Podcast Audio Transcript
Mimi: Hello listeners, this is Mimi; thank you for tuning in to yet another exciting and informative podcast from us at Infosys BPM.
By now, most leaders agree on one thing. The real business case for AI is not about efficiency alone.
It’s not about shaving a few minutes off a process. It’s not about how many FTEs a chatbot can replace. And it’s definitely not about vanity metrics like “number of models employed.
The real promise of AI lies somewhere else — in business value, in measurable outcomes, and in impact that shows up clearly on the P&L.
To talk about all of these and more, we have Sourav with us here. Sourav Ghosh, is a Senior Industry Principal with Infosys BPM.
Welcome Sourav. How are you?
Sourav: I’m doing great Mimi. It’s nice to be back in this podcast again.
Mimi: Sourav, let’s start where we left last time. You spoke about the need to scale AI beyond pilots and why organizations need to think of business value as a metric for AI Return on Investment.
But today, AI conversations still often start with automation and efficiency. So why do you believe that’s no longer the right primary business case?
Sourav: Let’s start with a hard truth.
Efficiency in some cases is a necessary outcome of AI—but it’s a terrible primary business case.
Why?
Because efficiency, rarely creates sustained competitive advantage. Its easy for competitors to replicate and often doesn’t translate into real financial impact unless headcount is actually removed or redeployed—which most organizations are reluctant to do.
For example, many enterprises proudly say “Our AI solution improved productivity by 20%.”
But when you ask - Did revenue increase? OR did cost actually come out of the system? Or did customer lifetime value improve? The answers are often… unclear.
In contrast, the strongest AI transformations focus on value creation, not effort reduction.
Think - Revenue uplift; Margin expansion; Risk avoidance; Faster decision cycles; Better capital allocation.
AI is not a cost cutting tool. Neither it is an automation program or a technology upgrade. For example in a Bank, AI would be about better decisions at scale, not just faster processes.
Mimi: Thank you for sharing Sourav. So does that mean productivity improvements from AI don’t matter at all?
Sourav: They matter—but only as an enabler. AI often frees up capacity, but unless that capacity is redeployed toward higher value activities—like better customer engagement, smarter underwriting, or proactive risk management—the value never materializes. Productivity without business action just creates unused capacity.
Mimi: So how should organizations identify the right metrics?
Sourav: The answer is simple—but not easy. Start where the business already measures success. Let me give you a few real-world examples.
Example 1: Sales & Revenue
Instead of:
- “AI reduces sales admin time by 30%”
The right metrics are:
- Conversion rate improvement
- Deal velocity
- Average deal size
- Revenue per sales rep
- Win loss ratio
One global B2B company reframed its AI CRM initiative this way:
“If AI cannot move quarterly revenue per seller by at least 5%, it is not a successful project.”
That single statement changed:
- How the model was designed
- What data was prioritized
- And who was accountable
Example 2: Risk & Compliance
Instead of:
• “AI automates compliance checks”
Measure:
- Losses avoided
- Fraud leakage reduced
- Regulatory penalties prevented
- Insurance premiums lowered
These are P&L and balance sheet outcomes—not IT KPIs.
The rule is this:
If the metric doesn’t already exist in a business review or earnings discussion, it’s probably the wrong metric for AI.
Mimi: Why is it so important to tie AI initiatives directly to P&L and balance sheet accountability?
Sourav: AI metrics must live where business accountability already exists—the P&L.
AI success cannot sit in:
- Innovation dashboards
- Or “digital transformation” reports
That means:
- Business sponsors own the outcome
- AI benefits are baked into annual operating plans
- And delivery success is reviewed like any other business initiative
One financial services firm did something radical: Every AI project required:
- A named business sponsor
- A quantified financial outcome
- And a quarterly benefit realization review
If benefits didn’t materialize, the question wasn’t:
“Why didn’t IT deliver?”
It was - “Why didn’t the business change how it operates?” Because AI does not create value on its own. Changed decisions and behaviours do.
Mimi: Many organizations still rely heavily on IT or vendors to define AI use cases. Why is that risky?
Sourav: This is where many AI programs quietly fail.
If we ask an IT organization or worse, an AI vendor to identify use cases or define value and justify ROI, remember that IT does not own revenue, does not own margins or owns customer outcomes.
So what do they optimize for?
- Technical feasibility
- Data availability
- Model accuracy
Its necessary but not sufficient.
Business value identification must be co owned, with the business in the lead.
IT and partners should ask:
- “What decisions do you want to change?”
- “What economic lever does that decision move?”
- “What happens if we’re wrong?”
When business leaders are forced to answer those questions, AI initiatives become sharper, fewer—and far more impactful
Business leaders should own the “why AI” and the “value”—what outcome is changing and by how much. IT and partners should own the “how.” This separation ensures AI initiatives are anchored in business reality while still being technically sound.
Mimi: Let’s shift to commercial models. Why do traditional IT pricing models struggle with AI?
Sourav: Because AI value in BFSI emerges over time through improved decisions, not at go live. Paying for models built or licenses consumed ignores whether credit losses fell, fraud reduced, or customer profitability improved. Traditional models reward activity, not outcomes.
In an AI world, traditional delivery models break down.
Why? Because the biggest value from AI doesn’t come from:
- Code delivered
- Models built
- Or hours billed
It comes from realized ROI over time.
That’s why we’re seeing a growing shift toward:
- Outcome based pricing
- Benefits sharing models
- Gain share and risk share constructs
For example:
- AI partners tied to revenue uplift
- Analytics fees linked to cost reduction achieved
- Platforms paid based on value realized, not licenses consumed
This aligns incentives correctly:
- Business pushes for adoption
- Partners push for outcomes
- AI investments are treated like growth capital, not IT spend
In many ways, AI is forcing organizations to mature how they think about value itself.
Mimi: How do these models change internal organizational behaviour?
Sourav: They force discipline. Business teams focus on adoption, data quality, and decision integration. Partners focus on impact, not just algorithms. AI investments start being treated like risk weighted capital allocation decisions, not discretionary IT spend.
Mimi: That makes perfect sense!
Do you have any final words, Sourav? What thoughts would you like to leave our listeners with?
Sourav: So let’s close with this.
The future of AI will not be decided by:
- The best models
- Or the largest data sets
- Or the biggest technology budgets
It will be decided by organizations that can clearly answer three questions:
- What business outcome are we changing?
- Who owns that outcome on the P&L?
- How will we measure success in financial terms?
Mimi: Thank you so much, Sourav, for sharing your expertise on Scaling AI beyond prototypes. It's been an enlightening discussion, and our listeners will undoubtedly find these insights valuable.
Sourav: It was indeed a pleasure, Mimi.
Mimi: Dear listeners, if you enjoyed our podcast today, please don’t forget to share and like it on social media. Our social handles are mentioned in the podcast page. The podcast will be available on various platforms like Google Podcasts and Spotify, in addition to our website. Which is www.infosysbpm.com
Also, if you have any queries, do reach out to us through the email address on the podcast description. Watch this space for more exciting podcasts coming up. Once again, thank you for tuning in, Have a nice day, stay sharp and stay safe.
