The Founder’s Filter Stack: A Hierarchy for What to Ignore, Skim, Read, and Act On

The founder’s information problem is not a shortage—it is a flood with no drain. The practical answer is a filter stack: a personal hierarchy that routes every incoming signal to one of four dispositions—ignore, skim, read, or act—before it ever consumes cognitive bandwidth. Build it deliberately and your decisions improve; leave it to chance and the feed decides for you.

Why the feed is a decision tax

In 1971, Herbert Simon observed something that has only grown more consequential with time. “A wealth of information creates a poverty of attention,” he wrote, “and a need to allocate that attention efficiently among the overabundance of information sources that might consume it.”¹ Simon was describing the structural condition of every founder operating today: information is no longer scarce, but the capacity to process it is. Attention is the constrained resource.

The numbers make this concrete. According to the McKinsey Global Institute’s The Social Economy report, knowledge workers spend 28% of the workweek on email alone—roughly 11 hours—and a further 20% searching for and gathering information.² More recent data from Microsoft’s 2025 Work Trend Index finds that employees now experience interruptions every two minutes during core work hours.³ For a founder who is simultaneously the chief decision-maker, the lead salesperson, and the product strategist, this is not an inconvenience. It is a structural tax on the quality of every decision made after 10 a.m.

The tax compounds. Research by Gloria Mark, Chancellor’s Professor of Informatics at UC Irvine, found through nearly two decades of field study that people now spend an average of just 47 seconds on any screen before shifting attention elsewhere—down from two and a half minutes when she began measuring in 2003.⁴ Critically, it takes up to 25 minutes to return full attention to a task after an interruption.⁵ Every unfiltered notification, every ambient newsletter, every reflexive tab-open is not a small distraction. It is a 25-minute withdrawal from the account that funds your best thinking.

The cognitive science reinforces this. Information overload—defined as the condition in which incoming information exceeds working memory capacity—demonstrably damages decision quality, increases decision delay, and pushes decision-makers toward heuristic shortcuts rather than deliberate analysis.⁶ A comprehensive review published in Frontiers in Psychology in 2023 notes that cognitive load theory identifies working memory as limited to approximately seven (plus or minus two) units of information; once that threshold is crossed, performance degrades predictably.⁷ Founders who treat their inbox as a to-do list are, in effect, pre-loading their working memory with other people’s priorities before the day’s real work begins.

The decision-value hierarchy: how to classify every signal

The filter stack is built on a single classification question: does engaging with this signal increase the quality of a decision I need to make? Not “is this interesting?” Not “might this be useful someday?” Not “did someone important send it?” Decision value only. From that question, four tiers emerge.

Tier 1: Ignore

The most powerful tier is the one that requires no reading at all. Ignore is not avoidance; it is a deliberate architectural choice to prevent certain categories of signal from ever entering your cognitive system. This means unsubscribing, not just archiving. It means removing apps from the home screen, not just silencing notifications. It means building source whitelists—a small, curated set of inputs you have pre-committed to reading—so that everything outside the whitelist is structurally blocked rather than individually resisted.

The distinction matters because willpower is a poor filter. The ego depletion literature, originating with Roy Baumeister’s research and illustrated vividly in Danziger, Levav, and Avnaim-Pesso’s 2011 study of Israeli parole boards, suggests that decision-making quality degrades as cognitive resources are consumed across a session.⁸ In that study, the probability of a favorable parole ruling fell from roughly 65% at the start of a session to near zero by its end, resetting only after a food break.⁹ The mechanism remains debated, but the directional finding is consistent with a broader body of evidence: the more decisions and micro-judgments you make early in the day, the worse your consequential decisions become later. Choosing to engage with a piece of content is itself a micro-decision. The Ignore tier eliminates those micro-decisions at the source.

Practically, the Ignore tier should capture: social feeds not directly tied to customer intelligence or deal flow; general news not relevant to your specific market or regulatory environment; newsletters you have not read in the past 30 days; any channel where the signal-to-noise ratio requires more than two seconds of assessment per item. The test is not whether content might be valuable in the abstract. It is whether the expected decision value exceeds the certain cognitive cost.

Tier 2: Skim

Skim is not lazy reading. It is a disciplined protocol for extracting the headline claim of a piece of content in under 60 seconds, then making an explicit binary decision: promote to Read, or discard. The skim tier exists because some categories of signal carry genuine option value—competitive intelligence, regulatory updates, investor communications, customer feedback aggregates—but do not require deep engagement on every instance.

The structural rule for the Skim tier is time-boxing. Skim sessions happen at fixed, pre-scheduled windows—not on demand. This is the operational translation of Paul Graham’s 2009 observation that founders and operators work on a maker’s schedule, where a single unplanned interruption does not cost one hour but can fracture an entire half-day into pieces too small for deep work.¹⁰ Scheduled skim windows—say, one at the start of the day before deep work begins, and one after the primary work block ends—preserve the maker’s schedule while still processing the signal volume that a founder genuinely needs.

Tier 3: Read

The Read tier is reserved for content that has passed two gates: it is relevant to a current decision, and it requires sustained comprehension rather than headline extraction. This is a small category. Most content that feels like it belongs here actually belongs in Skim or Ignore. The discipline is in the gate-keeping, not the reading itself.

Read-tier content should be batched and consumed in dedicated blocks, away from the communication stack. The reason is neurological as much as practical. Research in cognitive neuroscience has identified the prefrontal cortex—the seat of executive function, planning, and deliberate judgment—as precisely the system most vulnerable to resource depletion under cognitive load.¹¹ Deep reading, the kind that builds mental models and generates genuine insight, requires the prefrontal cortex to be operating at capacity. That capacity is not available when the same session has already processed 40 Slack messages and three email threads. Read-tier content deserves its own protected time slot, not the gaps between other tasks.

A practical heuristic: if you cannot articulate, in one sentence, the specific decision this piece of content will inform, it does not belong in the Read tier today. It may belong there next week, when the relevant decision is live. Defer it explicitly—to a dated folder or a read-later queue with a review trigger—rather than reading it speculatively.

Tier 4: Act

The Act tier is the only tier that generates output. Everything else in the filter stack exists to protect the quality and timing of what reaches here. Act-tier items are signals that require a decision, a response, or a change in direction within a defined time window. The critical discipline is that Act-tier items must be processed when cognitive resources are at their peak—not when they arrive.

The inbox is not a decision queue. It is a staging area. The distinction is architectural. An inbox managed on the sender’s timeline means your highest-stakes decisions are made at whatever hour the email arrived, in whatever cognitive state you happen to be in. An Act tier managed on your timeline means consequential decisions are batched to your peak cognitive window—typically the first two to three hours of the working day, before the interruption load accumulates.

Building the stack: structural defaults over willpower

The filter stack fails when it is implemented as a set of intentions rather than a set of structures. Intentions require willpower to execute. Structures execute automatically. The practical architecture has four components.

Source whitelists, not blacklists. Rather than trying to block bad sources reactively, define a small, explicit set of sources you have pre-committed to reading. Everything outside the whitelist is ignored by default. This inverts the cognitive burden: instead of deciding what to block, you have already decided what to allow.

Scheduled intake windows. Communication tools—email, Slack, LinkedIn, industry newsletters—are checked at fixed times, not continuously. The McKinsey data showing that knowledge workers spend 28% of the workweek on email is not an argument for faster email processing.¹² It is an argument for radical consolidation of when email is processed. Two or three fixed windows per day, each time-boxed to 30 minutes, replace the ambient monitoring that fragments the maker’s schedule.

Action-only inboxes. The Act tier should have its own physical or digital container, separate from the general inbox. Items are moved there deliberately, not because they arrived. This separation makes the Act queue visible and finite, which is itself a cognitive relief: a bounded list of decisions is less depleting than an unbounded stream of potential ones.

Weekly filter audits. Once a week, spend ten minutes reviewing what reached the Act tier. Ask: which sources generated genuine decision value? Which generated noise that passed through the skim filter? Adjust the whitelist accordingly. The filter stack is not a one-time configuration; it is a system that improves through iteration.

The compounding return on protected attention

The case for the filter stack is not merely defensive—it is not only about avoiding bad decisions made under cognitive load. It is also about the compounding return on protected attention. Research in cognitive load theory distinguishes between extraneous cognitive load (load generated by poorly designed information environments) and germane cognitive load (the productive load that builds mental models and genuine expertise).¹³ The filter stack does not reduce total cognitive engagement; it shifts the composition of that engagement from extraneous to germane. Founders who protect their attention are not reading less—they are reading better, and building the kind of pattern recognition that compounds into judgment over time.

This is the deeper argument for founder agency in information management. The feed is not neutral. Algorithmic curation is explicitly designed to maximize time-on-platform, not decision quality.¹⁴ Every platform that serves you content has an objective function that is orthogonal to yours. The filter stack is the mechanism by which you substitute your own objective function—decisions, not engagement—for the platform’s. It is, in the most literal sense, an act of agency.

What this means

The founders who will make the best decisions over the next decade are not the ones with the fastest internet connections or the most comprehensive news feeds. They are the ones who have built systems that protect their attention as a finite, strategic resource—and who have the discipline to let most of the world’s information pass by unread. The filter stack is not a tool for staying informed. It is a tool for staying sharp. In a market where everyone has access to roughly the same information, the competitive advantage belongs to the person who processes less of it, better.

At Business Growth Accelerator (a FounderWise brand), the operators and founders we work with consistently identify information overload as one of the most underestimated drains on decision quality. Building a filter stack is one of the first structural changes we recommend—not because it saves time, but because it protects the cognitive conditions under which good judgment is actually possible.