siift
Back to Blog
Evidence First Innovation for Founders: Four Repeatable Moves
SS

Author

Samim Safaei

Founder @ siift ~ 5x entrepreneur with >10 years of startup experience as a CEO, Product Leader & Engineer.

Connect on LinkedIn

Evidence First Innovation for Founders: Four Repeatable Moves

Evidence first innovation for founders: OECD and ISO definitions, peer reviewed test methods, and a four step checklist to validate and ship ideas.

Innovation moving from idea to implementation

Innovation is a new or improved product, process, or business model that has actually been put into use and creates measurable value, not just a clever idea sitting in a notebook. That distinction between invention and implementation comes straight from the standards bodies that measure this stuff for a living. This guide walks through the types, the processes, the metrics, and the practical steps you can run this week.

TL;DR:

  • Innovation requires actual implementation and market adoption, not just a new or improved idea that remains in development or concept stage.
  • Different types of innovation, such as product, process, business model, or service, have specific risks, costs, and timelines that influence the chosen validation approach.
  • Structured, hypothesis-driven testing methods like landing pages or concierge MVPs are most effective in high-uncertainty environments and should be guided by clear kill criteria set before testing.
  • Measuring real diffusion and outcomes, including customer adoption, revenue impact, or cost reduction, provides a more accurate picture of innovation success than activity metrics alone.
  • Founders should focus on rapid, inexpensive validation of the riskiest assumptions as a habit, rather than over-investing in broad brainstorming or unstructured experimentation.
siift
Turn Evidence Into Business Traction
siift guides you through ideation, validation and go-to-market, helping turn promising ideas into viable businesses with greater clarity.
Explore siift

Table of Contents

What do the Oslo Manual and ISO actually say innovation is?

Most people use “innovation” to mean “cool new idea.” The people who actually measure innovation for a living disagree, and their definition is the one that matters if you want to talk about business value instead of vibes.

The Oslo Manual, now in its fourth edition and used to run innovation surveys across more than 80 countries, defines business innovation as a new or improved product or process that differs significantly from what the company previously offered and has been introduced to the market or brought into use. Read that last part again: brought into use. A prototype in a drawer is not innovation. A patent nobody licenses is not innovation. A pitch deck full of “disruptive” language is definitely not innovation. The Oslo Manual’s whole framework exists to separate the idea from the shipped thing, because economies (and investors) don’t grow from ideas, they grow from adoption.

ISO 56002 takes a complementary angle: instead of defining the output, it defines the system that produces it. ISO’s innovation-management guidance treats innovation as a repeatable management discipline built on a Plan-Do-Check-Act cycle, with principles like realized value, future-focused leadership, and deliberate management of uncertainty. In other words, ISO doesn’t ask “was this novel?” It asks “did you build a system that reliably turns uncertainty into value?” That’s a founder-friendly framing, because it means innovation isn’t a lightning strike of genius. It’s a process you can install and improve.

Here’s how the two ideas fit together, and where popular usage gets it wrong:

  • Invention is a new idea, technology, or design; it requires no market impact and no adoption.
  • Innovation requires implementation: the new or improved thing has to actually be introduced and used, per the Oslo Manual’s standard.
  • Popular usage conflates the two, calling any novel-sounding idea “innovative” even if it never ships.
  • ISO’s contribution is process, not output: a system for managing the uncertainty between idea and adoption.
  • The practical takeaway for founders is that “innovative” is not a compliment you give yourself. It’s a status you earn once real customers are using the thing.

None of this is academic hairsplitting. If you’re building a startup, this definition is your accountability check. Did you ship something, or did you just brainstorm something? Founders who internalize the implementation requirement stop mistaking activity for progress, and that single mental shift saves a lot of runway.

What types of innovation should founders know?

Not every good idea needs the same playbook. Knowing which category your idea falls into tells you how much it will cost, how long it will take, and how much risk you’re actually signing up for.

Product innovation is a new or meaningfully improved good or service, the kind customers can touch or use directly. Think of a new feature in your app or a redesigned physical product.

Process innovation changes how you make or deliver something, not what you deliver. Automating your onboarding flow so it takes five minutes instead of two days is process innovation even if the product itself doesn’t change.

Business-model innovation changes how you capture value, not just how you create it. Moving from a one-time purchase to a subscription, or from direct sales to a marketplace model, falls here.

Service innovation covers new or improved ways of delivering a service experience, like concierge onboarding or asynchronous customer support built around chat instead of calls.

Organizational innovation changes internal structures, like how teams are organized, how decisions get made, or how knowledge moves across a company. It rarely shows up in a product demo, but it can be the reason a smaller team consistently outpaces a bigger one.

Layered on top of all five types is a second axis: incremental versus radical (or disruptive) innovation. Incremental innovation improves what already exists, faster checkout, better onboarding, a cleaner UI. It’s lower risk, cheaper to test, and it’s how most companies actually grow year over year. Radical or disruptive innovation creates something categorically new, or serves a market in a way that makes the old approach obsolete. It’s higher risk, higher cost, and far less predictable, which is exactly why most successful companies run a portfolio of small incremental bets alongside a much smaller number of radical ones.

Five innovation types across risk spectrum

Worth noting: innovation isn’t only an economic activity. Public-sector and nonprofit innovation, like a new way to deliver a government service or a new delivery model for humanitarian aid, still counts as innovation as long as it’s implemented and creates measurable value for the people it serves, even when that value isn’t measured in dollars.

Pro Tip: Before you build anything, name which type of innovation you’re actually attempting. A “new pricing model” is business-model innovation, not product innovation, and testing it requires a completely different experiment.

Which process actually turns an idea into a shipped product?

There’s no single “correct” innovation process, but there are a handful of models that show up again and again because they work under specific conditions. Picking the wrong one for your situation is one of the most common ways good ideas die slow deaths.

  1. Design thinking starts with deep empathy for the user, then moves through ideation, prototyping, and testing in tight loops. It’s strong when the problem itself is fuzzy and you need to understand human behavior before you can define a solution.
  2. Stage-gate breaks development into defined phases (concept, feasibility, development, testing, launch) with a formal go/no-go review at each gate. It fits regulated industries, hardware, or anything where mistakes are expensive to unwind.
  3. Lean startup treats a new venture as a series of falsifiable hypotheses, tested through a build-measure-learn loop with the smallest viable experiment at each step. It’s built for high uncertainty, low cost of iteration, and speed.
  4. Open innovation brings external partners, customers, or even competitors into the development process instead of keeping it entirely in-house. It works well when your team lacks a capability internally or when the market moves faster than you can hire.

The real decision isn’t “which process is best,” it’s which one matches your constraints. Structured approaches like stage-gate reduce the risk of a costly, irreversible mistake, but they’re slow and can suffocate a small team in process overhead. Iterative approaches like lean startup and design thinking move fast and cheap, but they can generate noisy signals if you’re not disciplined about what counts as evidence.

A useful heuristic: the higher your cost of change (regulatory approval, manufacturing tooling, safety certification), the more structure you need. The higher your uncertainty about what customers actually want, the more you should lean on cheap, fast experiments before committing to any of it. Most early-stage founders are operating in high-uncertainty, low-cost-of-change territory, which is exactly why lean, hypothesis-driven testing tends to be the default starting point, a theme we’ll get concrete about in a few sections.

How does innovation actually create business value?

Innovation sounds nice in a mission statement. It only matters to your business when it shows up in numbers you can put in a board deck. The translation from “we innovated” to “we grew” runs through a specific set of KPIs, and vague enthusiasm doesn’t substitute for tracking them.

Revenue lift, cost reduction, adoption rate, customer satisfaction (often tracked as NPS), and time-to-market are the five outcomes that most innovation efforts are ultimately trying to move. A process innovation that cuts fulfillment time in half shows up as cost reduction. A product innovation that solves a real pain point shows up as adoption and, eventually, revenue. If your innovation effort can’t be mapped to at least one of these, you should ask hard questions about why you’re doing it.

Because not every bet pays off, mature innovation teams think in portfolio terms rather than betting everything on one idea. That means weighting investment by stage: small, cheap experiments early to kill bad ideas fast, and larger, risk-adjusted investment reserved for ideas that have already cleared a few rounds of evidence. This is the same logic a venture investor applies across a fund, just scaled down to a single company’s project list.

Ongoing customer feedback loops are one of the more reliable ways to keep that pipeline honest. Structured feedback programs are linked to substantial revenue growth when companies act on what they hear instead of just collecting it, which is the whole point of building innovation on evidence rather than opinion.

For founders who don’t have a full analytics team, a handful of starter metrics go a long way:

  • Time-to-first-test: how long from idea to first customer signal, a proxy for how fast your team actually learns.
  • Adoption or activation rate: the share of users who actually use the new thing after launch, not just sign up for it.
  • Cost-per-validated-learning: roughly, how much you spent to get a clear yes/no signal on a risky assumption.
  • Revenue or retention delta: the before/after change tied directly to the innovation, isolated from other variables where possible.
  • Time-to-market: how long from a validated concept to something customers can actually pay for.

None of these require expensive tooling. They require the discipline to write the number down before you get excited and after you get the result.

How do you actually test an idea before you build it?

This is the part most “innovation” advice skips: the actual mechanics of testing whether your idea deserves more of your time and money. The good news is that the research on this is more concrete than most startup folklore suggests.

  1. Frame a testable question. Not “will people like this?” but something falsifiable, like “will at least 10% of visitors who see this landing page enter their email?”
  2. Write a small, falsifiable hypothesis. State what you believe, what would prove you wrong, and what threshold counts as a real signal versus noise.
  3. Design the cheapest test that could produce that signal. This is almost never a fully built product.
  4. Collect the signal and be honest about what it says. A handful of enthusiastic friends is not a signal. A stranger paying money is.
  5. Decide: pivot, persevere, or scale. Make the decision before you run the test, not after you see results you like.

This sequence isn’t just founder folklore, it’s backed by actual research on how early-stage teams behave. A peer-reviewed study of 1,022 informants across 129 firms found that rapid-validity testing and hypothesis-based customer probing correlated with stronger innovation-program performance than open-ended, unconstrained idea generation. Separate Stanford research on 152 early-stage teams running an eight-week I-Corps program found that structured customer interviews motivated real idea convergence and pushed teams toward new hypotheses, though team composition (whether the team included MBAs, for instance) affected how deeply they engaged with the method. Translation: interviews work, but only when they’re built around a specific hypothesis instead of a vague “let’s go learn about our customers” mandate.

Wharton’s executive-education research backs this up from the other direction. It found that brainstorming alone doesn’t produce breakthrough innovation. What does: framing the right question, disciplined testing of assumptions, and following through to actual implementation, while also watching for experiment-design boundary conditions that can quietly poison your results.

For low-cost tests, a few formats show up again and again in this research and in practice:

  • A concierge MVP, where you manually deliver the service behind the scenes before automating anything.
  • A landing page with real preorders, which tests willingness to pay, not just willingness to click.
  • A simple clickable prototype, useful for testing usability and comprehension before a single line of production code gets written.

Each fits a different kind of risk. Use a concierge MVP when you’re unsure whether people even want the outcome. Use a landing page when you’re unsure whether they’ll pay for it. Use a prototype when you’re confident about demand but unsure whether your specific design communicates the value.

Pro Tip: Write down your kill criteria before you launch the test, not after you see the data. It’s the only reliable defense against talking yourself into a bad idea because you already fell in love with it.

If you want a deeper walkthrough of running these tests in practice, our guide on validating a business idea step by step covers the tactical templates founders reuse most often.

What should you actually measure and report?

The Oslo Manual’s own framework for data collection is a useful blueprint for internal reporting too: track inputs, activities, outputs, use or diffusion, and outcomes, in that order, because skipping straight to outcomes is how teams end up gaming vanity metrics.

  • Inputs: time, budget, and headcount committed to the effort, tracked before you get excited about results.
  • Activities: the actual tests, interviews, or builds completed in a given period.
  • Outputs: what got shipped, even in prototype form.
  • Use or diffusion: whether people are actually using it, and how many.
  • Outcomes: the business result, revenue, cost, retention, tied back to the specific innovation.

A simple monthly dashboard with these five rows, reviewed on a predictable cadence, beats a quarterly innovation report full of anecdotes almost every time. The trap most teams fall into is measuring activity (how many experiments we ran) instead of diffusion and outcomes (whether anyone is actually using what we built and whether it moved a number that matters). Activity metrics feel productive. They just don’t pay rent. For a more detailed walkthrough of validation metrics specifically, see our full guide to early startup validation.

When does experimentation stop working?

Rapid testing is powerful, but it’s not universally the right move, and pretending otherwise causes real damage. When feedback signals are noisy (small samples, biased respondents, leading questions), teams draw false confidence from tests that were never designed to produce a clear answer. When pivots are expensive (regulatory refiling, manufacturing retooling, contractual lock-in), the cost of being wrong outweighs the speed benefit of testing fast, and a more structured, stage-gate approach earns its overhead. Over-testing is its own failure mode too: some teams run experiment after experiment as a way to avoid the harder work of actually committing to a direction.

Organizational barriers compound all of this. A culture that punishes failed experiments quietly trains people to stop reporting bad news, which poisons your data before you even see it. Under-resourced innovation efforts (a side project bolted onto someone’s full-time job) rarely produce clean signals, because nobody has the time to run the test properly. And weak governance, no clear decision-maker, no defined kill criteria, means “we’re still testing” becomes a permanent state instead of a phase.

The decision rule that cuts through most of this: if adaptation costs are high and your signal quality is low, skip the endless testing loop and either commit to direct entry or find a cheaper, correlated proxy to test instead.

A founder’s checklist for putting this into practice

Strip away the theory and the whole innovation process comes down to four repeatable moves: define the riskiest assumption behind your idea, design the cheapest test that could prove it wrong, run it and interpret the result honestly, then map what you learned onto your next go-to-market step.

  • Define the risky assumption first, not the feature list. What has to be true for this to work?
  • Design the cheapest test that could produce a real signal, not the most impressive-looking one.
  • Run it, then interpret the result against the threshold you set before you started, not after.
  • Map validated learning to go-to-market, so testing has an actual destination instead of becoming a loop.

This is exactly the workflow an agentic AI platform can support: guiding founders through ideation, validation, and go-to-market planning as one connected sequence instead of three disconnected tools. Where generic AI chat tools give you generic answers, such platforms can map specific business context, flag blind spots in assumptions, and turn validated learning directly into actionable strategy. If you’re at the stage of testing what to build, our startup idea validation tools are built around this exact checklist.

What actually matters if you’re trying to innovate on purpose

Here’s my honest take after all the frameworks and citations: most founders don’t fail because they lack ideas, they fail because they never build the muscle to test an idea cheaply before falling in love with it. The research is consistent on this point. Structured, hypothesis-driven testing beats open-ended brainstorming, and the teams that win aren’t the ones with the flashiest idea, they’re the ones with the tightest feedback loop.

If I had to allocate a founder’s limited time and budget, I’d put the bulk of it into cheap, fast, falsifiable tests on your riskiest assumption, not your favorite one. Save structured, expensive processes for the moment you actually have evidence worth protecting. Innovation isn’t a personality trait. It’s a habit of testing before you commit, and it’s learnable by anyone willing to be wrong quickly and cheaply.

— Samim Safaei

Where to go for deeper reading

If you want to go past this article, these are the sources worth your time. The Oslo Manual is the measurement baseline nearly every serious innovation statistic traces back to, so it’s worth reading even just for its definitions section. ISO’s 56000/56002 family is the practical management-system counterpart, useful if you’re trying to build repeatable process rather than one-off wins.

Sources

FAQ

What is the best definition of innovation?

The most authoritative definition comes from the Oslo Manual: a new or improved product or process that differs significantly from what came before and has actually been introduced to the market or brought into use. The key requirement is implementation, an idea alone doesn’t count until it’s adopted.

What is the innovation of meaning?

Innovation of meaning refers to changing what a product or experience signifies to people, rather than changing its physical function. It’s a less standardized concept than the Oslo Manual’s implementation-based definition, and definitions of it vary across design and management literature.

What does it mean to be innovative?

Being innovative means consistently turning new or improved ideas into things that are actually adopted and used, not just generating novel concepts. Under the Oslo Manual’s standard, an organization only earns the label once implementation and use actually happen.

What is an example of innovation?

A company that redesigns its checkout process to cut cart abandonment, then rolls that change out to every customer, is practicing process innovation under the Oslo Manual’s definition. A subscription model replacing one-time purchases is a common example of business-model innovation, since it changes how the company captures value rather than what it sells.

How can founders start fostering innovation today?

Start with a single falsifiable hypothesis about your riskiest assumption and design the cheapest possible test for it, following the sequence supported by Stanford’s lean-startup research. Tools like siift’s startup idea validation platform are built to guide founders through that exact process from ideation through go-to-market.