Output Versus Outcome Based Model

by

During the recently concluded NASSCOM Technology & Leadership Forum 2025 conclave held in Mumbai, the OGs of Indian IT industry noted that the time is ripe for the industry to pivot to Output-based Model.

I shared this article on LinkedIn (and X fka Twitter) with the following comment:

“Output-based Model” is correct. Kudos to Indian IT Industry OGs for making the subtle but critical distinction between output-based and outcome-based model.

In response, a couple of people reached out to me asking what’s the difference between output-based model and outcome-based model.

I thought that many others might also have the same question and decided to write this post comparing and contrasting these two engagement models.

Let’s get on with it.

Here’s a working definition of “output” and “outcome” in the context of ERP as an example:

Output: I deliver ERP, you use it to reduce your DSO.

Outcome: I deliver a reduction in your DSO by using ERP.

where:

  • “I” means vendor
  • “You” means customer
  • ERP means Enterprise Resource Planning software, which supports invoicing, AR, collections, dunning, among other features, and tight integration between them
  • DSO means Days of Sales Outstanding i.e. the average time to collect payments from customers. In B2B technology business, DSO is typically in the 0-45 days range

Output is a vendor’s product whereas outcome is a business process of the customer organization.

In output-based model, the vendor charges a certain amount of money (typically a flat fee) for the output and Total Contract Value (TCV) is knowable in advance.

In outcome-based model, the vendor charges a certain amount of money (flat or ad valorem fee) for the outcome and Total Contract Value can only be computed ex post facto.

IT products business traditionally operates on output-based model. In the case of onprem / COTS software, the fees is typically a onetime fixed fee for perpetual license of the software or recurring annual subscription fee. In the case of SAAS software, the fees is typically a monthly subscription fee.

IT services business is typified by the T&M model:

Time-and-Material: I supply consultants matching your specifications, you manage them to get the output you want.

In T&M, the vendor contracts for various resources along with their hourly billing rates, logs the timesheet for the number of hours worked by the resources in the month (typically 160 per month), bills the total person hours supplied during the month (typically 160 times number of resources), and raises an invoice for the total amount (total person hours times hourly billing rate).

Keeping up with the times, suppose the vendor starts using Cursor, Claude Code, GitHub CoPilot, and other AI coding assistants. Going by early success stories of these tools, the vendor can generate the same output in a fraction of time as human coders. So, billing for 160 hours a month will become untenable. The only way the vendor can get the same amount of money is to change the model from T&M to output-based model where the vendor tells the customer “I’m giving you the same amount of code as before, so you pay me the same amount of money as before”. This is what the OGs of the Indian IT services industry are saying and it makes eminent sense.

ILLUSTRATION: WSJ

On a side note, IT Services is not the only one that will need to make this switch in engagement model. Others include law firms, accounting firms, ad agencies and other Professional Services industries.


As Wall Street Journal notes in AI Saves Ad Agencies a Lot of Time. Should They Still Charge by the Hour?:

Some ad agencies are trying to move to new compensation models that focus on the scope of work they perform, not the hours it takes for them to do it.

Moving from charging for hours to the work product is the essence of the pivot from T&M to output-based model.

To some extent, it’s not such a big pivot for the IT Services industry. While the industry is typified by T&M, many IT services firms – including two of my old companies – have been generating 20-25% of their revenue from fixed price projects for 15+ years. The real pivot would be to generate 20-25% of their profits from fixed price projects (IYKYK!).

A lot of people are talking about AI driving a change to outcome-based model.

I’m not so sure.

Customers who want outcomes don’t deal with technology providers. They buy them from managed services providers. Since times immemorial, managed services providers have operated on outcome-based model e.g. payroll processors charge per payslip processed and payment processors charge per payment processed. This model has been in vogue in this industry well before AI entered the scene.

You might think that the $2 price of AgentForce is an example of outcome-based model. But it’s not. While it differs from the traditional seat-based model of CRM and consumption-based model of Data Cloud products of Salesforce, the $2 price for AgentForce is for a conversation, which is closer to the output of AgentForce rather than outcome achieved by its customer (resolved ticket). Ergo, this is output-based, not outcome-based, model. See this this Reddit thread for more details from actual users of AgentForce.

According to me, output-versus-outcome is driven more by the nature of industry than technology. If someone predicts that AI will drive technology providers to pivot to managed services providers, I’ll point them to the vastly different P/E or P/S valuation ratios accorded to AI technology startups versus managed services providers. AI technology providers get $10B valuation with zero revenue. A managed services provider will get zero valuation if it didn’t make revenues.

tl;dr: No AI technology provider in his right senses will pivot to a managed services.

Some people seem to believe that customers prefer outcome-based models, while vendors often push for output-based models.

This is so not true.

The mainstream market does not buy technology for the sake of technology but to leverage technology to get manifold returns in business outcomes like revenue, profits, etc. As a result, customers are loath to part with a chunk of the monetary value of the outcome in return for technology. Ergo they opt for output based model where they can get to extract unlimited value in return for a fixed cost.

Accordingly, all leading enterprise software vendors like Microsoft, Oracle, and Salesforce have traditionally operated on outcome-based model – and, for the reasons stated in the earlier section, will likely continue to do so even for their AI products.

Let’s now take a few examples of vendors who do operate on output-based model:

  1. Google Play Store: No fees for listing of apps, charges variable fee for every payment collected by the app.
  2. Swiggy: No fees for listing of restaurants, charges variable fee for every food order delivered by it.
  3. Uber: No fees for listing of drivers, charges variable fee for every taxi ride facilitated by it.
  4. Visa: No fees for enabling merchants to collect credit card payments, charges variable fee for every credit card payment successfully processed by the merchant.

If customers really preferred outcome-based model, they should be jumping at joy with the above vendors. But, in actual practice, most of them are unhappy with their fees, many of them are appealing to the government to cap their fees, and a few of them are suing these vendors for price gouging.

Closer home, I once came across a prospect (actually prospect of our customer) who was skeptical about the value of our product and kept asking for discount. Sensing that it was only a negotiation tactic, I threw the gauntlet and offered outcome-based pricing. When he realized how much more money he’d have to cough up with the pivot, he quietly switched to the output-based model and placed the order with us without seeking any discount! Click here to know more about this deal.

In my experience, the output-versus-outcome discourse is a mere negotiation tactic. Customers will demand outcome-based model but, when offered it, will reject it and opt for output-based model.

To that extent, output-based model is like security. As I remarked in my blog post titled Why Do People Make Payments Without A Password After Obsessing Over Security?,

Users want security, but only until they get it.

On a side note, I never cease to be amazed at how the very same merchant who tells Visa “I don’t care what is your infrastructure cost, I’ll pay you only when you process my credit card payment” when he enters the relationship suddenly changes his tone later to “Your infrastructure cost is the same whether you process $100 or $1000 credit card payment, why should I pay you $22 to process a $1000 payment when I pay you only $2.2 to process a $100 payment?”