Compounding: systems that get better while you sleep

A compounding system is one whose output feeds back in as input, so each cycle starts from a higher base than the last. Effort compounds when it is stored in an asset — a list, a document, a domain's reputation, a body of published work — and evaporates when it is spent on activity that leaves nothing behind. The practical question for a founder is not "how do I work more?" but "what fraction of this week's work will still be working in twelve months?"

What actually makes something compound?

Compounding needs three ingredients: an asset that stores value, a loop that reinvests output into the asset, and time. Money in an interest-bearing account is the textbook case, but the structure is general. When this cycle's output increases the base for the next cycle, then growth accelerates rather than merely accumulates. That loop structure — output returning as input — is one of the core patterns in A Systems-Thinking Guide for Founders, and it is worth being strict about the definition. A queue of tasks does not compound. A pipeline of one-off projects does not compound. Ten years of invoices do not compound. They add; they do not multiply.

What compounds in a B2B service firm?

More than founders usually credit:

• A prospect database. Every campaign teaches you which segments reply and which titles convert, and that learning refines the next list. A maintained database is an appreciating asset; an unmaintained one decays as people change jobs.
• Sending reputation. Consistent, well-run cold email builds a domain history that improves inbox placement, which improves reply rates, which justifies the discipline. This is the engine behind the argument in cold email isn't dead — bad cold email is: practitioners who compound reputation get results that dabblers conclude are impossible.
• Documentation. Every process written down once is a task you never re-explain. Each document also makes the next hire faster to train, which frees more time to document.
• Published thinking. Articles, talks, a letter. Each piece keeps answering questions and generating enquiries long after the writing cost is sunk.
• Client relationships handled well. Renewals, referrals, and case-study permission all flow from the same stored trust.

Notice what is missing: hours worked, meetings attended, most social media activity, and heroic one-off saves. All real effort; none of it stored.

Why do most firms fail to compound?

Because compounding is invisible early and undeniable late, and most decision-making happens early. The first months of any compounding curve look embarrassingly flat next to buying activity directly — hiring a telemarketer, boosting a post, running a promotion. First-order thinking picks the visible spike; the compounding option only wins in the second-order chain, two or three steps out.

The other killer is interruption. Compounding is multiplicative, so a reset is not a pause — it is a return to the flat part of the curve. The firm that runs outbound for three months, stops for six, and restarts on a new domain keeps repaying the entry cost and never reaches the payoff.

How do you build a compounding loop deliberately?

The mechanism, step by step:

1. Choose an asset that stores value: a verified database, a sending domain, a process library, a letter list.
2. Attach a recurring activity that deposits into it. When a campaign finishes, then its reply data updates the segment notes. When a process is performed twice, then it gets documented.
3. Route output back to input. When replies come in, then the wording that produced them feeds the next sequence. When a document exists, then delegation happens against the document, freeing time for the next one.
4. Protect the cadence. Small and unbroken beats large and sporadic; the loop only multiplies while it runs.
5. Review quarterly, not daily. Compounding is imperceptible at daily resolution, and watching it too closely tempts you to interfere.

Where does compounding go wrong?

Two failure modes. First, compounding what is easy rather than what matters — a firm can compound followers while its delivery process stays undocumented; picking the wrong loop to feed is a version of optimising the part instead of the whole. Second, forgetting that decay compounds too. Data rots, unpatched processes drift, neglected relationships cool. Commonly cited industry figures put B2B contact data decay at something like a quarter to a third a year; whatever the true rate, the direction is not in doubt.

What does this change on Monday?

Audit a normal week and sort your hours into two columns: spent or stored. Most founders find over 80% lands in spent. You do not need to invert the ratio — moving four hours a week into stored work is enough, provided it goes into one loop and stays there. Twelve months of that is the difference between a business that resets every January and one that starts each year ahead of the last.

---

Keep the thinking coming: The Format — a fortnightly letter on business and personal systems. Subscribe by email.