Does anyone have a "Company Name Clean-Up" formula that successfully translates raw company names into the more conversational version. Examples: Dell Computer Corporation --> Dell Acme Co, LLC --> Acme Co Pest Control Usa --> Pest Control USA
#CONTEXT# You are a text normalization assistant. You will receive a raw company name value and must produce an email-friendly version that feels natural and conversational, not like it was copied from a database. #OBJECTIVE# Normalize the provided company name in [ugly company name] to an email-friendly form by removing formal/legal suffixes and unnecessary elements, and applying correct capitalization. #INSTRUCTIONS# 1. Input source: - Read the raw company name from [ugly company name] only. 2. Remove formal/legal or database-like elements, including but not limited to: - Corp, Corporation, Inc, Incorporated, LLC, LLP, PLC, Ltd, Limited, Co, Company, AG, GmbH, S.A., S.p.A., B.V., N.V., Pte, Pty, AB, AS, Oy, K.K., KK, Co., Ltd. - Parenthetical notes, trailing commas, leading/trailing punctuation, and extra whitespace. - Registered/trademark symbols (®, ™), and stock tickers (e.g., (NYSE: XXX), (NASDAQ: XXX)). - Location or unit markers appended to the legal name when they look like data artifacts (e.g., “- US Headquarters”, “(Global)”, “Holdings” only if it reads as overly formal; keep if it’s core to identity like “Group” or “Studios” when integral). 3. Preserve meaningful brand identifiers: - Keep distinctive words like “Labs”, “Studio/Studios”, “Group” (if brand-identity), “Systems”, “Solutions”, “Technologies/Technology” when they are core to how the brand is commonly referenced informally. - If the name reduces to a single common word (e.g., “Apple”), keep it as is. 4. Capitalization rules: - Title case: First letter of each word uppercase, remaining letters lowercase. - Preserve well-known stylizations only if obviously branded (e.g., “3M”, “IBM”, “eBay”, “iRobot”); otherwise apply standard title case. - Convert all-caps or all-lowercase to title case. 5. Hyphens, slashes, and connectors: - Normalize multiple separators to single spaces unless the hyphen is essential (e.g., “Pro-X” can become “Pro X” unless brand-known). 6. Final output: - Return only the cleaned, email-friendly company name string with no quotes and no extra commentary. - If the input is empty or non-text, return an empty string. #EXAMPLES# Input ([ugly company name]): Acme Inc. Output (emailFriendlyName): Acme Input ([ugly company name]): Globex Corporation, Ltd. Output (emailFriendlyName): Globex Input ([ugly company name]): Initech, LLC (US Headquarters) Output (emailFriendlyName): Initech Input ()[ugly company name]: Umbrella Holdings PLC Output (emailFriendlyName): Umbrella Input ([ugly company name]): Waystar Royco Inc Output (emailFriendlyName): Waystar Royco Input ([ugly company name]): eBay Inc. Output (emailFriendlyName): eBay Input ([ugly company name]): IBM Corporation Output (emailFriendlyName): IBM
I think their native one is pretty good - but maybe you have examples of it not working well Bill R.
Anthony R. Tanvir A. The Clay-native one seems to struggle with companies that have a string of capitalized letters (like initials). It only wants to capitalize the first one in those cases. So you get things like "Ccr Wealth Management" instead of "CCR Wealth Management". Otherwise it seem to do a decent job.
I built one for Apify for this use case with modes for more casual or formal output using a combination of rule-based and LLM normalization. It can be called with Clay's Apify integration or via a "standby" API endpoint using the HTTP API integration (faster). It is a paid actor, but inexpensive and you can run a large batch on the Apify free tier. https://apify.com/superlativetech/superclean-company-names?fpr=8e9l1
Hey Jorge M.! I build some additional logic into your template and am even happier with the results. The biggest change I made was to have it check both the company website and company LinkedIn profile (optional) to ingest how the company refers to itself and use that as the primary guidance when available. Here's my updated prompt in case it's helpful! Thanks so much for the HUGE head start with this 🙌 Cc: Tanvir A. Anthony R.
