SERP Axis
Insights

What we learn, we publish.

No fluff, no curiosity-gap hooks, no ghostwritten link bait. Long-form playbooks from the operators doing the work — with data, code, and templates.

FeaturedAI Engineering·Jul 2, 2026

Prefill vs decode: the real economics of LLM inference

Generating a token doesn't cost you compute — it costs you memory bandwidth. Once you understand that one sentence, KV cache, batching, GQA, and speculative decoding stop being tricks and become the obvious consequences of the hardware.

AD
Aman Dhyani — Co-founder, SERP Axis
13 min read
DatabasesJun 25, 2026

MVCC, dead tuples, and the vacuum you're scared of: how Postgres really stores your rows

A precise mental model of PostgreSQL MVCC: row versions with xmin/xmax, why UPDATE is really insert-and-mark-dead, how VACUUM reclaims space, HOT updates, index-only scans and the visibility map, why one idle transaction bloats your whole table, and the transaction-ID wraparound that can shut a database down.

Aman Dhyani
13 min read
Distributed SystemsJun 18, 2026

The dual-write problem: why "save to the DB, then publish to the queue" is a lie

Writing to your database and your message broker in sequence is a distributed transaction in disguise — and it has no rollback. A deep look at partial-failure modes, why 'exactly-once delivery' is impossible, and how the transactional outbox and log-based CDC make it correct.

Aman Dhyani
12 min read
Data ScienceJul 1, 2026

Why your A/B test is lying to you: peeking, alpha inflation, and always-valid inference

Most teams run fixed-horizon significance tests but monitor them continuously and stop early — silently inflating Type I error far above the nominal 5%. A rigorous look at why peeking breaks p-values, why 'Bayesian is immune' is a myth, and the sequential / always-valid methods (and CUPED, SRM checks) that make continuous monitoring correct.

Abhishek Aggarwal
12 min read
Engineering2026-06-28

WooCommerce vs Shopify in 2026: an honest decision guide for founders and operators

A practitioner's WooCommerce vs Shopify comparison for 2026: ownership and control, true total cost (platform + apps + dev), performance, Black Friday scalability, content+commerce vs pure commerce, B2B and subscriptions, headless options (Hydrogen vs headless WordPress), migration, and exactly when each platform is the right call.

SERP Axis Engineering
17 min read
Engineering2026-06-28

Sanity vs Contentful vs Payload: choosing a headless CMS in 2026

Sanity, Contentful, and Payload are the three headless CMS choices most teams actually shortlist in 2026. This is the honest comparison: GROQ vs GraphQL/REST, schema-as-code vs UI, localization depth, preview, self-host vs SaaS economics, governance, and a decision flow grounded in production deployments.

SERP Axis Engineering
15 min read
Marketing2026-06-28

Paid search vs paid social: where the next dollar should go

Paid search harvests existing intent; paid social manufactures demand. They sit at different points in the funnel, measure differently, and scale differently. This is a practitioner's framework for deciding where the marginal dollar goes — including a decision table, budget heuristics, and how paid fits with SEO/GEO and lifecycle.

SERP Axis Growth
15 min read
Marketing2026-06-28

CRO that actually moves revenue (not button colors)

Most CRO programs run button-color tests and report "wins" that don't survive contact with the P&L. Here's what actually moves revenue: the high-leverage test areas, ICE/PIE prioritization, sample-size and statistical honesty, qualitative plus quantitative inputs, and how CRO compounds with paid and SEO.

SERP Axis Growth
15 min read
Marketing2026-06-28

Lifecycle email that converts (and doesn't read like spam)

Most brands send broadcasts and call it email marketing. The money is in lifecycle: welcome, onboarding, abandonment, post-purchase, and winback flows that trigger off behavior. Here's how to build them, segment them, keep them in the inbox (SPF/DKIM/DMARC, list hygiene, warmup), and measure them in revenue per recipient — plus how lifecycle compounds with CRO and paid.

SERP Axis Growth
15 min read
GEOApr 22, 2026

Generative Engine Optimization (GEO) in 2026: the complete playbook

AI Overviews now appear on well over a quarter of US informational queries, and a growing share of buying journeys begin in a generative engine. Here's the engineering playbook to be cited — not bypassed — across every major LLM-powered search surface.

Aman Dhyani
18 min read
SEOApr 4, 2026

Programmatic SEO without spam: the 12-step QA gate that survives Helpful Content Updates

Most programmatic SEO gets deindexed in 90 days. This is the engineering + content QA gate designed to keep templated pages indexed through Helpful Content Updates and core updates.

Aman Dhyani
22 min read
Technical SEOFeb 28, 2026

Core Web Vitals in 2026: the INP edition (and what actually moves rankings)

INP under 200ms, LCP under 2.5s, CLS under 0.1. The thresholds are simple. The engineering is not. The exact patterns we use to take a 4s LCP / 540ms INP site to 1.4s / 84ms in 6 weeks.

Aman Dhyani
16 min read
Data & BIMar 22, 2026

Power BI anti-patterns: 7 silent bugs that cost you CFO trust

Bidirectional relationships everywhere. SUMX over fact tables without filtering. Auto-generated date hierarchies. The silent Power BI bugs that work for two months, then don't — with the DAX fixes that actually hold up under scrutiny.

Aman Dhyani
20 min read
StrategyMar 14, 2026

The revenue model we attach to every audit (template + assumptions)

Most agencies sell SEO on traffic estimates. CFOs don't fund traffic. They fund revenue with a confidence interval. Here's the dollar-weighted opportunity model we attach to every audit, and the assumptions that survive board-level pushback.

Aman Dhyani
14 min read
EngineeringApr 14, 2026

Headless CMS in 2026: Sanity vs Contentful vs Payload vs Strapi vs Storyblok

The headless CMS comparison that doesn't cherry-pick its examples. Real engineering, content modeling, and editor experience tradeoffs across Sanity, Contentful, Payload, Strapi, and Storyblok — based on production deployments in 2024–2026.

Aman Dhyani
19 min read
AI EngineeringMar 30, 2026

Production RAG: chunking, reranking, evals, and cost (a field guide)

RAG (Retrieval-Augmented Generation) sounds simple. Production deployments fail on chunking strategy, embedding model choice, retrieval ranking, and citation accuracy. The patterns that work, the patterns that don't, and the eval discipline that prevents regressions.

Aman Dhyani
24 min read
EngineeringApr 2, 2026

Shopify performance: from 3.4s LCP to 1.4s in 6 weeks (the engineering playbook)

Slow Shopify stores cost ranking and AOV. The 'install a speed app' fix doesn't work. Here's the 6-week engineering playbook: app audit, theme architecture, image strategy, third-party tax, Liquid optimization, and CWV monitoring.

Aman Dhyani
16 min read
EngineeringMar 28, 2026

Shipping vs. shipping fast: a senior engineer's three-variable heuristic

Move-fast culture optimized for one variable and pretended the other two didn't exist. Here's the three-variable model — velocity, reversibility, observability — and the DORA-aligned framework we use to calibrate it for any project.

Aman Dhyani
12 min read
OperatingFeb 12, 2026

11 agency anti-patterns we refuse to participate in

Agency relationships break for predictable reasons. Eleven anti-patterns we've watched destroy budgets and trust — and the contract clauses that protect you from each one.

Aman Dhyani
13 min read
More long-form pieces shipping every two weeks. Subscribe at the footer for new-article emails.
Free 48-hour audit · no lock-in

The cost of waiting
is your competitor.

Every 90 days you delay is 90 days of authority compounding for someone else. Get the audit. See the math. Then decide.

Claim a free $2,400 auditTalk to a strategist
No lock-in
Weekly invoicing
Reply within
3 hours
Audit value
$2,400 yours, free