Values of same report different on the service and on desktop

[pedrocDAcamara]

Hello everyone

Have anyone happened this? My visual shows different values on the service

than on desktop. 

Anyone can help me? I've tried also clear cache on the Edge but I don't think it worked.
Thank you all in advance

[luisrh]

This is one you can look at.  https://youtu.be/3mox85RpySU

Also suggest you read this -  Introduction to dataflows and self-service data prep - Power BI | Microsoft Learn

good luck!

[luisrh]

I assume you are using Import mode.   If you are truly looking to see the same data in PBI desktop and a pbi report in the service,  the best way to manage this and save time in loads and consistency is to create Dataflows in PBI.COM and point PBI desktop to them as well as in the service.   This way you load data in one place once and don't have to worry about consistency on PBI desktop as well.

[pedrocDAcamara]

Hi @luisrh 
Can you please tell me how do I do that? Do you have some videos that teach us how?
Thanks a lot

[edhans]

Either the service isn't up to date (fully refreshed) or there is a filter set. Hover over the visual and look at the filters shown.

It may be a slicer or filter pane setting has been selected. You can reset all filters by clicking the counter-clockwise button in the upper right of the report.

Otherwise we will need more info. But this isn't a bug in the engine. It is just how the data is being filtered, or the dataset in the service is out of date.

[pedrocDAcamara]

Hello @edhans 
First of all thank you for the answer.
I've already checked all the filters in the desktop.
I'm not sure if this might be the reason or not, but I believe we already refresh the data more than 8 times when I tried here to do one more. Maybe that's why it didn't show me the correct values. 
Does it make sense to you?

[edhans]

If you are connected to a live system, you will almost never see the same results in the service as you do in the desktop because they won't refresh at exactly the same time so one will have more/less records than the other.

[pedrocDAcamara]

Hi @edhans 
I'm not. It's an import mode.
But let me tell you what I did. I've created another workspace, published and it worked. Now the million dollar question : Why does it work on one and not on another workspace? I don't think it has anything to do with RLS or incremental refresh because the value is exactly double what it should be. Any thoughts on this? Anyone?

[edhans]

Sorry, when I said a live system, I meant is the Power BI report refreshing against an ERP system that is functional. So if it refreshes at 12:00:00pm, it will get one number, whereas if you refresh at 12:00:02pm you might get a different number because in those 2 seconds, an invoice could have been entered.

The workspace issue isn't it. There is something else going on, and I'd just republish to the original workspace. There is no difference in the way workspaces calculate data.

[pedrocDAcamara]

Hi @edhans 
I'm 99% that the problem is about the incremental refresh. I've defined to update all my records for the last 5 days, and I'm sure it is doing it but it double it, it duplicates all the values. I don't think it makes sense.
I have never heard anything about this nor I've seen any videos to show us what could happen. 
Let me know your thoughts plase

[edhans · Accepted solution]

What is your filter set as? You must ensure the = sign is only on one side.

Good:

[SomeDate] > RangeStart and [SomeDate] <= RangeEnd

Bad:

[SomeDate] >= RangeStart and [SomeDate] <= RangeEnd

The bad version will cause duplicated data. For example, the range for March is March 1 to April 1. The first example will return March 2 - April 1. That is fine. In April it wrill return April 2 - May 1. No overlap.

The bad example will return March 1 - April 1. The problem is in April, it will return April 1 - May 1. The problem is April 1 will be in both partitions, and thus doubled.

[pedrocDAcamara]

Hi @edhans 
Thanks for your help. It was the configuration of incremental refresh that was duplicating my values. In order to not happen that again, this is how I've done it.

Not choosing anything it will avoid dups.
Thank you so much

[edhans]

Yeah, I think it would be possible to get dupes if you have Detect Data changes selected but were using the same date field as the incremental policy used and not a specific "date updated" field.