Long-form essays on the empirical mechanisms of product, marketing, and platform economics.

Every essay here is written to be useful to a practitioner trying to act on the topic — not to be comprehensive in the encyclopedic sense. Five-thousand-word average length is not a target; it is the length at which these particular ideas hold their shape. Shorter and the hard questions get waved away; longer and the argument loses discipline.

The publication operates on three commitments. Every claim with a number points to a primary source or to data shared by an industry partner. Every counter-intuitive finding is either replicated in the public literature or cross-checked against partner data. Every essay is updated when the evidence moves, with the revision history left visible.

On the data

Some of the figures in these essays come from the published academic record and are cited inline. Many come from a different source: operating-company data shared with me under non-disclosure agreements while I was advising or working inside those organizations. That second category is why you will see numbers like conversion lifts, churn curves, or attribution deltas attributed to “a marketplace,” “a subscription SaaS,” or “a Series-B fintech” — not to a named company.

This anonymization is a condition of the access, not a rhetorical device. The alternative — declining to write about the lived work of the past decade — would leave the essays stripped of the empirical grounding that distinguishes them from the thousands of think-pieces that recycle the same half-dozen public case studies. Where feasible, aggregate distributions, ranges, and methodologies are described in enough detail that a practitioner with access to similar data can replicate the result. Where that would re-identify a partner, it stops at the narrowest claim the evidence supports.

Datasets listed in the data catalog follow the same rule. Each entry describes how the data was collected, the sample window, the variable list, and the licensing constraints. The raw rows themselves remain with the originating organization; what is published here is aggregation, summary statistics, and methodology — enough to audit the argument, not enough to re-identify the source.

What this publication is

A continuously updated archive of 51 long-form essays on the empirical mechanisms of product, marketing, and platform economics. The archive is organized into six departments covering the full stack — from the behavioral substrate of consumer choice to the machine-learning infrastructure that runs modern commerce.

Underlying the essays are three supporting artifacts, each independently useful. A glossary of 62 canonical definitions — 40 to 60 words each, written to be the reference you can send to a colleague when they ask what a term actually means. A catalog of 17 anonymized datasets — aggregated from partner companies under NDA and published with methodology, sample size, licensing note, and a machine-readable JSON endpoint. And a set of side-by-side comparisons for the decisions that recur: Bayesian versus frequentist experimentation, MMM versus MTA, Cox proportional hazards versus deep recurrent survival models.

What this publication is not

It is not a newsletter. There is no weekly cadence, because the essays take the time they take. It is not a consulting pitch — nothing here is gated, nothing leads to a demo request, nothing is written to generate calls. It is not a collection of opinions. Where I have opinions, they are clearly marked as such. Where I have evidence, the evidence leads.

Nor is it an attempt to be comprehensive across the entire marketing and analytics universe. Six departments, narrowly scoped. A platform economics essay will not teach you Salesforce administration; an attribution essay will not explain what Google Ads is. Prior familiarity with the adjacent literature is assumed, because the essays are written for people who already work in these fields and want more depth, not a different starting point.

On the voice

The voice is first-person plural in research mode, first-person singular when a claim rests on my own observation. I tell you when I have seen something directly, when I am citing others, and when I am speculating. This is not a stylistic choice; it is the only honest way to write about a field where most published numbers are someone else's marketing.

On the method

Every essay begins with a question that matters to a practitioner. The question is interrogated against the published literature, against proprietary data where it exists, and against the counter-positions. The conclusion is the narrowest defensible claim the evidence supports — not the most exciting claim available.

This sounds obvious and is routinely violated. Entire categories of business writing optimize for virality over accuracy, or for the author's positioning over the reader's decision. Product Philosophy is a bet that an audience exists for the opposite trade-off.

About the editor

Murat Ova, founder of Product Philosophy. I spent the past decade building and advising growth, analytics, and machine-learning systems at marketplace and SaaS companies. The essays here are the documents I wished existed while doing that work. The short bio is on LinkedIn.

If this is your first essay

The three essays most commonly read first, by domain:

• Product-Market Fit, Quantified — the retention-curve test that replaces “you'll know it when you feel it.”
• Bayesian A/B Testing in Practice — how to stop experiments early without inflating the error rate.
• Multi-Touch Attribution and the Causal Inference DAG — why every MTA system overstates ROAS, and what to replace it with.

If none of those are in your domain, the full archive is the better starting point.