{"id":10476,"date":"2025-10-03T17:18:32","date_gmt":"2025-10-03T09:18:32","guid":{"rendered":"https:\/\/www.salesforce.com\/?p=10476"},"modified":"2025-10-03T17:18:33","modified_gmt":"2025-10-03T09:18:33","slug":"data-cleaning-how-to","status":"publish","type":"post","link":"https:\/\/www.salesforce.com\/ap\/blog\/data-cleaning-how-to\/","title":{"rendered":"How To Clean Your Data for AI Agents Without Breaking the Bank"},"content":{"rendered":"\n

Bad data costs companies good money \u2014 especially if that data\u2019s fueling AI. <\/p>\n\n\n\n

How much are we talking about? A 2024 survey<\/a> by data integration company Fivetran found that artificial intelligence<\/a> (AI) trained on inaccurate, incomplete, or low-quality data can cost large businesses 6% of their revenue, or an average of $406 million annually.\u00a0<\/p>\n\n\n\n

\u201cIf you don’t have good data, you\u2019re probably not going to be making the best business decisions,\u201d said Karim Habbal, vice president, data management solutions, at Salesforce. \u201cThere\u2019s a real impact on both the day-to-day tactical decisions and the multiyear decisions being made.\u201d <\/p>\n\n\n\n

With that big a bite out of revenue, you\u2019d think leaders would rush to get going on data cleaning. But time, labor, and tech tools are expensive, and some companies don\u2019t want to make the investment. That\u2019s shortsighted. Investing even a small amount can pay off later. Our list of five cost-effective ways to clean data can get you started. <\/p>\n\n\n\n

\n\t
\n\n\t\t
\n\t\t\t

What’s your agentic AI strategy?<\/h2>\n\t\t\t\t\t\t\t

Our playbook is your free guide to becoming an agentic enterprise. Learn about use cases, deployment, and AI skills, and download interactive worksheets for your team.<\/p>\n\t\t\t\n\t\t\t\n\t\t\t\t\t\t\t

\n\t\t\t\t\tThe future starts now<\/a>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\n\t\t
\n\t\t\t\t\t<\/div>\n\t<\/div>\n\n\t\t\t
<\/div>\n\t\n\t\t\t\n\t\t\"\"\n\n\t\t\n\t\t\t\t\t\"\"\n\t\t\n\t\t\n\t\t\t\t\t\"\"\n\t\t\n\t\t\n\t\t\t\t\t\"\"\n\t\t\n\t<\/div>\n\n\n\n

The cost of bad data<\/h2>\n\n\n\n

Many big brands have learned that bad data hurts reputations and bottom lines. One major airline ended up in court<\/a> after its chatbot explained its bereavement travel policy incorrectly, telling a customer he was eligible for a refund when he was not. Elsewhere, a data error in an automated air traffic control system cancelled 2,000 flights<\/a> in the U.K. and Ireland, leaving thousands of travelers stranded and airlines suffering as much as $135 million<\/a> in losses.<\/p>\n\n\n\n

The costs can also be more subtle. A minor typo in a customer\u2019s address could lead to missed communications, missed deliveries, and lost sales. And then there\u2019s customer trust, which is too valuable to put a price tag on. If an AI agent hallucinates or answers questions incorrectly, customers might take their business elsewhere. They don\u2019t care if the AI screwed up. They\u2019ll only remember it was your company. <\/p>\n\n\n\n

How to clean your data in a cost-effective way<\/h2>\n\n\n\n

Your AI agent will only be as good as the data you feed it. But it may be easier \u2014 and less expensive \u2014 to get your data ready<\/a> than you think. Here\u2019s how:
<\/p>\n\n\n\n

1. Prioritise which data needs to be cleaned<\/h3>\n\n\n\n

Start by cleaning only the data your agent needs. <\/p>\n\n\n\n

Salesforce does this with its own agents, which are powered by Agentforce<\/a>, the company\u2019s platform for building and deploying AI agents. When the product team builds an agent, they focus on the task (or tasks) they want the agent to perform. \u201cThose jobs are called \u2018topics,\u2019 and the topics are a way of routing a user query to a specific thing the agent can do,\u201d said Daniel Zielaski, vice president, data science, at Salesforce. Once the product team has identified a topic, they build a \u201ccorpus,\u201d which is the knowledge base an agent needs to carry out its task.\u00a0<\/p>\n\n\n\n

Zielaski pointed to Salesforce\u2019s new sales development representative (SDR) agent as an example. The SDR agent needs clean and updated account, lead, and contact information to write outreach emails to prospects. But it doesn\u2019t need information on how to solve a tech problem. \u201cWe identify the data that will be consumed by a specific topic, and then we focus on improving its overall quality, versus boiling the ocean and trying to clean all our data,\u201d he said. <\/p>\n\n\n\n

2. Manage your labor costs<\/h3>\n\n\n\n

For many companies, the largest data-related cost is labor. A data engineer in San Francisco, for example, earns a median salary of $178,000 per year<\/a>.  And when you build an entire in-house data team, the cost of salaries, training, and benefits can add up. <\/p>\n\n\n\n

Internal teams are crucial for handling sensitive data like health or financial information. They also offer continuity and institutional knowledge. But for less sensitive data, you could use an outside provider or freelancers, which would allow you to pay only for the services you need<\/a>. Or you could use a combo of both, a hybrid approach. <\/p>\n\n\n\n

You can also use Salesforce\u2019s Data Cloud<\/a>, which solves one of the biggest problems companies face: pulling data from different software systems into one place for an AI agent to read. \u201cThe product has been designed so that you don\u2019t have to pay for a large data engineering team,\u201d said Zielaski. \u201cYou don\u2019t have to pay for an architecture team. You don\u2019t have to pay a group of people to go in and use code to move data from one place to another.\u201d<\/p>\n\n\n\n