Design for input of time-series data

Question

I'm designing an application page where the user can input multiple time series.

As an example, this can mean updating the daily product forecast for a 5 stores over the next 14 days. This results in 5 different time-series to be updated, each with 14 data-points. Users are planners and analysts, so they know their data.

All time-series data in the application are inducted via ETL data pipelines. In our case, this would be system generated forecast time-series. The user wants to override or modify an existing time-series.

• In some cases, the modifications can be limited (just for a few time-periods)
• In some cases, user wants to override (e.g. all time-periods set to a new value)
• Sometimes the modification can be based on a formula (e.g. increase by 10%)

I've looked in the archives on past questions, and a few of them was insightful.

1. One of them let the users download an Excel/CSV template, populate and then upload.
2. Another one had an interesting suggestion to let users maintain data in Google sheets and import the data via API integration. I do not know how this would work in the context of a larger app.
3. A third option would be to develop a custom UX/UI based on data-table components like ag-grid. The disadvantage to this will be a lot of development that would most likely be not as good as manipulating data on a spreadsheet. Advantage would be that we get a consistent look and feel.

My question is which of these 3 options would provide the easiest/most natural user experience. What I care about the most, in the decreasing order of importance:

1. User-experience, where the UI does not stand in the way of the user.
2. Faster to develop, easier to maintain
3. Modern look and feel. This is subjective, I know, but I'm trying to make the application stand-out from existing enterprise apps which look awful compared to modern consumer web apps.

Answer 1

Purely as a matter of approach, I think that creating a custom UX is by definition supposed to provide better UX than using spreadsheet software, precisely because your UX can be tailored to your needs and users, while spreadsheet software is spreadsheet software. Not to say that spreadsheet software is bad (I think that Excel is a wonder of creation), but it's extremely generic and meant to fit the lowest common denominator of countless different use cases. Whether you have the resources to go ahead and really design and develop the perfect UX for your use cases is another story altogether.

A big caveat to the above is that analysts might be so used to Excel that initially it would be the best way for them regardless of how great your UX is, but I think that it's a matter of time and if the tailored one is indeed better, they could be won over.

And as for designs, this is one potential direction you could explore (blue=selection, single-cell edit is inline)