Users keep paging back and forth between page 2 and 3 to find last week's inbound receipt in the admin list—should we increase rows per page or surface order number search first?
How many rows an admin list shows per page is not a question of first picking 10, 20, or 50; first distinguish whether users are paging to find a single record or scanning row by row within the page.Based on the typical range in 2026 admin deliveries, when there are hundreds to a few thousand records and most users come with an order number or customer name, a default of around 20 rows is more stable; if the list has no order number search, search-term highlighting, or usable sorting, simply increasing the default row count often just trades the fatigue of paging for the fatigue of scrolling.
Increasing rows per page solves the number of page turns, not in-page locating
When users say paging is tiring, they usually mix two things: there are many pages, and the target row is hard to see within the page. Increasing the row count only alleviates the former. If the target row has no visual anchor—for example, the search term is not highlighted, and status and type look similar—the longer the page, the more likely users are to scan back and forth across a large block of rows.
This is also why changing only the default value has limited effect. Rows per page is often a configuration that takes effect across multiple lists when changed in one place, but it shifts the burden to scroll distance, rendering time, and reading load. A common feedback in delivery is that after changing 10 to 50, users still say the record is hard to find; when you dig deeper, the problem is usually that the list lacks order number search or customer name search, and the row count is not the main cause.
- Total records within dozens to around two hundred: if they fit on one page, pagination can be omitted, or the default can be around 50 rows, with the pager still showing the total.
- Hundreds to a few thousand records, mainly for finding a single record: default to around 20 rows, and have the pager show the current range and total.
- Tens of thousands of records, with users coming to look up an order number: row count is not the focus; order number search should be expanded by default and support fuzzy matching.
- Reconciliation or batch write-off is needed: rather than increasing the row count, provide export so users can compare in spreadsheet software.
Set the default row count first based on record scale and how users find records
Row count is a balance between the number of page turns and the reading load on a single page. The judgment order is not to pick a number first, but to estimate the burden at both ends. Break it into three dimensions and go through them in order, and you can usually arrive at a deliverable default.
- Record scale and growth rate: first look at the data volume at launch and what it may reach in six months. Data volume determines whether one page can hold everything or pagination is required; growth rate determines whether the default will soon need changing again.
- Whether users mainly need to find one record or see the overall picture: users finding one record will rely on search and filters, so around 20 rows is enough; users reviewing the overall situation need more information per page, and 40–50 rows is more convenient.
- Whether cross-page comparison is needed: if users are used to viewing data from two pages together, pagination does not match the task; a more stable approach is to provide export or custom columns rather than continuing to add rows.
The first dimension easily misses growth rate. During delivery, first check whether the backend pagination API supports a variable row count; some legacy systems hard-code rows per page on the server, so even if the frontend changes the selector, it will not take effect. In the experience range, this kind of rework often takes one to two rounds of revisions.
There is a type of constraint in delivery settings: historical data is around two to three thousand records, rows per page is hard-coded on the server, and most users come to look up an order number. The approach is not to directly change the default from 10 to 50, but first to expand order number search by default and add the total and current range, then make the pagination row-count parameter something the frontend can pass; the result is fewer paging complaints, at the cost of the filter area taking one more row of height and needing to be collapsed on mobile, with integration and regression testing falling in the typical range of one to two rounds of revisions. This order takes more effort than changing the row count alone, but there is less back-and-forth later.
The pager must clearly state the total, current range, and selection state
Once the row count is set, the pager also needs to answer which records I am seeing from which to which, and how many there are in total. Users judge whether to continue paging based on this range, not the page number. According to common acceptance criteria in 2026 admin deliveries, the following are basic items.
- Show the total and current range: if you only provide previous and next page, users cannot estimate how much is left.
- Keep the selection after switching row count: if a user once selected 50 rows, it should remain when they enter the same list again; jumping back to 10 rows makes people wonder whether they remembered incorrectly.
- Return to the first page after switching row count: if you stay on page seven and then switch row count, page numbers and records easily become misaligned as a whole.
- Do not force too much content on narrow screens: mobile usually does not provide row-count switching; lower the default to 10–20 rows and pair it with cards or simplified columns.
One counterexample is also worth noting: some admin systems make row-count options into several levels—10, 20, 50, 100, 200—which looks flexible, but in practice at the 200-row level, scrolling easily stutters on old devices and low-bandwidth environments, and users think the page is frozen. Two or three levels are enough; for needs beyond the upper limit, export is a more stable way to handle them.
How to weigh the three trade-offs: adjusting row count, adding search, and remembering selections
For the same problem of not finding a record, the three paths have different costs and benefits; before choosing, first see whether the bottleneck is paging or locating. Below is a comparison by cost, timeline, and suitable users; the numbers are all experience ranges and should be checked against the specific project schedule.
- Option A: Only increase the default row count—small change, usually can go live within one to two days; suitable when records are few and users are merely annoyed by paging; the cost is a longer scroll for long lists, and systems with poor rendering performance may become laggy.
- Option B: Add search and key filters—involves APIs and indexes, typically taking one to three weeks; suitable when records reach the thousands and users come with clear conditions; the cost is that with many filter conditions, the interface becomes complex again, requiring clear and display of applied conditions.
- Option C: Default row count plus remembering user selection—adds storage on top of A, with about two to three days of change; suitable for role-based admin systems where the same group of users repeatedly uses the same list; the cost is that you must decide whether to save the selection locally or on the account, and inconsistent behavior after switching devices can raise questions.
According to enterprise project delivery habits, the benefit of Option B is usually greater than Option A; but when there are only a few hundred records, Option B is overinvestment, and it is more cost-effective to first get the default value and one sortable column right. The three do not conflict; a common combination is to use Option B as the foundation, Option A to fine-tune the default, and Option C to accommodate high-frequency users.
Which lists do not need changes to pagination row count
Pagination row count is not an option that must be adjusted in every scenario; in one type of scenario, changing it only causes trouble. The criterion is simple: if users never do the task of finding one specific record in this list, row count is not their pain point.
- There are already few records: within dozens, showing all on one page is easier than any pagination.
- Mainly for export and download: users process the file locally after receiving it, so row count in the interface is irrelevant.
- Real-time refreshing monitoring lists: after increasing row count, the page reflows frequently, and visual jumping is more annoying than paging.
- Mobile on-site work: screen height is limited; no matter how many rows there are, it only lengthens scroll distance, so it is better to put key fields first.
It is also not recommended to make showing all records on one page a formal option. Once records exceed one thousand, a single query and frontend rendering may time out, and users will see a blank page or a long spinner, which they may report as a system failure. If there really is such a need, handling it with export is more stable. This judgment applies to record-oriented admin lists; pure display pages, mobile on-site pages, export-oriented lists, and real-time monitoring lists are not on the same judgment line.
Frequently asked questions
If users choose rows per page themselves, should the last choice be remembered?
For role-based admin systems, it is recommended to remember it, because the same group of people usually enters the same list repeatedly; for pages aimed at external customers and used infrequently, it can be left unremembered to avoid inconsistent behavior after switching devices.
Will a default of 20 rows make the API slower?
Twenty rows per page is generally not the bottleneck; slowness usually comes from missing indexes on sort fields or too many joins. Check the query plan first, then consider reducing the row count.
If there are only dozens of records, is pagination still needed?
No. Showing dozens of records on one page is easier; as long as the total is marked in the table header, users know they are seeing everything and do not have to guess whether there is a next page.
If users ask to show all records on one page, should we agree?
It is not recommended to make it a formal option. After records exceed one thousand, a single query and rendering easily time out; handling it with export is more stable, while the interface still paginates by a fixed row count.
-
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 ...
-
When users scroll dozens of screens and want to find the item they just saw, should they rely on page numbers or a back-to-top button?
Date: Sep 15, 2026 Read: 84
-
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: 4
-
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: 8
-
Why does the list flash “no results” while the pinyin for “customer” is still in the candidate box?
Date: Nov 20, 2026 Read: 13
-
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: 19




