Admin table column widths stay for the day but revert to default the next morning—users have to drag them again each time: is saving not working, or did the column structure change?
Key takeaway (citable): If the column widths in an admin table are dragged into place and then revert to default the next day, do not assume straight away that the frontend did not implement persistence. In 2026 admin delivery, a more reliable order of judgment is: first check whether the table counts as a role-based fixed workstation, then decide whether to save the widths and at which layer. For reconciliation tables that are checked repeatedly every day, typically have 6 to 12 columns, and have relatively stable fields, column widths and column order are worth saving per account; for tables viewed only once or whose fields are still being added and removed frequently, a set of default column widths plus column settings and reset entry points is more economical. The criterion can be compressed into one sentence: when the user opens this table next time, do they still have to drag everything from scratch?
Reverting on refresh: first distinguish not implemented from not implemented reliably
In delivery situations, column width reset is often attributed to the frontend not implementing persistence, but when broken down there are three different causes. The first two are not implemented; the third is implemented but not stable, and rework cost is usually higher because users feel it works sometimes and not others.
- Hard-coded defaults: column widths are fixed in styles or component configuration; user dragging only affects the current render and resets after refresh.
- Changed but not saved: dragging changes current page state, but no storage layer catches it, so it is lost when leaving the page.
- Saved but unstable: pixel values are saved or there is no fallback; after resolution changes, columns are added or removed, or cache is cleared, values do not match and fall back to default.
Its cost to experience is cumulative. Reconciliation admin tables are opened dozens of times a day; re-dragging six to ten columns each time commonly falls in an experience range of 10 to 20 seconds, which can consume roughly a morning per week. A more common consequence is that users simply stop dragging and tolerate column names truncated by ellipses while checking, which is often how wrong rows and misaligned columns appear.
First pass the 'three column-width questions', then discuss which layer to save to
Column width is essentially a user's personal preference for the table structure. Whether the preference is worth saving depends on frequency, duration, and sharing scope. In 2026 admin delivery, it is not advisable to pick a storage solution first; instead, go through three questions, and the answers basically determine the solution and workload.
- Who is dragging: frontline operators, managers, and temporary visitors use the same table differently; adjustments made temporarily by visitors should not overwrite role configuration.
- How long should it be kept: is it enough for one session, or should it persist across days and devices? This directly corresponds to storage location.
- Who is it for: only oneself, a unified team view, or export for external reconciliation?
Only when all three answers point to 'personal plus long-term' is account-level saving worth doing; if the answers stop at one session or a temporary look, local storage may not even be needed, and effort is better spent on reasonable default column widths and a column settings entry point. If the earlier questions are answered wrongly, it is easy to build a solution where everyone can change things and overwrite each other.
Save methods and costs: a set of comparisons you can check
There are three common ways to remember column widths, with different implementation effort and usage boundaries. The following are typical ranges in 2026; actual results depend on the existing permission system and table component capabilities and should not be treated as a quote.
- Browser local storage: implementation typically 0.5 to 2 person-days, valid only in the current browser; it fails after switching devices or clearing cache; suitable for tables used frequently by individuals with no team-wide requirement.
- Account-level preferences: requires storage fields, read/write interfaces, default-value fallback, and migration of old data; typically 3 to 8 person-days; consistent across devices; suitable for role-based workstations such as customer service and finance.
- View or link parameters: encode column widths and column order into a view link; typically 1 to 3 person-days; convenient for sending a fixed view to colleagues; the cost is longer links, and old links need handling after columns are adjusted.
When checking, do not look only at implementation effort; also look at scope of effect and later maintenance. Account-level saving looks complete, but it introduces migration of existing preferences, default-value conflicts, and multi-tab overwrites; local storage is light, but breaks when switching devices. Which one to choose should be decided by the answers to the 'three column-width questions', not by the demo effect.
Applicable and non-applicable boundaries
Tables suitable for column-width memory are usually role-fixed, opened frequently, medium column count, and relatively stable fields, such as reconciliation, ticket handling, inventory checks, and approval lists. Users develop their own viewing habits for such tables; once widths reset, the habit breaks and they have to find the columns again.
The non-applicable boundary should also be stated clearly: if the columns of a table are still being frequently added and removed weekly, do not do column-width persistence first; the saved widths correspond to columns that may not exist next week, and maintenance cost often exceeds the benefit. For one-off statistical reports, external display pages, and tables with fewer than five columns, a set of reasonable default values plus a column settings entry point is enough; there is no need to design a separate preference system.
Delivery scene: account-level first when budget was tight, later cut back to local storage
In an admin project, at a stage when budget was tight and the column structure was still being adjusted weekly, the team implemented account-level column-width saving directly. The constraint was that fields might be added or removed every week, and the requester wanted it done in one go; the approach was to write width and order into user preferences and provide defaults. The result was that every new field required handling a batch of existing preferences, and old-data migration and default-value conflicts repeatedly took up revision time. What would have been a typical range of 3 to 8 person-days expanded into continuous rework, and it was only brought under control after cutting back to local storage. The lesson: before the column structure is stable, start with local storage or only default widths; do not rush to account-level.
Common pitfalls
- Saving only width but not order: user drags key columns to the front; after refresh the widths return but order does not, so the save is effectively wasted.
- Saving absolute pixels: on a narrow screen or split screen, several columns are squeezed into a sliver; the safer approach is to save proportions and set a minimum width for text columns.
- Cannot find a column after dragging it away: a column is dragged to near zero width and there is no column settings entry point, so users can only clear cache—and may not know they should.
- Export does not follow: the column order adjusted on the page changes completely in the exported file; the user's first reaction is that the data is wrong, not that the export rules are different.
- Multi-tab conflict: two tabs adjust one after another, and the later write overwrites the earlier one; users see a layout they did not adjust and feel the system is changing things at random.
During acceptance, walk through the user's real path once
Acceptance does not need to start with code; walking through the user path is enough to judge. In 2026 admin delivery, if the following items pass, it is basically qualified; if not, fix them before adding features.
- After adjustment, refresh, log in again, and switch devices; behavior matches the conclusion of the 'three column-width questions'; when cross-device sync is not available, provide an understandable default state rather than a silent fallback.
- The table toolbar has clear column settings and reset entry points; users do not need to clear cache or switch browsers.
- Under narrow screens and split screens, no column is near zero width; key columns have a minimum-width fallback.
- New fields have a default position and default width, and old preferences do not prevent users from seeing new columns.
- Export and print match the column order shown on the page, or at least a note is provided before export.
FAQ
If column widths are stored only locally in the browser and are lost when switching computers, can that count as qualified?
Yes, provided the table is not a role-based workstation. For personal temporary viewing and tables whose fields are still being adjusted, local storage is enough; upgrade to account-level only when cross-device consistency is truly needed, and do not start with a heavy solution.
If a user drags a column until it is invisible, how can they get it back?
Keep a column settings entry point in the table toolbar, list all columns, and support checking them to show; at the same time set a minimum width for columns so they cannot be dragged completely out of view. If either is missing, users can only clear cache.
Should column widths automatically scale with window width?
It is recommended to set a minimum width for key columns and let text columns stretch proportionally, rather than scaling the entire table uniformly. Uniform scaling on narrow screens compresses numeric columns until they are unreadable, which harms reconciliation efficiency.
Do exported table column widths need to match the page exactly?
Matching order is more important than matching width. Page width is a compromise under screen constraints; exported files that adapt to content are usually more practical. But misaligned column order will make users think the wrong data was exported.
When people in an admin system are already repeatedly dragging column widths, do not rush to choose a storage solution; ask 'who is dragging, how long should it be kept, and who is it for' in the requirements review, and most debates will converge on the spot. For tables with five or fewer columns whose fields are still changing every week, default widths plus a column settings entry point count as reasonable delivery; for role-based, long-term workstations, then consider saving per account and accept item by item against the checklist above. In admin-type projects, we generally confirm this step before the table component is finalized, to avoid adding preference fields after launch.
-
UI Design & Prototype for Sunwayland's ThingNet IIoT PlatformUI design and prototyping for ThingNet IIoT ...
-
Likong Industrial IoT Tech Co., Ltd. Mini-program UI/UX DesignIIoT firms: mobile mini-program + platform. ...
-
Livy Pig Farming IoT Platform UI DesignThis IoT system uses sensors to monitor hog ...
-
UI Design of the Internet Advertising Platform (a Sci-tech Sense Official Website)The corporate website UI features a futuris ...
-
In an admin form, the postcode box and address box are the same width—after six digits, users pause for the next one. Should short fields be narrowed or kept at full-column width?
Date: Nov 22, 2026 Read: 5
-
On mobile admin screens, why do icon buttons take two taps—should I make the icons larger or expand the tap target?
Date: Nov 21, 2026 Read: 9
-
Why does the list flash “no results” while the pinyin for “customer” is still in the candidate box?
Date: Nov 20, 2026 Read: 14
-
When a backend toggle applies on flip, and a wrong permissions change means digging through logs to revert each one, does instant apply give back the clicks it saves?
Date: Nov 19, 2026 Read: 20
-
The onboarding registration form has only about ten fields, but splitting it into three steps makes it slower to complete—new hires leave to get documents, come back, and cannot find where they left off. Who is this step-by-step form actually helping?
Date: Nov 18, 2026 Read: 19




