Skip to content

Archive

Growth engineering, written for Ariadne Growth Systems.

A nine-stage series on how demand, conversion, capture, routing, follow-up, sales, measurement, automation, and compounding behave as one system. It belongs to the company rather than to this site, so it lives here rather than in the main writing feed.

Why this is a separate archive

This site is an AI research and engineering portfolio. These 20 articles are growth-engineering writing for Ariadne Growth Systems, a separate company. Every article keeps its original /blog/… URL so nothing that already links to it breaks; they are simply no longer part of the main feed, the RSS feed, or the navigation. For AI and research writing, see Field notes.

  1. Auditing the Systems That Audit You

    Why I named the company after a paper on causal faithfulness — and what measuring 'reasoning theater' in LLM agents taught me about trusting business automation.

  2. What a Growth System Audit Actually Looks For

    A diagnostic is a search for one binding constraint, not a list of everything wrong. Here is the instrument, the arithmetic, and the deliverable.

  3. Leave the Keys: The Economics of Not Being Locked In

    Switching costs are worth roughly a customer's entire future profit stream to the vendor holding them. The industrial-organisation literature says so precisely.

  4. Measure Honestly: Goodhart's Law Inside Your Own Dashboard

    Why the agency model produces optimised parts and an unimproved whole — traced to a formal result in contract theory, not to bad intentions.

  5. Stage 09 · Compounding: The Loop That Pays for Itself

    Retention elasticity is 3–7 while acquisition elasticity is 0.02–0.3. Also: the most mis-cited statistic in business, traced to its actual source.

  6. Keep Humans in Control: The Ironies of Automating a Sales Process

    Bainbridge's forty-year-old result on why automating the easy parts makes the human job harder — and what that implies for automated follow-up.

  7. Stage 08 · Automation: Amdahl's Law for Your Operations

    The ceiling on every automation project, why Gustafson's rebuttal matters more than the law itself, and the 95% glue-code problem.

  8. The Unfavorable Economics of Knowing If Your Ads Work

    Twenty-five large randomised advertising experiments, $2.8M of spend, and confidence intervals over 100 points wide. Why ad measurement is hard for reasons no dashboard can fix.

  9. The Experiment You Cannot Run

    What A/B testing actually costs in sample size, why most small businesses can never afford it, and the four things to do instead.

  10. Stage 07 · Measurement: Last-Click Is a Story You Tell Yourself

    Markov removal effects, Shapley attribution, and the peer-reviewed evidence that last-touch does not merely mis-measure — it changes behaviour for the worse.

  11. Stage 06 · Sales: Variance Is the Tax You Never Budgeted For

    Kingman's V term, Shewhart's two kinds of variation, and why standardising a sales process buys more speed than hiring does.

  12. Stage 05 · Follow-Up: Persistence Is a Probability, Not a Personality

    A geometric model of contact attempts, the point where extra dials turn negative, and why 'we followed up' is not a measurable claim.

  13. Stage 04 · Routing: Your Lead Queue Obeys Erlang, Not Optimism

    Little's Law, Erlang C, square-root staffing and the abandonment model — the queueing results that decide whether an inquiry ever reaches a person.

  14. Stage 03 · Capture: Every Inquiry Needs a Next Step

    Multi-channel intake is a reliability problem in disguise. Series yield, silent failure modes, and why the leak you can't see is the expensive one.

  15. Core Web Vitals Are a Revenue Variable

    The published business-impact numbers for web performance, sorted by how much you should trust them — and the one that was a genuine controlled experiment.

  16. Stage 02 · Conversion: The Page Is a Decision, Not a Design

    Response-time limits from human-factors research, what actually gets tested, and why 'most ideas fail' is the most useful conversion statistic there is.

  17. Stage 01 · Demand: Being Findable When the Searcher Is a Machine

    How retrieval actually selects sources, what the first peer-reviewed study of generative engine optimization found, and why AI search cites so few domains.

  18. The Five-Minute Window

    Everyone quotes the speed-to-lead statistic. Almost nobody quotes it correctly. Here is where each number actually comes from, and what survives scrutiny.

  19. Nine Stages, One Constraint

    Growth systems fail at one place at a time. Kingman's equation explains why that place is almost never where it feels like it is.

  20. Growth Is a System, Not an Effort

    Why revenue stalls the moment the founder stops pushing — and what fifty years of feedback-loop research says to build instead.