Encouraging users to 'group' very large data sets

Question

We have a product that requires users to import large data sets (100+ rows) and then 'group' those datasets into buckets within the product. This is part of the set-up process so that they can start getting value out of our analytics tool.

The first issue we want to address is that users are accidentally creating duplicate buckets with small errors (e.g. spelling or capitalisation inconsistencies). We're exploring a solution that forces users to pre-define the buckets, and then when they're in the bucketing process, only allowing users to select from the pre-defined list of options.

The second issue is that this is inherently a very time-consuming process, and users are not motivated enough to complete the process accurately. I'm wondering what an effective UX solution might be for making this process as quick and easy as possible.

Currently users are given a free text field for each row with which to assign the data to an existing bucket or create a new bucket. They can bulk add rows to the same bucket by selecting each row, or searching for a keyword and then selecting all rows that contain that keyword. This solution works OK but it feels like there might be a faster and more intuitive option out there.

Has anyone dealt with a similar requirement and found a strong solution?

Answer 1

A more concrete example would be better to discern a valuable approach. Especially to learn if different columns of the data might be beneficial in creating buckets. Also to find ways of increasing incentives or joy to deal with the task. Since special columns of a table are not mentioned, I take the scenario as a "one-column-only line" display of 100+ lines.

Quick analyse:

One point to consider is that there will be probably a lot fewer buckets than rows. I assume a maximum of 10 here (but the solution I present allows to add more).

Another point is the "entering of free text" to add rows to a bucket - that is more than detrimental in terms of speed/usability. Also forcing the user to decide about bucket names and number of them needed before having access to the data to work with (splitting into creating buckets and sorting rows phases) is not a good approach.

Ideas to improve:

Present buckets as a concept to the user:

• Prepare a toolbar with 10 buttons - initially numbered from 1 to 10. These are the buckets.
• If user clicks one, don't ask for a name, but present a naming field so the user can decide when to name things on the fly.
• Upon clicking that button, all selected lines are added to the corresponding bucket.
• Additionally provide keyboard shortcuts like Ctrl+1 through Ctrl+0 to allow "clicking" the bucket button quickly (Alt+1 to Alt+0 would be more beneficial because easier to reach - providing both does not hurt).
• Provide something to allow working with the already assigned content of the buckets in case a user wants to correct things. You can indicate the associated bucket with a column which would allow being filtered by and this way a correction to assignments could be made in bulk (as I understand, this column is currently present as the "input bucket name" column and can be re-purposed here).

Benefits:

This way you save typing in any names if not necessary and also prevent typo issues while speeding up assignment to buckets to a great deal.

If bucket names are mandatory ask for them after the assignments are done. Usually a user will have entered names during the process anyway.

For the pro users or to provide more buckets quickly, allow any shortcuts with Ctrl+ or Alt+ to automatically add a button/bucket with that shortcut assigned to it for further adding rows as described.