AI spend surges, but only 12 per cent of firms can prove marketing returns: Comviva report

NEW DELHI: As artificial intelligence races into marketing budgets, many organisations are discovering that proving its value is easier said than done.

In a classic case of “all AI and no ROI”, a new report from Comviva reveals that while 90 per cent of organisations have increased their AI marketing investments over the past two years, only 12 per cent can clearly demonstrate a measurable business impact from those investments.

The findings come from Comviva’s latest Global CMO Survey Report, The AI Efficiency Divide: Measuring AI’s Real Value Beyond the Hype, which paints a picture of an industry embracing AI at speed while struggling to quantify what it is actually delivering.

The study highlights a widening accountability gap as marketing leaders face growing pressure from management teams to justify AI spending. While enthusiasm for AI remains high, only 16 per cent of respondents said they were confident in defending AI investments with clear business evidence. The majority continue to rely on estimates, assumptions or incomplete measurement frameworks.

According to the report, 35 per cent of organisations depend on rough estimates to assess AI performance, while 32 per cent track campaign activity without connecting it directly to revenue outcomes. A further 21 per cent admit they lack consistent measurement infrastructure altogether. At the same time, 86 per cent of leadership teams are demanding stronger proof of return on investment, leaving chief marketing officers under increasing scrutiny.

One of the biggest obstacles is understanding the true cost of AI. The report found that 67 per cent of organisations cannot accurately determine their total AI expenditure, while 79 per cent rely on approximations rather than precise calculations.

Cost fragmentation emerged as the leading challenge, with 62 per cent of respondents saying AI-related expenses are spread across cloud infrastructure, talent, data management and third-party vendors. Revenue attribution remains another major hurdle. Around 58 per cent said AI influences too many customer touchpoints to accurately isolate its impact, while 55 per cent reported difficulties linking customer experience improvements to commercial outcomes. Half of those surveyed also pointed to governance and integration issues that hamper consistent measurement.

Commenting on the findings, Comviva chief executive officer Rajesh Chandiramani said, “AI is rapidly moving from experimentation to enterprise-wide adoption, and the industry is entering a phase where accountability and outcomes will define success.”

He added that organisations will increasingly focus on tying AI investments directly to business metrics such as revenue growth, customer lifetime value and operational efficiency, with robust measurement frameworks becoming critical to long-term success.

Despite the challenges, the report identifies several areas where AI is already producing tangible returns. Customer segmentation and targeting topped the list, cited by 57 per cent of respondents, followed by campaign automation and optimisation at 43 per cent. Predictive personalisation and recommendation engines were highlighted by 41 per cent as key contributors to stronger customer engagement.

Meanwhile, pricing and offer optimisation was cited by 39 per cent of respondents, while 36 per cent pointed to demand forecasting as a valuable application delivering measurable business benefits.

The report also sheds light on the hidden costs that often distort AI return calculations. While organisations generally track software, API and cloud infrastructure expenses, many overlook talent acquisition, training and integration costs. As a result, total AI investments may be underestimated by as much as 30 per cent to 50 per cent, potentially creating an inflated picture of returns.

Even promising AI projects frequently struggle to scale. More than half of respondents said they face challenges defining deployment timelines, delaying time-to-value. Others cited explainability, trust and governance concerns as barriers to broader adoption.

The findings suggest that the next phase of AI adoption will be less about experimentation and more about accountability. As organisations move from pilot projects to enterprise-wide deployments, the winners may not be those investing the most in AI, but those best equipped to measure what it actually delivers.