Embedded Insurance ROI: What Banks and NBFCs Actually See in Attach Rate and Persistency

Why ROI conversations about embedded insurance get vague

Ask most vendors about embedded insurance ROI, and the answer tends to stay at the level of "higher revenue, better retention," without landing on numbers a finance team could actually put in a board deck. That vagueness is a real barrier for banks and NBFCs deciding whether to invest, because the technology and integration costs are concrete and upfront, while the benefit is often described only in adjectives. The purpose of this piece is to walk through what the return actually looks like in numbers: where the revenue comes from, what a realistic attach rate is, how persistency changes the math over time, and what a worked calculation looks like for a lending business specifically.

The revenue model in one sentence

Embedded insurance earns commission per policy as fee income, with essentially zero balance sheet risk to the distributor, since the underwriting risk sits with the insurer, not the bank or NBFC placing the sale. This is the foundational reason the ROI case is usually favorable even before volume assumptions get involved: the distributor is not underwriting anything, is not holding reserves against claims, and is monetizing a moment that its own lending or checkout journey was already creating. The cost side is correspondingly narrow. There is a platform fee for the insurance API layer, some integration engineering time, typically front-loaded, and ongoing operational attention to the journey. Customer acquisition cost for the insurance sale itself is close to zero, because the journey generating the insurable moment, the loan, the checkout, and the account opening already exists and was already being paid for.

A worked example: digital lending

The clearest way to make this concrete is a single worked example drawn from a typical digital lending scenario. A lender processing 50,000 personal loans a month, achieving a 25% attach rate on credit-linked insurance, at an average premium of 2,400 rupees, and earning roughly 20% commission, generates approximately 7.2 crore of annualized fee income from that one lending journey alone. That figure assumes no growth in loan volume and no improvement in attach rate over the year, both of which are realistic to expect to improve once a program is live and tuned. It's worth being clear about what drives that number and what does not. Loan volume is largely fixed by the underlying lending business and not something the insurance program controls. Attach rate and average premium are the two levers the embedded insurance implementation actually influences, and they are the two numbers worth watching most closely once a program launches.

What attach rate actually looks like in practice

Attach rate is context-dependent, and the range is wide enough that a single benchmark number is not that useful on its own. In lending journeys specifically, digitized and inline consent typically reaches 18 to 30%, with personal loan credit life often landing around 28% against a manual, non-integrated benchmark closer to 12%. In commerce checkout contexts, attach rates are meaningfully lower, typically 3 to 6%, reflecting that a device protection add-on at checkout is a smaller, more optional decision than credit life on a loan a customer is actively taking out. The gap between digitized and manual attach rates is the single clearest ROI argument for investing in a proper embedded integration rather than a bolt-on insurance sales process: roughly doubling attach rate on personal loans, from a manual benchmark near 12% to a digitized 28%, more than doubles the fee income from the same loan book without needing any additional loan volume at all.

Beyond the first sale: why persistency compounds the return

A single year's attach rate tells only part of the ROI story, because credit-linked and embedded policies that lapse early stop generating value, and because renewal business on standalone products compounds commission income year over year if it is retained. This is where persistency, the share of policies still active at defined anniversaries such as the 13th or 25th month, becomes as important to the ROI case as the initial attach rate. Managed renewal automation, meaning automated reminders, pre-filled quotes, and one-tap renewal flows rather than manual chase-ups, has been shown to lift 13th-month persistency by up to 12 points above typical industry baselines of 65 to 70%. Applied against a real book of business, that lift represents more retained premium meaningfully and repeat commission without any new customer acquisition at all, which is why a serious ROI evaluation of embedded insurance should include the renewals and commission management layers, not just the initial issuance flow.
There is a second, less obvious retention effect worth naming: customers holding two or more products with an institution, a loan plus an insurance policy, for example, churn measurably less than single-product customers. This means the ROI of an embedded insurance program is not fully captured by the insurance commission line alone; some of the return shows up as improved retention on the core lending or banking relationship itself.

A second reference point: bancassurance at scale

The lending example above is not the only proof point worth citing. A private-sector bank running a digitized bancassurance program reported going from zero to 40 crore of attached premium within two quarters, starting from a four-week white-label deployment, with 60 branches live in the first quarter, and reaching a full attach rate tripling by the second quarter, all without adding operations headcount. That last detail, zero added headcount, is worth sitting with because it means the ROI calculation does not need to net out a growing operations team against the revenue gain; the technology layer absorbed the additional volume.

The KPIs that actually determine whether ROI is being realized

A handful of numbers separate a program that is genuinely working from one that looks fine on a dashboard but is quietly underperforming. Attach rate on the relevant journey is the headline number, but it needs a denominator that matches the actual eligible population, not just total transactions. Thirteenth-month persistency indicates whether the initial sale is converting into retained, compounding revenue. Fee income per customer, tracked quarter over quarter, shows whether the program is actually growing or has plateaued. Claim Net Promoter Score matters more to ROI than it might initially seem, since a poor claims experience directly damages renewal willingness and referral behavior on future cross-sell. And complaint rate needs to be watched alongside attach rate specifically, because a high attach rate paired with a rising complaint rate usually indicates a program winning short-term volume by under-disclosing or over-pushing the offer, a pattern that tends to produce regulatory attention and customer churn later, not durable ROI.

What actually kills ROI in embedded insurance programs

The programs that underperform financially tend to fail in a small number of recurring ways. Offer volume gets prioritized over offer relevance, meaning insurance gets pushed into every possible journey rather than the two or three moments where it genuinely fits, which drives complaint rates up without proportionally growing attach rate. Issuance friction, forms with dozens of fields instead of reusing KYC data already on file, quietly suppresses attach rate below what the underlying demand would support. Servicing gets neglected once the sale is made, and a poor claims experience directly damages the renewal and referral behavior that persistency depends on. And compliance gets treated as paperwork rather than product design, which does not show up as an ROI problem immediately but tends to surface later as regulatory friction that halts or slows an otherwise working channel.

Building the business case

For a bank or NBFC building an internal business case, the calculation worth presenting is straightforward: eligible transaction volume, multiplied by a realistic attach rate for that specific journey type (18 to 30% for lending, 3 to 6% for commerce checkout), multiplied by average premium and commission rate, gives a first-year fee income estimate. A second, equally important line should model the persistency-driven compounding effect in year two and beyond, since a program that only measures year-one attach rate materially understates the return of a well-run renewal and retention layer. The full mechanics of how the underlying journey and integration work are covered in DeployIT's embedded insurance guide and use case detail.