Marketing Chiefs Followed AI After Altering Campaign Numbers

Validity’s new survey finds 92% of SVP and VP marketers have acted on an AI suggestion they later suspected was wrong because of bad CRM data. The Boston firm released the State of CRM Data Management in 2026 on August 25, 2026, after polling 500 B2B and B2C marketing professionals in the United States, the United Kingdom, Brazil, Australia, and New Zealand.

The same study found that 67% of C-suite respondents admit campaign figures are sometimes altered to look better to leadership. Those two facts belong on the same slide. The people most willing to follow the model are the people most willing to bend the numbers it trains on.

C-Suite Marketers Acted on AI They Already Distrusted

Seniority tracks with the habit. Nearly 78% of C-suite respondents said they had acted on an AI recommendation they later suspected was wrong because of bad underlying data, against 41% of individual contributors. The SVP and VP band sits at 92%.

WHO ACTED ON AI THEY ALREADY DISTRUSTED

The people closest to the records are the least likely to treat a suspect score as a decision. That is the split the rest of the report keeps reproducing. Individual contributors live in the duplicates, the empty fields, and the Friday cleanup queue. Their bosses live in the dashboard the model writes.

Ninety-one percent of marketers in the survey said data readiness is critical for adopting AI. Just 21% called their CRM data “very well prepared” for the tools they use or plan to use. The gap is not a secret inside these companies. It is a choice about who has to live with it.

The Campaign Numbers Feeding Those Models Are Already Bent

Validity asked about a quieter practice than missing phone numbers. Across the whole sample, 38% said campaign data is, at times, manipulated to make results look better to leadership. Among C-suite respondents the share is 67%.

An altered conversion rate does not sit in a deck. It sits in the CRM row an agent will read as ground truth the next morning. If leadership has already asked the numbers to look a certain way, the model has no second source. It only has the story it was handed.

I grew up as a scientist and was trained to never just believe, but instead always verify with data and think critically about the results. We take this approach constantly at Validity, from the boardroom to our customers’ meeting rooms. In this sea of AI-inspired change, I am amazed at how many other folks don’t take the time and steps to trust their data and remain proud of fellow leaders who make sure to get this step right.

Mark Briggs, founder, chairman and CEO at Validity, in a statement

Briggs is selling a data-quality company, and the survey is his. The finding still has to be dealt with on its own terms. A C-suite that both alters campaign results and then follows AI built on those results has closed the loop. The agent is not catching the fiction. It is scaling it.

Agents Treat Bad CRM Records as Orders

Two out of three organizations in the survey increased the number of marketing decisions they handed to autonomous AI agents in the past year. Forty-five percent already use agentic AI that can act without human review. At that point a bad data point stops being a report error and becomes an instruction.

The agents are only as sound as the CRM rows they read. A stale job title, a duplicate account, or a padded conversion rate becomes the next send, the next bid, and the next forecast, and it moves before a person can pull it back. Speed is the feature. It is also how an old error travels farther than it used to.

WHAT A BAD RECORD CAN TRIGGER

• A campaign send: The agent mails or messages a contact whose status, consent, or job is wrong in the CRM.
• A lead score: A duplicate or incomplete account gets ranked as if it were a real opportunity.
• A personalized offer: The model writes copy and pricing off a history that never happened that way.
• A budget shift: Spend moves toward a channel whose conversion rate was altered for a leadership review.

Just 21% of marketers said their CRM data is very well prepared to support that kind of autonomy. The rest are putting a machine in front of records they would not defend in a meeting.

Why Boards Push AI Onto Unready CRM Data

Nearly 60% of C-suite respondents, and 52% of SVP and VP respondents, said they feel pressure to put AI tools in place now even though they know the underlying data is not ready. The board sees the dashboard. It does not see the duplicate leads, the missing consent flags, or the campaign field someone edited before the QBR.

Salesforce research across the United Kingdom and Ireland found that 51% of CMOs cite unrealistically high CEO expectations for AI in marketing. Seventy-seven percent of those CMOs expect AI agents to force a redesign of the department. The brief from above is to show movement. Clean records do not photograph as well as a new agent demo.

Gartner, in a February 26, 2025 note based on 1,203 data-management leaders, said companies will abandon 60% of AI projects that lack AI-ready data through 2026. Sixty-three percent of those leaders said they do not have, or are not sure they have, the right data practices for AI. Marketing is running the same experiment from the other end, shipping the agent first and promising the cleanup later.

Poor Data Already Shows Up in Revenue and Compliance

The 2026 sample is not waiting on a future write-down. 62% of organizations said they have already lost revenue directly because of poor CRM data quality, through missed renewals, bad forecasts, lost deals, and campaigns aimed at the wrong people. Poor data has also contributed, to some degree, to compliance exposure at 63% of organizations and to delayed or scrapped campaigns at 67%.

WHAT THE 2026 SAMPLE ALREADY PAID

• Revenue: 62% report a direct loss tied to poor CRM data quality.
• Compliance: 63% say poor data has contributed to regulatory or privacy exposure.
• Campaigns: 67% have delayed or scrapped work because the records could not support it.
• Cleanup time: Nearly a third of teams spend six or more hours a week fixing and reconciling data instead of selling.

Only 41% of organizations said they have a dedicated data governance team or owner for regulatory and privacy risk. The other 59% are asking agents to move faster through a file that nobody owns. Nearly 69% of respondents said a revenue, pipeline, or performance number they or their team presented was challenged or walked back because the underlying data was wrong, and that share rises to nearly 75% among C-suite executives, SVPs and VPs, department heads, and directors.

Twenty-six percent of respondents said more than three-quarters of their CRM data is accurate and complete. The rest are running agents, and presenting numbers, on a file that fails a basic completeness test.

Marketers Want a Watch on the Records

Asked what would most raise their confidence in CRM data for strategy, reporting, and AI, 39% of the sample chose continuous, automated monitoring that catches and fixes issues in real time. That share rises to 47% among C-suite respondents. A single unified platform came second, at 23%. Third-party validation or enrichment came third, at 19%.

WHAT WOULD RAISE CONFIDENCE

• Live monitoring: 39% overall, and 47% of C-suite respondents, want automated checks that fix issues as they appear.
• One platform: 23% would rather consolidate onto a unified system.
• Outside enrichment: 19% want third-party validation layered on top of the CRM.

The ranking is a tell. Leaders who feel pressure to ship AI are not, in this sample, asking first for another suite. They are asking for a watch on the file they already have, because that file is already in production. An agent that can send mail without review makes a quarterly cleanup ritual look as useful as a fire drill after the alarm.

Validity’s 2025 Survey Already Flagged the Same Split

This is the second year Validity has put the same tension on paper. On July 10, 2025, it released a survey of 602 CRM users and administrators in the United States, the United Kingdom, and Australia. Ninety percent called CRM data a cornerstone of operations, yet 76% said less than half of CRM data was accurate and complete. Forty-five percent said their CRM data was not prepared for AI, even as 54% of those organizations were already deploying generative AI tools.

The two surveys are not the same poll. The 2025 sample was CRM users and admins. The 2026 sample is marketing professionals at companies with at least 100 employees, and it adds Brazil and New Zealand. The questions also shift. What repeats is the pattern: executives rate the file higher than the people who work in it, then put a model on top.

TWO VALIDITY SURVEYS, TWO SAMPLES

In 2025, 68% of executives believed their teams had adequate data, while workers spent 13 hours a week hunting for basic information in the CRM. Only 19% of CRM users said leaders actually changed course when presented with countering data, even though 84% of leaders claimed they did. One in four companies in that sample reported a 20% or greater drop in annual revenue tied to data quality, and companies lost an average of 16 sales deals a quarter.

A year later the marketing sample is still describing the same file, only now two out of three shops have given more of it to agents. The monitoring those marketers want is the step they skipped when the first generative tools arrived. Gartner’s forecast that unsupported AI projects get dropped through 2026 is the bill for that skip, arriving while the agents are already live.

Frequently Asked Questions

When Was Validity’s 2026 CRM Survey Fielded, and Who Qualified?

Validity fielded the survey in July 2026 among 500 B2B and B2C marketing professionals at organizations with at least 100 employees in the United States, the United Kingdom, Brazil, Australia, and New Zealand. That is a marketing-leader sample, not a CRM-admin sample, which is why it can be read next to the 2025 study of 602 CRM users without treating the percentages as a single time series.

How Often Did Marketers Act on AI Advice They Later Distrusted?

Across the full 2026 sample, 19% said they frequently presented or acted on an AI-generated recommendation they later suspected was wrong because of poor underlying data, and another 43% said it happened occasionally. The seniority figures in the body (92% of SVP and VP respondents, nearly 78% of C-suite respondents, 41% of individual contributors) count anyone who has done it, not how often.

How Many Marketers Are Very Confident Their CRM Shows True Campaign Revenue?

Only 28% of respondents said they are very confident their CRM provides an accurate view of campaign performance and revenue impact. That is the measurement problem underneath the 62% who already report revenue loss: companies believe bad data is costing them money, and they also doubt the file they would use to prove how much.

Do Marketing and IT Teams Share Ownership of CRM Data Quality?

Only 39% of C-suite executives, SVPs and VPs, department heads, and directors said marketing and IT or RevOps collaborate very well to keep data usable for campaigns, and among senior managers and individual contributors that share falls to 27%. Combined with the 41% who have a dedicated governance owner, most of the sample is automating decisions on a file that still has no single adult in the room.

Until that watch exists, the agents will keep executing the version of the customer the CRM already contains, including the version someone edited for the last leadership review.