Seventy-eight percent are doing the math before they answer
Seventy-eight percent of frontline workers calculate costs such as gas, commuting and childcare before deciding whether to accept a shift. That comes from research Workday and DailyPay released on September 17, 2026, called The Frontline Worker Expectation Gap. Sit with the number for a minute, because it changes what a shift offer actually is. The notification your scheduling module fires off carries an economic proposition, and the system sending it has no idea what the answer costs the person receiving it.
The same study says 46 percent borrow money between pay periods and 69 percent check their pay after each shift. Those three findings describe one behaviour. Money gets tracked shift by shift by the people working the shifts, while most HCM configurations still reason in pay periods. The mismatch sits on our side of the screen.
The gaps land on settings your team already owns
Workday and DailyPay build the study around distance between what workers say matters and what they actually get. Schedule control is important to 72 percent while only 51 percent are satisfied with it. Pay clarity is a priority for 84 percent and 57 percent say they are very satisfied. Mobile tools are valued by 70 percent, with 53 percent satisfied. Every one of those gaps has a configuration decision underneath it.
Schedule control, translated into settings, means how far in advance schedules publish and how much of a swap a worker can complete without a manager's phone call. Pay clarity means whether someone can see the detail behind a single shift's pay on a phone, without a desktop login and a password reset they will never bother with. Mobile satisfaction depends less on the app than on which security group that worker sits in and what that group can actually see. We have written before about security groups that quietly multiply, and the same sprawl decides what a frontline worker gets to read at six in the morning.
None of this needs a new product. A schedule that publishes ten days out instead of three is a policy change and a configuration change, and it lets someone book childcare. That kind of fix rarely reaches a roadmap slide, because it has no licence attached.
Borrowing between pay periods is a pay timing signal
The 46 percent who borrow between pay periods is the figure DailyPay most wants you to read, and we would still take it seriously. Andrew Brandman, Chief Operating Officer at DailyPay, said it this way: "Employee expectations regarding their time, money and jobs have evolved. Yet, the systems supporting them haven't kept pace." He is selling on-demand pay when he says it. The observation underneath survives the sales motive.
For an HCM owner, the honest version of that finding is narrower and more useful. If half your hourly population is bridging the gap between the work and the money, your pay frequency has become a credit decision you never intended to make. Changing it is no small lever. It moves the payroll calendar, the posting schedule, deduction timing, and every reconciliation finance runs after a cycle. Anyone who has lived through a payroll cutover and the reconciliation that follows knows the cost of getting it wrong. So settle the cheaper question first. How many of your own hourly workers would say they borrow between pay periods?
Who paid for the research, and why the gaps still stand
This is vendor-sponsored research from a company whose business is paying people faster, published with a vendor whose business is the system of record. Read it with that in mind. The methodology is at least stated plainly. The Harris Poll ran the survey online in the US on behalf of DailyPay among 2,208 hourly-paid employed adults ages 18 to 45, between April 23 and May 2, 2026, accurate to within plus or minus 3.1 percentage points at 95 percent confidence.
The age ceiling is the part we would ask about. Frontline workforces we see are not capped at 45, and the workers above that line often carry the heaviest caregiving costs the study is measuring. The sample leaves them out. That does not void the gaps, it means the percentages describe a younger slice of your headcount than your roster holds. Sponsored research earns the same handling as a vendor demo run on your own terms. Accept the direction, re-measure the size yourself.
The AI split holds two different questions
Half of frontline workers, 50 percent, worry AI will replace their jobs. Among frontline workers whose employers already use AI, 64 percent report a positive impact. Those are two populations answering two questions, and the second turns on the employer's use of AI, not the worker's own. The study does not tell you whether exposure changed anyone's mind.
Josh Secrest, Vice President of Paradox at Workday, said: "Frontline workers want tools that make their daily lives easier, not more complex." Fair enough, and the test of that claim is whether an agent removes a step a worker performs today or adds a surface they now have to learn. A scheduling assistant that still requires a manager call to approve a swap has added a screen and removed nothing. The same discipline applies that governs what an agent is allowed to complete.
Pull four of your own numbers before the next scheduling review
Four measurements will tell you more than the survey does. Median lead time between schedule publish and shift start, by location. Percentage of shift offers declined on first send, and whether that rate rises with distance from the site. Share of your hourly population that has ever opened pay detail on a phone. Days between the end of a shift and money landing in an account.
If the decline rate climbs for sites further from where people live, you have measured the 78 percent inside your own tenant without commissioning anything. The follow-up costs nothing and stings a little. Ask a supervisor at your highest-turnover site what a worker has to do to find out why last week's pay came in lower, then time that path yourself on a phone. The exercise tends to surface the failures employees stopped bothering to report. What do you do with the answer if it turns out to be eleven minutes and two logins?
