Roster was founded in London in 2009, in a single room above a bookshop, by two people who had spent the previous decade at larger search firms and wanted to run searches differently. We spent our first two years mostly on contingent work. Our first retained shortlist went to a client in October 2011. This month marks fifteen years since then.
Anniversaries are a poor excuse for self-congratulation, so this is not that. It is an attempt to set down what the numbers say, and what we have learned from them.
The numbers
We have kept records of every search since the beginning, sometimes more carefully than others. As of the end of September 2026:
1,240 placements since 2009, across six sectors and two offices
Around 1,410 searches completed or closed, of which 88% ended in a placement
Just over 7,300 candidates presented on shortlists
94% of placed candidates still in post twelve months after starting
21 days median from signed brief to shortlist, down from 33 days in 2012
42% of placements were people who were not looking for a new role when we first called them
112 placements in 2025, against 14 in 2010
The New York office opened in 2016 and now accounts for just under 40% of our work.
What fifteen years taught us
Shortlists should be short
Our early shortlists ran to eight or nine names, because we assumed clients wanted choice. They wanted confidence. Since 2015 our median shortlist has been five, and the share of searches where the appointed candidate came from the first shortlist has risen from 71% to 86%.
The brief is most of the work
When we look back at searches that failed, very few failed because the market had nobody in it. Most failed because the brief changed in week five, or because two people on the client side wanted different things and had never said so. We now spend more time on the brief than we did in 2011, and less on everything after it.
Speed comes from preparation
The 21-day median is not the result of working faster. It comes from research that has already been done. Every consultant keeps a live map of their market, and most shortlists begin with people we have known for years.
The twelve-month figure matters more than the fee
We track whether people are still in post after a year because it is the only honest measure of whether a placement worked. When the figure dipped to 89% in 2017, we traced it to searches where we had let clients compress referencing. We stopped doing that. It has not been below 92% since.
People remember how you declined them
Of the 7,300 people on our shortlists, around 6,000 did not get the job. A good number have since become clients, or been placed by us on a later search. We make a point of telling unsuccessful candidates why, quickly and in person where we can. It is the part of the job that pays back most slowly and most reliably.
Who we are now
There are 31 of us across London and New York, and eight consultants lead searches. Four of them started here as researchers. The average tenure of a consultant at Roster is nine years, which in our industry is unusual and, we think, a large part of why clients come back. Just over 60% of last year’s searches were for clients who had hired through us before.
Our shortlists look different too. In 2014, 34% of our shortlists included at least two women. This year it was 71%. We did not get there through targets. We got there by widening where we looked, and by challenging specifications that described the last person in the role rather than the next one.
What comes next
We have no plans to open a third office or add a seventh sector. Growth, for us, has meant going deeper into the markets we already know. Over the next few years that means more interim work in energy and healthcare, a larger research team in New York, and holding the median time to shortlist where it is while the searches themselves get harder.
Mostly it means doing what we did in October 2011: understanding what a client actually needs, and finding the four or five people who can do it.
