๐๐จ๐๐ ๐ข๐ฌ ๐๐ข๐ซ๐ญ ๐๐ก๐๐๐ฉ, ๐ง๐จ๐ฐ.
LLMs are responsible and there is no looking back.
The last time I was coding for a living was 20 years back and it was a long painful process of learning the skill / language, applying and getting proficient in it.
But now things are different and let me explain with a use case.
As a marketer we get a lot of lead form fills. Some of the fields are free text like designation.
If I want to segment this list based on departments, I face an uphill task as filters donโt work. Key challenges include
โณ Spelling mistakes
โณ Short and Long form of same designation
โณ Special characters
To solve this, I now enter the challenge / problem statement into LLM and ask for a solution with code.
๐๐จ๐ข๐ฅ๐, I have the code which I can just copy and paste in apps like Jupyter notebook. I only need to know the basics of programming with a good understanding of input data and expected result.
I applied the same process, and got a result which was 85-90% accurate. Had I done it manually, I would have ended up spending 3-4 additional hrs.
LLM gave me 3 options to solve the problem.
โณ Excel formula
โณ Python program
โณ NLP (Natural Language Programming)
I tried excel first but soon realized that excel cannot manage to many conditional statements and I dropped it.
Being an AI / ML enthusiast, I tried to implement the NLP but soon was disappointed with the result. Maybe I chose the wrong model and hence I dropped that too.
I then executed the Python code on ๐๐ฎ๐ฉ๐ฒ๐ญ๐๐ซ ๐ง๐จ๐ญ๐๐๐จ๐จ๐ค.
Donโt get me wrong as it didnโt give me the exact output I was expecting but then I turned back to LLM and it gave me the required changes which I then implemented.
Finally, I got the desired output and was happy with the result.
Knowing the near accurate department wise segmentation of my prospects will help me target them better with my marketing message.
Attaching the code for reference. I donโt claim it is the best code written but it got the job done.
When asked whether programming will remain a helpful skill in the age of generative AI prompts, Huang said, “๐ ๐ต๐ฉ๐ช๐ฏ๐ฌ ๐ต๐ฉ๐ข๐ต ๐ฑ๐ฆ๐ฐ๐ฑ๐ญ๐ฆ ๐ฐ๐ถ๐จ๐ฉ๐ต ๐ต๐ฐ ๐ญ๐ฆ๐ข๐ณ๐ฏ ๐ข๐ญ๐ญ ๐ฌ๐ช๐ฏ๐ฅ๐ด ๐ฐ๐ง ๐ด๐ฌ๐ช๐ญ๐ญ๐ด,” However, โ๐ฑ๐ณ๐ฐ๐จ๐ณ๐ข๐ฎ๐ฎ๐ช๐ฏ๐จ ๐ช๐ด ๐ฏ๐ฐ๐ต ๐จ๐ฐ๐ช๐ฏ๐จ ๐ต๐ฐ ๐ฃ๐ฆ ๐ฆ๐ด๐ด๐ฆ๐ฏ๐ต๐ช๐ข๐ญ ๐ง๐ฐ๐ณ ๐บ๐ฐ๐ถ ๐ต๐ฐ ๐ฃ๐ฆ ๐ข ๐ด๐ถ๐ค๐ค๐ฆ๐ด๐ด๐ง๐ถ๐ญ ๐ฑ๐ฆ๐ณ๐ด๐ฐ๐ฏ.”
Should marketers learn coding in the Age of AI? Comment below.
p.s. Drop your email in the followit box to receive my latest blog. We can also stay connected on various social media platforms, just click the links to follow.
# Segmentation of Prospects based on department using Python
import pandas as pd
# Load the data
df = pd.read_csv("designations.csv")
# Define classification functions
def classify_department(designation):
if “finance” in designation.lower() or “cfo” in designation.lower() or “accounts” in designation.lower() or “controller” in designation.lower() or “tax” in designation.lower() :
return “Finance / Accounts”
elif “hr” in designation.lower() or “human resource” in designation.lower():
return “HR”
elif “sales” in designation.lower():
return “Sales”
elif “legal” in designation.lower() or “counsel” in designation.lower():
return “Legal”
elif “secretarial” in designation.lower() or “cs” in designation.lower() or “compliance” in designation.lower() or “secretary” in designation.lower() :
return “Secretarial”
elif “information” in designation.lower() or “cio” in designation.lower() or “it” in designation.lower() or “cto” in designation.lower() :
return “IT”
elif “procurement” in designation.lower():
return “Procurement”
elif “auditor” in designation.lower():
return “Audit”
else:
return “Other”
# Apply classification functions
df[‘Department’] = df[‘Designation’].apply(classify_department)
#df[‘Seniority’] = df[‘Designation’].apply(classify_seniority)
# Save the result
df.to_csv(“classified_data.csv”, index=False)
