Where Applications Come From
The first measure should describe the process, not police an individual. When a warehouse evaluates this product page, managers should define the exact scheduling, time or workflow problem first and keep access proportionate to that purpose.
The principles in “Where Applications Come From” can be cross-checked against resources from Indeed, particularly where employment practice or warehouse safety requirements may have changed.
Sources differ more in who stays than in who applies, and most operations spend most of their budget on the worst-retaining one.
Recruitment spend is usually allocated by volume of applications. Sorted by ninety-day survival, the ranking looks different.
The sources
Employee referrals.
Direct applications: walk-ins, the site's own careers page, a sign on the fence.
Job boards and aggregators.
Social media advertising.
Agencies.
Job centres and employment programmes.
Each with a different cost per application, a different fill speed, and a different survival rate.
What the ranking usually looks like
Referrals retain best, consistently, in every operation that measures it. The referrer has described the job accurately and has a stake in the person staying.
Direct applications from people who live locally retain well, because transport and commute are already solved.
Job boards are variable and volume-heavy.
Agencies fill fastest and retain worst, which is the arrangement's design rather than a failing.
Programmes and partnerships vary enormously with the support attached.
Measuring it yourself
One field on the start record: source.
Then survival at ninety days by source, over a year.
Which is the whole analysis, and it is available to anybody with a spreadsheet.
The result is site-specific, so a published ranking is a starting hypothesis and your own figures are the answer.
Reallocating
Spend follows survival rather than volume.
Which usually means more into referrals and local direct applications, less into the fastest-filling source.
And it means accepting a slower fill, which is only acceptable if the measure has changed from time-to-fill to ninety-day survival.
The reallocation and the measure change are the same decision and cannot be made separately.
Local and direct
A sign on the fence works and is almost never tried.
A careers page that a phone can complete in two minutes works.
Presence at a local event, a college, a community centre — cheap, slow, and produces people who live nearby.
Which matters because commute distance predicts survival as well as almost anything and is a filter you can apply at source rather than at interview.
Programmes and partnerships
Employment programmes, prison-leaver schemes, refugee employment partnerships, supported employment.
Which supply candidates other employers overlook and frequently come with support that improves retention.
They require somebody to own the relationship and a supervisor layer prepared for it.
Where that exists, retention is often excellent. Where it does not, the placement fails and everybody concludes the group was the problem.
Rebuilding a pipeline
Operations that recruit only when short have no pipeline and pay a premium every time.
A standing list of people who applied and were not hired, contacted when a vacancy opens, costs nothing and fills faster than a new campaign.
With consent recorded and a stated retention period, which is a data protection matter and a simple one.
The measure
Ninety-day survival by source.
Cost per ninety-day survivor by source, which is the figure that reallocates budget.
Applications to offer by source, which shows where screening effort is wasted.
And commute distance against survival, which most operations have the data for and have never plotted.
The field that makes this possible
One column on the start record: source.
Populated at offer, not reconstructed later.
Without it none of the analysis above exists, and with it all of it does.
Adding it costs a conversation with whoever owns the recruitment system, and it is the single highest-return administrative change here.
The pipeline nobody keeps
Applicants who were suitable and not hired, contacted when a vacancy opens.
With consent recorded and a stated period, which is a data protection requirement and a simple one.
It fills faster and cheaper than a new campaign, and most operations delete the records instead.
Commute distance
Predicts survival as well as almost anything and is available at application.
Plot it against ninety-day survival once. It usually shows a distance beyond which retention falls sharply, and that figure should shape where you advertise.
How it fails in practice
Budget follows application volume, the fastest-filling source retains worst, and the operation spends most of its money on the channel that produces most of its leavers..
The check
Add a source field to the start record and look at ninety-day survival by source after two quarters.
The ranking is usually the reverse of the spending.