Data Extraction Made Easy: Using AI to Populate Missing Company Information
Have you ever found yourself facing a mountain of data, only to realize that some of the fields are missing valuable information? If so, you’re not alone. As marketers, we rely heavily on data to make informed decisions, but sometimes that data can be incomplete or inaccurate. Fortunately, there is a solution: artificial intelligence.
We were recently approached by a client at Ivenium Marketing with a problem that’s all too familiar – missing information on their list of companies. They had the basic stuff, but the juicy details like industry, revenue, and headcount were nowhere to be found. In the past, filling in these missing data gaps would have required hours of tedious manual work and online investigations – or a pricey investment in data append services. But we decided to have some fun and put AI to the test instead.
Using some fancy AI wizardry, we were able to automate the whole process and get our client the missing data in record time. Sure, the results weren’t 100% perfect, but they were still incredibly helpful. Who knew that our machines had a knack for playing detective and digging up some seriously useful info?
If you find yourself in a similar situation, do not spend your valuable time on laborious data-entry tasks. Let us show you how AI can turn a headache into a breeze, and maybe even have some fun along the way. And the best part? While costly data append services were the only option in the past, now you can achieve the same results for a lower price or even for free using AI-powered tools like ChatGPT.
Prompt: I need you to fill in missing information in our dataset. Specifically, we have several CSV files containing values for Industry, Sub-Industry, Company Size, and Company Revenue. Please use these files to fill in missing values for these columns in our dataset.
As for the Website column, we don’t have any specific instructions for how to populate it, so please use your best judgement and skills to fill in this information.
I’m going to load the CSV files now so you can get started.
ChatGPT: Sure, I’m ready to assist you with filling in the missing information in your dataset. Please go ahead and load the CSV files, and I’ll get started on processing the data to fill in the missing values.
Prompt: Can I upload files here?
ChatGPT: Sure, here’s a summary of the possible ways a user can share CSV files with ChatGPT:
- Copy and paste the contents of each file into our chat window if the CSV files are small enough*
*Copy and paste those CSV files into the chat window like it's 1999! Sure, it might take a bit longer, but ChatGPT will appreciate the personal touch. Plus, it's the best way to ensure your data doesn't get lost in cyberspace like an astronaut's sock.
- Email the CSV files to yourself and then provide the information needed to access the files to ChatGPT*
*Well…don't give ChatGPT access to your email unless you want it to start sending emails to all your contacts with subject lines like 'I am a robot, beep boop!' and 'Greetings, fellow humans!'. Trust us, your reputation will never recover.
- Upload the CSV files to a cloud storage service such as Google Drive, Dropbox, or OneDrive and then share the link to the file with ChatGPT*
*Warning! If you give ChatGPT a link to a publicly accessible file, it might stare at it blankly, scratching its virtual head, and wondering why the file is not magically appearing in its chat window. Remember, ChatGPT can't upload files to the cloud and links can be a tricky business. So, let's stick to plain old copy-pasting, shall we?
Since we don’t want to use publicly accessible links to the CSV files and we want to avoid giving ChatGPT access to our emails or cloud storage, we’ve decided to try the old-fashioned method of copying and pasting the CSV file contents into the chat window. While it might take a bit longer, this solution ensures that our data stays private and doesn’t get lost in the vastness of the internet. Plus, who doesn’t love a bit of retro charm?
Check out how ChatGPT improved our dataset accuracy from 42% to 91%!
We limited the sample size for this AI-powered data enrichment to just 30 companies for the purpose of this article, to stay within the query limit. It’s entirely possible that ChatGPT’s performance might vary when dealing with larger datasets or different types of data. But based on what we’ve seen so far, I’m pretty impressed with its abilities. Ready to see how ChatGPT performed when put to the test? Buckle up, because the results might surprise you.
Here’s our initial set of data. Please note that for the purpose of our blog post, we opted to utilize publicly available data from Wikipedia to compile our dataset of companies and their corresponding industry, sub-industry, company size, company revenue, and website.
After receiving access to the CSV files, ChatGPT started to fill in the missing values and update the dataset using the techniques they had mentioned earlier.
Our initial dataset had a fill rate of only 42%, meaning that over half of the cells were empty. Although existing field values were 100% accurate, missing data was considered invalid.
We provided ChatGPT with standard picklist values for Industry, Sub-Industry, Company Size, and Company Revenue, and it was able to fill in missing values for these columns.
Additionally, we left the Website column blank, and ChatGPT used its best judgement and skills to populate this field.
After applying the AI algorithm, we were delighted to witness a significant boost in our dataset’s fill rate, resulting in a remarkable 100% rate. However, we must acknowledge that our accuracy rate is 91% due to some errors and non-standard values that were provided within the response. Nevertheless, the progress we made was substantial and proved to be a great time and effort saver compared to manual methods of filling in missing data.
There are several potential benefits of using AI data completion to fill in missing values. First, it can significantly reduce the time and cost of completing large datasets. This is especially true when dealing with datasets that are too large or complex for humans to handle within a reasonable amount of time. Additionally, AI can reduce the potential for human error, which can be costly and time-consuming to fix.
However, there are also some potential risks to consider when using AI to fill in missing values. One of the main risks is that the AI may not be able to accurately identify the correct value for every missing cell, resulting in inaccuracies or inconsistencies in the dataset. Additionally, if the AI is not trained on a representative sample of the dataset, it may make incorrect assumptions or fill in missing values with values that are not representative of the dataset as a whole.
Despite these potential risks, we believe that the benefits of using AI data completion outweigh the risks. By using AI to automatically populate missing values, we were able to improve our dataset accuracy from 42% to 91%.This saved us a significant amount of time and effort and allowed us to focus on other tasks that required our attention.
In conclusion, we recommend using AI data completion to anyone dealing with incomplete datasets. While there are some potential risks to consider, the benefits of using AI to fill in missing values are numerous and can result in significant time and cost savings. We look forward to using ChatGPT in the future and exploring other ways to incorporate AI into our workflow.
Bonus:
I have to admit, when I asked ChatGPT to fill in missing information in our dataset, I wasn’t expecting it to go beyond the CSV files I provided. But to my surprise, when I asked for more rows to be added, it conjured up an entirely new set of companies that weren’t in the original dataset! I mean, I knew AI was capable of a lot, but adding completely new companies? It was like having a data scientist on steroids. The next thing you know, ChatGPT is going to ask for a raise and take over my job. Just kidding, I’m pretty sure ChatGPT doesn’t need a raise or a job. It’s just happy to help us mere mortals with our data woes…
Summary:
Well, that’s a wrap on our exploration of data enrichment with ChatGPT. We hope you found this post as enlightening as we did! At Ivenium Marketing, we’re always on the lookout for innovative tools and techniques that can make our lives easier – and let’s be real, more fun. With ChatGPT (and other crazy-smart AI tools), we’re excited to see what new opportunities and insights we can uncover for our marketing automation teams.
We’d love to hear your thoughts on this post and the topics we covered. Did you find ChatGPT’s capabilities impressive? Have you had any experience using ChatGPT or other AI tools in marketing automation?
If you’re interested in exploring this topic more deeply, be sure to check out our upcoming blog posts on AI and its capabilities for the Marketing Automation industry.
Also, check out our latest articles:
Stay tuned for more content from us!
