the insight your data holds. This is how a publisher moves up the value chain, from describing what happened towards predicting and shaping what happens next, and increasingly where new revenue is found.
Q: How do you see it changing in the future?
A: AI raises the stakes considerably. Agents are starting to appear throughout the workflow, and over the next few years, they will take on a growing share of the low-value data tasks that currently eat up expensive expert time. That is no small thing: it frees your most knowledgeable people to spend more time with customers and the wider industry, where they add real value.
It also makes the foundations matter more, not less. There is a real risk of a widening gap here. Organisations with clean, well-structured, well-understood data will be able to deploy AI at scale, and they will pull away quickly, launching products faster, deepening their expertise and absorbing acquisitions with far less friction. Those whose data is a tangle will find that the same clever tools simply amplify the mess.
Skills will shift too. I expect to see far more data literacy in jobs we do not currently think of as data roles. The ability to prototype an idea, clean a dataset or pull in third-party data will become ordinary rather than specialist, and the publishers who help their people build those skills are the ones who will keep hold of their best talent.
Q: What are your three top tips?
1. Start with the outcome, not the technology. It is tempting to begin with a shiny platform, but the publishers who succeed begin with a clear thesis about the value they are chasing and work backwards from it. Technology is the easy thing to change later. Setting off in the wrong direction is far harder to recover from.
2. Invest early in shared meaning. A common data dictionary and a sensible ontology are the least exciting items on any roadmap and the most decisive of all. Get everyone understanding data the same way before you scale anything, because every downstream product, report and AI agent depends on that agreement holding.
3. Treat it as a change programme and ship value often. This is as much about people as it is about pipelines, so resist the urge to disappear for eighteen months and re-emerge with something enormous. Small, regular releases keep momentum up, surface problems while they are still cheap to fix, and give the sceptics in the building a reason to believe. Data transformation is won in increments, not in one grand reveal, and the publishers who remember that are the ones still standing when the work gets hard.
About 67 Bricks
67 Bricks is a data transformation and product development consultancy with 20 years in the information industry. We help B2B media, data and information businesses grow their value by uncovering the potential worth in content, building platforms and products that turn it into revenue, and leaving a more defensible, more valuable business.
Email: will.bailey@67bricks.com
Website: www.67bricks.com
This article was first published in Issue # 1 of Best Practice in Publishing, a new publication from InPublishing. Click here for links to the other ‘best practice’ articles from the publication.
