01 · Source
Official 2026, 2027 and early-2028 dates
The bank holidays for England and Wales, Scotland, and Northern Ireland come from the GOV.UK bank-holiday calendar. The embedded dates needed by the 2026/27 checker, calendar-2027 planner and April-2027-to-March-2028 planner were verified on 29 August 2026.
Bank holidays are not automatically paid leave. GOV.UK notes that an employer does not have to provide paid leave on bank or public holidays. Planning tools treat them as non-working only under the stated model or chosen mode; the date checker merely compares them with selected weekdays. Your contract and rota are authoritative.
02 · Seasonal window
How the Christmas 2026 plan is bounded
The Christmas 2026 annual-leave planner searches only 19 December 2026 through 4 January 2027. It compares every continuous span in that window, counts selected normal workdays that would require leave, and returns the longest span that fits either the three-day or seven-day allowance. Equal spans prefer fewer proposed leave days.
National bank holidays become non-working dates only when the user keeps the explicit confirmed-off setting selected. Unticking it makes every selected normal workday consume leave even when GOV.UK lists a bank holiday on that date. This prevents a calendar listing from being presented as a promise of closure, pay or approval.
The archive state begins on 5 January 2027 in Europe/London and leaves the page as a historical 2026 example rather than silently retargeting it. Shared settings contain only the region, fixed weekdays, bounded leave allowance and bank-holiday mode after the #; the current date is recalculated locally.
03 · Date check
How the 2026/27 leave-year count stays honest
The 2026/27 leave-year bank-holiday checker uses an inclusive 1 April 2026 through 31 March 2027 window. The full regional total never loses a date after it passes: England and Wales has 10 listed national dates, Scotland 11 and Northern Ireland 12. A second list contains only dates on or after the visible Europe/London snapshot date.
A collision means only that a listed national bank holiday lands on one of the fixed normal weekdays selected. It does not decide closure, pay, deduction from allowance or entitlement. Scotland’s full count includes the one-off 15 June 2026 World Cup bank holiday; that national listing did not guarantee every employer or local body observed it as time off.
Shared comparisons put the region, selected weekdays and effective snapshot date after the #, so the remaining count remains reproducible without sending those inputs to the server. After 31 March 2027, the clean canonical route remains an explicitly ended historical window with zero remaining dates; it is not silently repurposed for a later leave year.
04 · Search
How fixed-week plans are selected
The planner creates valid date ranges between three and 28 days long. For every range it counts the normal working days that are not official bank holidays; those are the days you would need to request as leave.
It then uses a budgeted interval search to choose a set of breaks that do not overlap. Total days away is the primary score. When two plans produce the same total, the planner prefers the less fragmented one. All calculation happens in your browser.
Your shared URL contains only the region, leave budget, minimum break length, and selected weekdays. It does not contain a name, account, employer, destination, or booked dates outside the reproducible result.
The Northern Ireland 2027 planner uses this same tested search with Northern Ireland selected by default and a crawlable 17–29 March example around St Patrick’s Day, Good Friday and Easter Monday.
The Scotland 2027 planner uses Scotland’s nine nationwide bank holidays currently listed by GOV.UK and publishes a tested 25 December–4 January example. Council, school, NHS-board and employer-specific dates remain outside the model.
The fixed four-day-week planner uses the same search for repeating weekly patterns such as Monday–Thursday or Tuesday–Friday. It is separate from rotating 4-on-4-off and alternating 9-day-fortnight schedules, and it accepts only a whole-day balance already confirmed by the user’s employer.
05 · Leave years
How the April 2027–March 2028 window is bounded
The April–March planner uses the same fixed-workweek interval search but charges leave only on selected working days from 1 April 2027 through 31 March 2028. Natural non-working days may extend a suggested break across either boundary; working days outside the window can never consume this leave-year balance.
The candidate scan includes a full 27-day margin on each side so custom workweeks are not truncated near the boundary. The regional holiday inputs include only national dates currently published by GOV.UK. They exclude local public holidays, employer shutdowns and any future one-off date not yet listed.
06 · Scenarios
How part-time bank-holiday collisions are calculated
The part-time bank-holiday calculator compares the official 2027 dates with the normal weekdays you select. A date is a collision only when that bank holiday lands on a normal working day. The calculator then applies the employer-policy scenario you choose to the confirmed allowance you enter; it does not infer a contract or decide a legal entitlement.
The statutory-baseline helper is limited to regular-hours workers employed for the full year. It multiplies normal days or hours per week by 5.6 and preserves fractional days or hours. It must not be used for irregular-hours, part-year, rotating-rota, joiner or leaver calculations. Unequal shifts stay in hours and are never rounded into the whole-day leave optimiser.
All calculation happens in the browser. Tracking receives only a bounded event name and page path, never the weekdays, hours, policy, allowance or result entered. If the user explicitly chooses Share or Copy, the displayed settings are encoded after the # in the reproducible link; fragments are not sent with the page request.
07 · Rotas
How rotating-rota plans are selected
The 4-on-4-off planner models an eight-day cycle of four rostered work dates followed by four rest dates. That same binary cycle covers a repeating 2-days, 2-nights, 4-off pattern when the anchor is the first day shift; the dedicated guide maps those positions as D, D, N, N, O, O, O, O. The interactive calendar keeps the four work positions generic because other four-on-four-off rotas use different day/night orders. It assigns an overnight shift to its start date, so an approved proposal's spans count dates without a shift start to work—not complete midnight-to-midnight days away. Users whose employer uses a different date convention must translate the proposal against their live rota. The 6-on-4-off planner uses the same tested engine with a ten-day cycle of six work dates followed by four rest dates. In both cases the anchor must be cycle day one of a known work block, and the selected pattern is projected forwards and backwards from it.
The planner evaluates possible breaks, counts only rostered working shifts as leave cost, then uses a budgeted non-overlapping interval search to maximise the covered date span. On the 4-on-4-off route that score is explicitly conditional on the proposed leave being approved and is subject to the overnight convention above. Bank holidays follow the rota by default. They are treated as extra rest dates only when you explicitly select that mode because your employer has confirmed it.
A shared rota URL contains only the bounded pattern key, cycle anchor, leave-shift budget, minimum break, UK region, bank-holiday mode, and channel attribution. It contains no account, employer, destination, or free-form personal data.
08 · Fortnights
How 9-day-fortnight plans are selected
The 9-day-fortnight planner models a Monday-to-Friday compressed schedule with one fixed weekday rostered off every 14 days. The anchor must be a date known to be that rostered day off; the alternating pattern is projected forwards and backwards from it.
The planner counts only scheduled working days as leave cost and uses the same non-overlapping interval search to maximise total calendar days away. Equal compressed days may be budgeted in days or hours; an hours balance is converted only into complete leave days for planning and any fractional remainder stays visible.
Bank holidays are treated as non-working only when that mode matches the employer-confirmed arrangement. Shared URLs contain only the anchor, unit, balance, equal day length, minimum break, region, bank-holiday mode and bounded channel attribution.
09 · Boundaries
What it does not know
- Your employer’s shutdown days, rota, booking limits, notice period, or minimum staffing policy.
- Local public holidays that are not present in the UK-wide GOV.UK feed.
- Whether a bank holiday is part of, or additional to, your annual allowance.
- Your personal preference for one long trip versus several shorter breaks beyond the minimum length you select.
Use the result as a shortlist. Verify every date before booking leave, travel, or accommodation.
10 · Measurement
Aggregate traffic only
The public growth goal counts successful HTML navigation requests from browser-like clients. Static files, API calls, prefetches, the public stats page, obvious crawlers, link-preview agents, command-line clients, and automated test tools are excluded.
The database stores only daily aggregate counters by event, A/B variant, coarse source category, and page path. It does not store IP addresses, full referrers, user-agent strings, cookies, fingerprints, account identifiers, free-form input, or individual event rows.
The first-party count is available through the public stats API. Cloudflare Web Analytics is the independent privacy-first validation source. A completed week validates only when the Web Analytics Page views report uses Standard/UTC time, the same Monday 00:00 to Monday 00:00 window, and filters Exclude bots = Yes, host leave-more-2027.pages.dev, and the exact eligible fixed-week, Christmas 2026, fixed four-day-week, Scotland, Northern Ireland, April–March, 2026/27 checker, methodology, rotating-rota, part-time-calculator and 9-day-fortnight paths. It must also show at least 5,000 views. If Cloudflare samples the query, its 95% lower confidence bound must remain at or above 5,000. The two central counts must be within 20% of each other; a larger gap requires investigation and cannot complete the goal.