Compensation gets the conversation started. It rarely finishes it. Associates in London and in the US already know the band. In New York the scale is public. In London the US firms publish, and within a City year group someone has usually compared notes before a recruiter is instructed. A firm that matches the market has cleared the first question. After that, the process is about the seat.

The surveys point the same way, for what they are worth. Chambers, reporting in 2026 on associates at 82 US firms, found culture given as a reason for joining about twice as often as pay. Law360 Pulse found associates more bothered by billable expectations, advancement and training than by the figure on the offer. Pay still matters. Past a certain point it stops separating anyone.

Work is what we get asked about most. Whether a mid-level is running a matter, or a real part of one, with a partner who will stand behind the draft. US firms in London are often picked for that responsibility coming earlier. City firms are often picked for the clients and the training around them. Houston, and the energy groups more generally, come down to something plainer: is there a transaction or a regulatory file with your name near it. That is also how you stay employable if you move again.

Partnership is harder to read, in both markets, and associates have got more direct about it. How many of the year made counsel. Who would sponsor them. Whether the people the firm trained are promoted, alongside the laterals. Chambers put the sharpest flight risk at two to four years. The firms that keep people are usually the ones where a partner will give a specific answer, rather than a speech about merit. It is worth asking for the last two or three promotions in the group, and what those associates had actually been doing. A vague answer is not a bad firm. It does tell you the path is not something the group has had to explain recently.

Culture, in these conversations, is whether a question gets answered and whether the team you meet is the team you join. Who last joined, and how they have found it. What a busy month looks like. Firms that can say are at an advantage. Plenty of good ones can.

A few worries come up often enough to be worth writing down.

Associates are warier about AI than partners, which is what Law360 found. The firm response has moved past a single town hall. Quite a few now treat fluency as part of the job, and some count the learning towards hours. Ropes & Gray lets first-years put nearly a fifth of the billable target into AI training and simulations. Others credit a smaller block, often around 100 hours, for courses and supervised work. Some expect you to learn it and do not credit the time. Worth asking what you still draft, what you review, and who checks it. The NALP Foundation has started asking whether AI training came up in exits. Fair question in a process. It is not evidence that firms have left it alone.

On security, the same NALP Foundation set put average associate attrition at 19 per cent in 2025, and most of those who left had been at the firm under five years. That is a market number. The associates who sound steadier are usually in a practice that is busy for a reason they can name, with experience another firm or a company can use. A niche helps. A general seat with no sector is a harder story, in a process and in a quieter year.

Hours have slipped for a lot of people. A third of the Law360 respondents said the balance had worsened. A light-hours job on the top US scale is not on offer, and nobody serious thinks it is. What some teams can offer, in London and in US offices outside New York, is a bit more predictability. Hybrid working is settled. It only marks a team out where they actually work that way.

In-house or private practice is a real choice, and neither is the grown-up one. In-house with a specialism, at the right year, is a proper option. US departments are more likely to have a counsel track and someone in legal operations beside it. London in-house is often a wider seat. Pay usually steps down. Hours often do too, until a deal or a dispute, when they do not. Private practice is still the better training if the work is real. The people who choose well have already decided which of those they want.