The Real Cost of Bad Data in Salesforce (and How to Calculate Yours)

Bad data feels free. Nobody sends you an invoice when a rep dials a disconnected number or an email bounces off a contact who changed jobs eight months ago. The cost is real, though. It just arrives in pieces, spread across wasted hours, misrouted deals, and reports the leadership team quietly stopped trusting.

This article gives you the famous numbers, explains what they actually measure, and then hands you a formula to calculate the cost inside your own Salesforce org. No vague hand-waving. A per-record model you can run against your own Account, Contact, and Opportunity data.

The famous numbers, and what they measure

Two statistics get quoted in almost every article on this topic. Both are worth knowing, and both are easy to misread.

Gartner’s widely cited figure puts the average cost of poor data quality at $12.9M per year for an organization. That number covers everything: rework, wasted campaigns, bad decisions, lost productivity. It is an average across large enterprises, not a per-record price and not a Salesforce-specific one. Treat it as a headline, not a calculator.

The second is IBM’s widely circulated 2016 estimate that bad data costs the US economy around $3.1 trillion per year. Again, this is a macro figure. It tells you the problem is enormous at national scale. It does not tell you what your org loses next quarter.

These numbers are useful for one thing: getting attention in a budget meeting. They are the wrong tool for sizing your own exposure. For that, you need a model built from your record counts and your team’s economics.

The 1-10-100 rule, applied to CRM records

There is a classic quality heuristic worth borrowing here. The 1-10-100 rule, attributed to George Labovitz and Yu Sang Chang, describes how the cost of a data error grows the longer it goes uncaught.

• $1 to prevent. Verify a record at the point of entry. A validation rule, a required field, a quick check.
• $10 to correct. Fix the error later, after it is already in the system. Someone finds it, investigates, and cleans it up.
• $100 to fail. Let the bad record cause a downstream failure. A deal routed to the wrong rep, a compliance breach, a contract sent to a dead address.

The ratio is illustrative, not a law of physics. The lesson holds regardless of the exact multiplier: a bad record gets more expensive every day it sits undetected. A missing Email on a lead costs almost nothing on Monday. Six weeks later, after marketing has run three campaigns past it and a rep has chased it twice, the same gap has quietly billed you many times over.

The cost buckets that actually hit a Salesforce org

Enterprise averages hide where the money goes. In a Salesforce org, bad data drains value through a handful of specific channels. Size each one and you have your number.

Wasted rep time. Reps work bad records: dialing dead phone numbers, re-keying duplicate Contacts, hunting for information that a complete record would have carried. Every minute here is paid time producing nothing.

Bad routing and assignment. A wrong Region, a stale Owner, a duplicate Account splits activity across two records. Leads land with the wrong rep or sit unassigned. Deals stall because nobody clearly owns them.

Marketing waste. Campaigns run against invalid or decayed contact data. Emails bounce, ads target people who left the company, and paid spend chases records that cannot convert. The spend is real even when the audience is not.

Compliance and risk exposure. Personally identifiable information in the wrong field, records you were supposed to delete, contacts who opted out but still get mailed. Each one is a potential fine or a broken trust obligation.

AI failure. This bucket is new and growing fast. Gartner projected in 2025 that through 2026, organizations will abandon 60% of AI projects not supported by AI-ready data. MIT’s 2025 report, “The GenAI Divide,” found that roughly 95% of GenAI pilots deliver no measurable P&L impact. When an agent reads a stale Opportunity or a duplicate Account, it does not average the error away. It acts on the individual record, confidently, and the wrong answer reaches a customer.

How do you calculate the cost of poor data quality?

Multiply the number of bad records by what each bad record costs you, then add up the cost buckets that apply to your org. The core formula is: (number of records × error rate) × cost per bad record = annual cost of poor data quality. Run it once per bucket (rep time, routing, marketing, compliance, AI) and sum the results for a total.

The hard part is honest inputs, not the arithmetic. You need three things: how many records you hold, roughly what share are wrong (missing, duplicate, stale, or invalid), and what a single bad record costs in each bucket. The first two come from measuring your org. The third comes from your own economics, a loaded hourly rate for reps, a cost-per-lead for marketing, an estimated exposure for compliance.

Start with a defensible error rate. If you have never measured, do not guess in your own favor. A conservative baseline is a fraction of your records; industry commentary often cites double-digit decay per year for B2B contact data. Measure to replace the estimate with a fact.

Worked example: a 100,000-record org

The numbers below are a hypothetical illustration, not a benchmark for your org. They show the shape of the calculation so you can swap in your own inputs.

Assume an org with 100,000 Contact and Account records and a 20% error rate. That is 20,000 records carrying a problem: missing fields, duplicates, invalid formats, or gone-stale data.

Cost bucket	Assumption (hypothetical)	Records affected	Cost per record	Annual cost
Wasted rep time	5 min lost per bad record, at $40/hr loaded	20,000	$3.33	$66,600
Bad routing	10% of bad records misroute a deal touch, $50 rework each	2,000	$50.00	$100,000
Marketing waste	40% of bad records emailed in dead campaigns, $2 wasted spend each	8,000	$2.00	$16,000
Compliance risk	1% carry a PII or opt-out exposure, $500 expected cost each	200	$500.00	$100,000
AI failure	5% feed an agent a wrong answer, $75 per incident	1,000	$75.00	$75,000
Total				$357,600

The headline for this hypothetical org is roughly $358,000 per year, from a data problem nobody was invoicing. Change the error rate to 10% and the total roughly halves. Change the rep rate or the compliance exposure and it moves again. That sensitivity is the point: the number is yours to build, and every input is a lever you can pull.

The reusable formula

Here is the model in a form you can lift into a spreadsheet today.

For each cost bucket:
  bad_records = total_records × error_rate
  bucket_cost = bad_records × affected_share × cost_per_record

annual_cost = sum(bucket_cost for every bucket)

Fill in five rows, rep time, routing, marketing, compliance, and AI, and you have a defensible figure for a business case. It will not match Gartner’s $12.9M, and it should not. It matches your org, which is the only number a budget owner can act on.

What fixing it costs, by comparison

Set your annual cost next to the price of finding the bad records in the first place. The gap is usually large, and it favors measuring early.

The 1-10-100 rule already told you why. Detection is the cheap end of that curve. You are paying $10-and-climbing costs today because errors sit uncaught; the fix is to move detection as close to prevention as you can. You cannot clean what you have never measured, so the first dollar goes to visibility.

This is where a data quality scan earns its place as the cheap first step. Data Quality Sense is a Salesforce-native tool that scans your records in batch and reports where the problems are: completeness gaps on revenue-critical fields, duplicate Accounts, stale Opportunities, invalid Email and Phone formats, and PII sitting in the wrong place. It does not silently rewrite your data. It surfaces the bad records and lets your team act on them through Tasks and Chatter, so a human decides what to fix. That measured error rate is exactly the input your cost model was missing.

Once you have a real error rate from a scan, drop it into the ROI calculator to turn the percentage into a dollar figure for your org. Same formula as above, done for you, grounded in your actual numbers instead of a hypothetical 20%.

Next steps

• Understand why the underlying problem compounds: Why data quality matters.
• Learn how to put numbers on your data health: Measuring data quality.
• Turn your measured error rate into a dollar figure: ROI calculator.