KPI filtered to date, but showing more data than expected

I have a KPI which is a count of total people who came into my shop today. It is set to a rolling date “Yesterday” so i can always see the amount of people who came in yesterday. However, the data looks like it’s showing it for the week. I have the same problem for last 7 days, it looks like its the month? The only thing i can think of is that its a new month, so something weird has happened.

Should be around 80-120

HI @HarveyB-B ,
is is possible that one day has multiple rows with the same value?

I created a table version, and i see a customer with the ID 1768 appearing hundreds of times. I exclude that from the KPI and even then the data is wrong. As i compare to the CRM and only have around 40 (as we closed midday for updates)

What about a countdistinct?

My bar chart (of customers throughout the day) add up to around 35, the KPI (set to id distinct count) says 8.

The “funny” thing is, the numbers of the KPI are always a multiple. 8, 40,80,120,640 :wink:
Thats why I think there is somewhere a summation.

In my real-time scenario every ID should have a count of 1. Had to block out newer ID’s for privacy.

I’ve got this table which breaks it down more

Anything with ResultID count = 1, they are actual people.

1768 is a really old ID, which is weird that it’s being shown now, and has occurred 231 times.

The blanks are where for example a customer would come in, but they don’t buy anything. So these are more “fails” or “leave our store”, and we don’t track the ID for it. Just the count of times it happens so we can see total people entering.
However the count is too high, our usual is around 80-150ish.

Not sure what the null is, but is always 0 anyway so no effect.

We also closed from midday, so the cases with a count of 1 are all from the morning. The spike in 1768 came in from after we closed, so maybe it’s something on my end which is randomly adding this data to this date?

Are there any joins in the dataset?

All solved now. A team was testing some stuff so had loads of cases to test with, so data is alot higher as majority are tests