Forecasting KPI in Dashboard
- Thread starter Kamarlon
- Start date
Probably different for everyone, but...
A dashboard isn't supposed to invent or be the source of business intelligence (opinion), it's job is to support and supplement what's already around in an informative and fresh way. If you are intended to project KPI, look for projections already out there to assist you. Good sources are budget plans, staffing plans, sales look-aheads, etc (accounting, finance, sales, human resources) internally. If you are lucky enough to have leading indicators from external data sources (stock prices, home sales, unemployment numbers, internet clicks, etc), go find that information. Then, look at how that information compares with current trends to see which factors can be used reliably to project things for your company.
I'd probably say it's more of an art than a science, but that doesn't mean science isn't involved...
: )
A dashboard isn't supposed to invent or be the source of business intelligence (opinion), it's job is to support and supplement what's already around in an informative and fresh way. If you are intended to project KPI, look for projections already out there to assist you. Good sources are budget plans, staffing plans, sales look-aheads, etc (accounting, finance, sales, human resources) internally. If you are lucky enough to have leading indicators from external data sources (stock prices, home sales, unemployment numbers, internet clicks, etc), go find that information. Then, look at how that information compares with current trends to see which factors can be used reliably to project things for your company.
I'd probably say it's more of an art than a science, but that doesn't mean science isn't involved...
: )
Thanks Jeremy. That's what I think too but they are insisting on me applying a statistical technique to forecast these KPIs which are sooo wide ranging. I have a few ideas though will decide on one to use. The second problem is my company isn't too big on data, you have to pull an arm and leg to get what you want. Your point is valid since each KPI is affected by different varibales.
Do you know of any course which teaches Business Intelligence?
Do you know of any course which teaches Business Intelligence?
No, I don't know enough to recommend a good BI course, but I'm sure Google and a few reviews online could give you a place to begin looking...
On, applying statistical models to predict future KPI, it's pretty much your call there. Every business is different, but I can ask a few focus questions...
-Do you have past performance information for this time last year?
-If you're predicting certain KPI with limited info, is there a couple "inputs" that mgmt-level people could drop into a few cells to give them a good look at some of the resultant KPI? I do this sometimes with sales/staffing plans, and it puts the onus on them to think about their own future plans without forcing them to do a bunch of calculations every time they want to see what the result would be.
-If you can't define a good prediction, can you at least define a range of values by looking at the risk of things that definitely won't change much versus things that have lots of variability? Think in terms of risk management/analysis.
-What pieces of information have they used in the past to predict future outcomes before they had you, the graphing genius, come along?
On, applying statistical models to predict future KPI, it's pretty much your call there. Every business is different, but I can ask a few focus questions...
-Do you have past performance information for this time last year?
-If you're predicting certain KPI with limited info, is there a couple "inputs" that mgmt-level people could drop into a few cells to give them a good look at some of the resultant KPI? I do this sometimes with sales/staffing plans, and it puts the onus on them to think about their own future plans without forcing them to do a bunch of calculations every time they want to see what the result would be.
-If you can't define a good prediction, can you at least define a range of values by looking at the risk of things that definitely won't change much versus things that have lots of variability? Think in terms of risk management/analysis.
-What pieces of information have they used in the past to predict future outcomes before they had you, the graphing genius, come along?
Jack
Member
Kamarlon dashboards can definitely show a target or forecast, what ever helps the business focus on the key issues. Re statistics because the past is not always an appropriate predictor of the future i think forecasting is more of a 'black art' and not so much a science. Some businesses may follow more closely a past pattern than others (eg like essential goods like milk) but others may not because of any number of factors ie are they luxury items, weather, new competition, the general state of the economy, what ever. Look how economists try to forecast exchange rates and how most are miles out. So you use what ever works on a case by case think. Books focussing on business analysis, business or investment valuation (this is the ultimate use of forecasting) and KPIs would be worth looking into on amazon. Perhaps your local accounting body has seminars on this topic. Also your local business council or management associations may have seminars on budgeting and forecasting.