Fixer of dirty data, classification and data accuracy - Susan Walsh

Podcast Show Notes

Fixer of dirty data, classification and data accuracy - Susan Walsh

This episode Susan Walsh - The Classification Guru gives us a whirlwind tour of what money is going missing in bad data with Cliff Notes: Lead manufacturing host Tristan Bailey.

She can clean it up quickly and affordably, saving your team hours of messing around in Excel and with product code mix ups. 

In offices and ordering systems for manufacturing businesses Susan often sees large overlap in miss entered data or different SKU from different suppliers. So clean up saves time and money.

Learn from the Cliff Notes on Data Classification

• Tends to be financial or spend data, for finance and procurement teams
• If I could show you have 50 suppliers in a category, there would be an opportunity
• Some people don't see it as an investment but a cost, but the % you pay to classify your data is a very small number vs what you could save is massive
• A big SKU problem 70,000, taken 30k sample, a lot were same and just named differently, so I got it down to 24k
• I can train the customer, in excel though it is very slow, so can be easier to sent it to me
• When suppliers change data, I can model that and let you just drop in the new data and merge it all
• Data has just exploded in the last year, so I'm working with clients more regularly
• No matter how many teams you have in your company there are always data issues even between teams
• Tip: Consistency, make sure all teams are all on same standards

Susan Walsh joined me on Cliff Notes this month, see what she can do for the data that flows through your business.

Resources

• Contact Susan work with data from The Classification Guru
• Tweet with @ClassificationG
• And I always like when you share your thoughts @cliffnotespod

Transcript Highlights

• Any one can work with data
• Worked with spend analytics company for 5 years
• I could offer this as a service, without the dashboards and reporting,
• For the last 3 years I have been doing that and became the "classification guru"
• Tends to be financial or spend data, for finance and procurement teams
• Depends on detail of the data I can classify it
• How much spend per cat, how many suppliers in each cat
• If I could show you have 50 suppliers in a cat, there would be an opportunity
• Often there is a preferred supplier list, but regional offices tend to like to still buy from their favorites
• Some people don't see it as an investment but a cost, but the % you pay to classify your data is a very small number vs what you could save is massive
• "How important clean data is"
• Q: is this just for enterprise?
• Even if you are a solo trader, you data is still important to you
• Loads of way to apply categorisation
• Q: sales warehouse, other business too?
• I'm a data trouble shooter, so I am able to fix problems businesses have had for a little while
• Once I have fixed a problem, I can share the general issue with my wider audience and they might think, I have that too
• A vig SKU problem 70,000, taken 30k, a lot were same and just named differently, so I got it down to 24k
• I work with data modeling and automation
• I can train the customer, in excel though it is very slow, so can be easier to sent it to me
• One of the biggest problems is cut and paste errors with filters on, and it makes a mess and have to start again
• I work with Omniscope (Supply Example) where you only work with the data you need
• The best you can do in Excel is conditional formatting and then filter on the colours
• If you are having issues with excel, send it to me and I can get it back to you in an hour
• I am used to working with small and large
• One had 9 different files 2.8mill rows, and de duped to 1.4m
• Q: do you have good privacy and protection policy?
• Fully GDPR compliant, and sign NDAs with client
• Two factor login and backed up as well
• Funny story, I have a bluetooth keyboard, and put it on a chair while I was watching a movie, I thought I had broken my computer, something was sitting on mouse in the other room
• Confusion, around part numbers, missing or half descriptions, unique codes without duplicates, part names have changed, there is no one reason people get in contact, just when there is a need
• When suppliers change data, I can model that and let you just drop in the new data and merge it all
• Before the first project I will have a call, to talk to them about the problems, and see a sample of the data
• I like to price per project and the sample helps me judge how long, when I follow with a first draft so we can review
• Its better for the Client and me, if there is a fixed price, regardless of how long it takes me
• Which makes it easier on budgets and to get approval as well
• The cleansing exercise took 5 days, classification is a bigger job maybe 2-3 weeks
• Data has just exploded in the last year, so I'm working with clients more regularly
• No matter how many teams you have in your company there are always data issues even between teams
• Q: tips?
• Consistency, make sure all teams are all on same standards

Posted on 14th September 2020

Tweet

Do you know someone who would make a good guest?

Send me your recommendation, recording is on a 4-6 weeks schedule so can book around their schedule.