Jason Berkowitz shows how to use GPT in Google Sheets to filter out typos, misspellings, and irrelevant keywords from large SEO lists. Paste a custom formula, run it across your keywords in about four minutes, then delete the false results. Converting to static values stops API charges and leaves you with a clean dataset.
Chapters
If your keyword research involves looking at existing organic opportunities and existing keywords that you're currently ranking for, you probably end up with a massive list of unrelatable, unapplicable keywords that are just adding more and more bulk to your data, whether it's typos, whether it's misspellings, whether it's weird versions of a brand name, it's just adding more and more bloat. And we know that it probably doesn't really matter and it could be excluded from a keyword research report because You're probably already ranking for the same similar term. Just want to share what we use internally here at Break the Web, how to integrate AI, GPT a little bit within Google Sheets itself. You can use GPT for Sheets, Sheet Magic, whatever your favorite tool is directly. And all you need is a specific prompt. In this example, our prompt is analyze the keyword. If it's not a part of the English language or contains a symbol, misspellings, maybe some brand names, non-English characters, or doesn't really resemble anything that would be typically SEO related. Give a false response. If it does include a recognized English word, maybe a related foreign term, plural forms with or without an apostrophe, maybe words with accents, respond with a true. So I'm going to copy this, and in Google Sheets itself, I'm going to paste and insert a column right, and then I'm going to use the formula. So in our case, the tool I use, I'm going to do equals AI to use the OpenAI API, highlight a cell, comma, quote, paste in the prompt, end quote, end comma, and bam, let it do its magic. We're going to see that this is going to be, of course, a true response because it makes logical sense that this is a normal keyword. And then we can just simply double-click and expand this to the entire keyword set. And behind the scenes, the API request and Google Sheets will work to classify true or false. I'm going to pause this and come back in just a moment. So this Google Sheet has about 1,000 rows. Took, I believe, about 4 minutes of letting it run behind the scenes. So once we have this data, of course, we just want to copy and replace as values only so that now everything is true values and we're not consistently making API calls and racking up the API usage costs. Now we can filter out by items that are just false. You might see some other things in there as well. That might have been missed. This is a weird one, but you can see a lot of misspellings, typos, things that again maybe have some weird symbols in there. Overall, we have a lot of keywords that now ensure that we can go ahead and delete this row, make sure that every single row that we are looking at and analyzing is nice, clean, and concise. Hopefully this helps you out a little bit.
