Why we will not publish a time-to-fill benchmark

There is no shared convention for when the clock starts. Some organisations start it when headcount is approved, some when the requisition is created in the system, some when the brief is complete, some when the job is first advertised. Between headcount approval and a complete brief there is frequently a fortnight of internal discussion, and whether that fortnight is in or out changes the published figure enormously.

The end point varies too. Offer accepted, contract signed, or first day worked are three different dates, often separated by weeks for senior roles and by months where notice periods are long.

So when two sources report different averages, most of the gap is definitional rather than operational. You are not looking at one industry hiring faster than another; you are looking at two different measurements wearing the same label.

Time to fill can only be computed for roles that were filled. Requisitions that were cancelled, withdrawn, put on hold indefinitely or quietly abandoned after four months contribute nothing to the average, because they never produced an end date.

This matters because the abandoned requisitions are systematically the hard ones. The roles that take longest are the ones most likely to be pulled, which means every time-to-fill average is computed on a sample from which the worst cases have been removed. The published number is not just imprecise; it is optimistic by construction.

The same effect operates inside a single firm, which is why your own average can improve in a quarter where you actually performed worse. If three difficult searches were cancelled, your remaining sample is easier and your average falls.

A number that mixes a warehouse operative and a chief financial officer is an average of two unrelated processes. Seniority, scarcity, geography, security clearance, notice-period norms and whether the role is contract or permanent each move the figure by more than any process improvement will.

Sector benchmarks are usually built by aggregating whatever respondents submitted, which means the mix is unknown and the comparison to your mix is unknowable. A firm that specialises in senior regulated-sector hiring will read the same benchmark as an indictment when it is simply a description of a different portfolio.

The corollary is that even your own internal figure should be segmented before it is compared. One number across a whole agency is a management artefact, not a measurement.

Most widely quoted figures come from voluntary surveys or from aggregated data across one vendor's customer base. Both have known problems. Voluntary surveys are answered by organisations that measure this at all, which is a self-selecting group, and the answers are frequently estimates typed by somebody who was not the person doing the measuring.

Vendor-aggregated data is more consistent internally but describes that vendor's customers, who share a size range, a sector skew and a way of working. It is also computed from whatever their system considers a stage transition, which is a product decision rather than an industry standard.

Neither source is dishonest. They are simply not measuring a population you belong to, using a definition you share, over a period comparable to yours. That is three independent reasons the number will not transfer.

Take your last thirty to fifty placements, or twelve months if you place fewer than that. For each one, record the requisition open date, first submittal date, first interview date, offer issued date, offer accepted date and start date. If those dates are not in your system, they are recoverable from email threads for a sample this size, and it is worth the afternoon.

Compute the intervals rather than the total, and segment by role family, seniority and placement type. Report the median rather than the mean, because a single eleven-month search will drag a mean into meaninglessness while the median tells you what a typical search looks like.

Then include the searches that failed. Count cancelled and abandoned requisitions separately as a fill rate, and report it next to the timing. A desk that fills fast because it walks away from difficult roles should be visible as exactly that.

Use it for three things. First, thresholds: an alert on a stalled submittal is only useful if it fires at a point that is unusual for that client, and your baseline supplies that point. Second, briefing: telling a client that roles of this shape typically take a given number of days on your own desk is a far stronger conversation than quoting an industry figure they can dispute.

Third, improvement. Because the baseline is segmented by interval, a change is attributable. If your submittal-to-response interval halved after you started presenting batched shortlists, that is a finding about your process rather than a fluctuation in the market.

What the baseline should not be used for is comparison against another firm. Their definitions differ, their mix differs, and the exercise generates argument rather than insight.

Surhires has no customers yet. That means we hold no aggregated hiring data, have run no survey, and have no basis on which to publish a benchmark. We could produce a figure by citing somebody else's, but repeating a number whose definition and sample we cannot inspect would make this page worse rather than better.

When there is a defensible basis for publishing timing data, it will be published with the definition used, the sample size, the segmentation and the date. Until then the absence is the accurate position.

If you need a figure today, the calculator page will compute the intervals from dates you already hold, which will tell you something true about your own desk rather than something unverifiable about an industry average.