[IPAC-List] Does time to staff REALLY matter, and if so, when?

[Mark Hammer]

I'm drowning in data here.  Sadly, it arrived too late to present in Nashville.  So here's a taste.

One of the things our organization and many others are besotted by is time-to-staff.  The assumption is that a faster turnaround time gets you better people.  In truth, I don't think anyone has ever put that hypothesis to the test, or if they have, they have never really done a regression analysis to examine the point at which faster turnaround time makes a difference.  If an organization typically takes 12 weeks to fill a position from the moment the ad goes up until the offer is made, but you have the same likelihood of missing out on your first choice at all points after 3.5 weeks, then why bang your head against a wall if you'll never realistically be able to reduce average time-to-staff below, say, 8 weeks?

So, I started looking at the estimated time between when the poster went up and when the offer of a job was extended.  We asked people if they were applying in seriousness or just doing so to get experience or practice, and I weeded out those who were not specifically pursuing *that* position.  Out of over 30,000 people answering the survey across government (only some of whom were applying to a competition in seriousness, and made it all the way through to the end), I ended up with 81 who had applied in seriousness but ultimately turned down the offer.  I compared their estimate of time between poster going up and the offer, to the average time of the nearly 2400 who accepted the offer of position, and there was no reliable difference.  Partly, I suspect, due to issues of statistical power, and partly because candidate estimates of time taken were highly suspect (managers estimated 21.8 weeks for the same interval that candidates were saying took over 35 weeks).

Even though we asked candidates if they were just dickering around or serious, those who said they were serious also had other irons in the fire. The folks who were turning down offers were more likely to have 1-2 other applications they were waiting to hear about at the same time, though only about 1/3 said they turned the offer down in favour of another.  So, people were turning down jobs for reasons other than it taking too long.

We looked at it from the other angle, and asked managers if their first choice had turned down their offer.  No difference in elapsed time between those managers who scored their first choice, and those who heard "Thanks but no thanks".

Do slower processes result in positions going unfilled?  Hard to say.  We asked managers if they still had unfilled positions at the end of the process, and compared elapsed time to staff for those where all positions were filled and those where positions went unfilled.  The completely-filled ones were clearly faster, but then they usually involved having to pick through a much smaller applicant pool to fill a smaller number of positions.

I've looked at it every which way the data permits (even looked at perceived quality of hire), but I see no unambiguous evidence that a faster process scores you more of what you want, whether in terms of quality or simply filling positions.

Perhaps this is a non-phenomenon, or perhaps it is simply a phenomenon restricted to a specific range of turnaround times, or perhaps to specific types of positions in specific labour/job markets.  So, maybe there is a huge difference in likelihood of snagging the best hire for job X, when comparing an offer made at 3wks post-application vs 3 days, but for job it simply doesn't matter because there is no shortage of suitably qualified people who are willing to wait as long as it takes.

In other words, the possibility that the faster-gets-you-more might be true does not preclude what I'm seeing in our data, nor does what I'm seeing preclude the faster-gets-you-more hypothesis being true....sometimes.  It really seems to be more of a case of identifying the context where faster provides advantage and where faster provides no advantage.  This may be the sort of thing I will not be able to sort out with our data until we've acquired several cycles' worth of cases, and I have hundreds and hundreds of cases where people turned an offer, and enough cases for different categories of job to see where it matters.

So....how fast do things need to be to make a difference?  And equally important, how would you know that speed had made a difference in outcome?

Mark hammer
Ottawa