An in-house legal team of ten to fifty has the hardest version of this problem. It is too small to fund a dedicated regulatory affairs function and too large for the general counsel to read everything personally. The result in most teams is a shared mailbox, three or four newsletters, and a quiet dependence on outside counsel to mention anything important. That works until the quarter it does not.
Start from your obligations, not from the regulators
The instinct is to list the regulators that touch your sector and watch all of them. That produces a long feed and a team that skims. Invert it. Write down the ten obligations that would actually hurt if they changed: the consent basis your marketing relies on, the permissions your product sits inside, the retention periods in your standard terms, the reporting deadlines you already struggle with.
Now work backwards to the sources that would publish a change to each. You will usually find that ten obligations resolve to three or four bodies, and that two of those bodies account for most of the risk. That is your watch list, and it is defensible in a way that a list of every regulator in your sector is not.
Three watches beat eleven
A watch that produces four items a week gets read. A watch that produces forty gets a folder rule. This is not a statement about attention spans; it is arithmetic about how much triage a team of twelve can absorb alongside its actual work.
So start with three: your primary conduct regulator, the ICO if you hold personal data at any scale, and whichever body publishes the rules your commercial contracts are drafted against. Run those for a quarter. Add a fourth only when the first three are being read and dispositioned every week, rather than because somebody suggested it in a meeting.
Materiality is a judgement about you
A consultation paper is not high materiality because it is long. It is high materiality if it changes something you do. Three bands are enough, and the definitions should be written in terms of your organisation rather than the document.
High: this changes an obligation we currently rely on, and somebody must do something before a date. Medium: this signals a direction that will affect a decision we are making this year, so a named person should read it properly. Low: we should know it happened, and no action follows. If your team cannot allocate a development to a band in thirty seconds, the definitions are too abstract.
Key dates, and the person who accepted them
Two dates matter and they are different. The publication date tells you what you knew and when, which is the date that matters if anybody asks later why nothing was done. The deadline, whether a consultation close or a commencement, is the one that goes in the calendar.
Never record a date without an owner, and never assign an owner without telling them. An unaccepted owner is a name in a spreadsheet, and an item with a date but no accepted owner is the most common way a horizon scanning process fails. It looks managed right up until the week the deadline arrives.
How the feature works here
Horizon Scanning in LegalAI Space runs watches on UK and EU regulator sources, including the FCA, the ICO and the SRA. Each watch produces three tabs. Developments carries the items, each with a materiality band, a publication date, a deadline where the source states one, and an owner. Key dates rolls the deadlines into one view. Sources shows what was actually read, which is the tab that tells you whether a quiet week was quiet or broken.
The item then needs to leave the system as something a human will read. Before a run you choose who the output is for: an executive brief for the board, a work-stream memo for the team, or a client update in plain English. That choice changes the writing rather than just the formatting, which matters when the same development has to reach a general counsel and a commercial director in the same week.
From development to board note in one pass
The pattern that works is four sentences and one date. What changed, in the regulator's own terms. What it changes for us, in ours. What we propose to do. Who owns it, and by when. Anything longer gets skimmed by the board and anything shorter gets sent back with questions.
Attach the source, always, and let it be the first link. A board that can click through to the FCA or ICO page is a board that stops asking whether the summary is reliable, which saves more time over a year than any other habit in this list. Every authority in an output carries a verdict, and a 'Where we looked' panel shows the sources searched, so the provenance travels with the note rather than living in the author's inbox.
A quarterly review that takes an hour
Once a quarter, pull the register and ask four questions. Which items did we band high and then do nothing about. Which owners have more than three open items, because that is a capacity problem rather than a diligence problem. Which watches produced nothing we acted on, and should they be switched off. And which of our known obligations had no watch pointing at it.
The fourth question is the one that finds real gaps. Everything else in the process is about handling what arrived. That question is about what never did.
The honest edge
Every source watched is a regulator or an official body publishing openly. A development that appears only in a subscription newsletter or a paywalled trade title will not appear, because paywalled sources are refused by design. If your sector's real early warning comes from a trade publication, this is not a complete substitute for it.
A materiality band suggested by a system is a starting point, not a decision, and dates come from the source or not at all: nothing is inferred. Coverage is UK and EU, so a team with US regulatory obligations needs another answer for those. And nothing here assigns an owner or accepts a deadline. Those remain acts by named people, which is precisely why they are the part that fails.
Read next
What credits cost, in plain English
