Pull Don't Push The Just-in-Time Approach to Innovation

Imagine racing two young children to stuff one hundred newsletters into envelopes. The kids decide to fold all the letters first, then apply all the stamps, then seal every envelope at once. They assume this is efficient because they are repeating the same task. You decide to complete each envelope one at a time, from start to finish.

You will win the race every single time because the pull vs push lean approach favors the flow of finished work over the accumulation of inventory. Completing one unit at a time is called single-piece flow. It eliminates the extra time wasted sorting and moving large piles of half-finished work.

This method allows you to discover problems almost immediately. If the letters don't fit in the envelopes, you'll know after the first one. In a large-batch "push" system, you wouldn't realize the mistake until you had already folded one hundred useless letters. In modern business, this difference determines whether a product succeeds or becomes expensive waste.

What Is the Pull Method?

In his book The Lean Startup, Eric Ries explains that most businesses operate on a push model. They build products based on a forecast or a massive plan. They push work from one department to another, regardless of whether there's a real need for it.

This creates huge amounts of work in progress (WIP) that sits around doing nothing. In a lean system, work is only performed when it is pulled by the next stage of the process. This concept originated in the Toyota Production System, where a part is only manufactured when a signal arrives from the next station.

Applying this to innovation means you don't build features because they were on a schedule. You build them because a specific experiment or customer need pulled that work from the development team. This prevents the ultimate waste: building something perfectly that nobody actually wants.

Why Small Batches Reveal Just-in-Time Production Flaws

Large-batch thinking is the silent killer of innovation. Most managers believe that having a designer work on thirty drawings at once is efficient. They think the designer stays "busy" and doesn't get interrupted. However, if the first drawing has a technical error, that error will be replicated across all thirty before anyone notices.

By switching to just in time production, you force the team to work on one item at a time. This reveals defects in the system immediately. In the early 1900s, 501 companies were formed to manufacture automobiles, but 60 percent folded within two years because they couldn't manage their production systems effectively.

Toyota solved this by focusing on changeover time. Their engineers, led by Shigeo Shingo, reduced the time to change a machine die from hours to less than ten minutes. This allowed them to produce a high variety of cars in small batches without losing efficiency. For a startup, this means you can test different product variations without a massive upfront investment.

Using Experiments as a Kanban Pull Signal for Teams

In a factory, a kanban pull signal is a physical card that tells the previous station to send more parts. In a business context, the pull signal is the need to run an experiment. Instead of a manager pushing a feature list, a specific hypothesis pulls the work from the development team.

This keeps the total amount of work in progress low. When a team has too many projects at once, everything slows down. Each project requires meetings, handoffs, and context switching. By limiting the number of active projects, you ensure that the team finishes what they start before moving to the next task.

At IMVU, the team used this to ship an average of 50 changes to the product every single day. This rapid flow was only possible because they had a "product immune system." If a change caused a defect, the system automatically rolled it back and signaled the team to stop and fix the root cause. This prevented small mistakes from turning into catastrophic failures.

Moving Toward a Lean Supply Chain of Ideas

Managing a lean supply chain of ideas requires a shift in how you measure productivity. Traditional managers look at how many hours an employee is "busy." Lean managers look at how fast an idea moves from a hypothesis to a validated fact.

Anything that doesn't contribute to learning is considered waste. If you spend months building a feature that customers don't use, that effort is lost. Small batches allow you to test your value hypothesis before you've spent your entire budget.

Startups that work in large batches often enter a "death spiral." They delay launching because they want the product to be perfect. The longer they wait, the more work they pile on, and the more they fear customer rejection. This leads to a massive batch that is almost guaranteed to fail because it hasn't been tested against reality.

How Alphabet Energy Scaled with Small Batches

Alphabet Energy is a clean-tech startup that generates electricity from waste heat. Many clean-tech firms require hundreds of millions of dollars to build factories before they ever ship a product. For example, BrightSource raised $291 million to build large solar plants before delivering power.

Alphabet Energy took a different path by choosing silicon-based materials. This allowed them to use existing semiconductor manufacturing infrastructure. Because they didn't have to build their own factory, they could produce physical prototypes in just six weeks.

They started with a hypothesis that power plants would be their primary customers. After just three months of small-batch testing, they realized power plants were too risk-averse. They pivoted to manufacturing firms, who were eager to experiment. Because they used small batches, they achieved this learning while only raising about $1 million, a fraction of their competitors' spending.

Three Actions to Accelerate Your Feedback Loop

Transitioning to a pull system requires changing the way your team handles daily tasks. You can implement these three changes this week to start reducing waste and increasing speed.

1. Shrink your batch size to the smallest possible unit. If you're designing a new service, don't plan the whole year. Design one tiny feature, test it with five customers, and use that feedback to decide the next step immediately.

2. Implement a limit on work in progress. Decide on a maximum number of active projects your team can handle at once. Do not allow anyone to start a new project until an existing one is either finished or discarded based on data.

3. Use the Five Whys to fix systemic errors. When a mistake happens, don't blame an individual. Ask "Why?" five times to find the flaw in your process and make a proportional investment to prevent it from happening again.

Why Large Organizations Resist Pull Systems

Established companies often struggle with pull systems because they have "muscle memory" for large batches. Their financial systems are built to track annual budgets and quarterly releases. When you try to introduce a weekly or daily release cycle, it creates friction with legal, marketing, and accounting departments.

Critics often argue that small batches are inefficient because they don't allow for economies of scale. They worry that constant switching between tasks will confuse employees or damage the brand. These concerns are usually based on a misunderstanding of how lean systems manage quality.

In reality, the high cost of rework in a push system is much more expensive than the setup costs of a pull system. While it may feel slower to fix every error as it occurs, it is the only way to maintain a sustainable pace. Ignoring quality in the name of speed always results in a total system breakdown later.

Successful innovation requires a relentless focus on the Build-Measure-Learn feedback loop. The faster you can get through this cycle, the more likely you are to find a sustainable business model. Switching to a pull vs push lean system is the most effective way to eliminate waste and ensure your team's hard work results in real value. Master the discipline of small batches to move from a culture of guessing to a culture of validated learning. Begin by reducing the batch size of your next project today.