S Structure Is the account architected to win before a dollar is spent?
48 checks Every ad group maps to one tight intent cluster (SKAG/STAG hybrid), not 200 keywords dumped together. Loose ad groups force one RSA to serve mismatched queries, tanking Quality Score and inflating CPCs. Campaigns are split by margin tier or funnel stage (prospecting vs. brand vs. competitor) with separate budgets. Blended budgets let cheap brand clicks starve high-intent non-brand of spend. Brand terms are isolated in their own campaign with a brand-only negative strategy on all other campaigns. Without isolation, brand steals credit and inflates blended ROAS while non-brand quietly loses. Shopping/PMax feed is segmented into product groups by margin, price band, and best-seller status via custom labels. Feeding all SKUs at one bid strategy overspends on low-margin junk and underfunds winners. A shared negative-keyword hierarchy exists (account + campaign level) and is versioned, not ad-hoc. Un-managed negatives leak spend to irrelevant queries and never get audited. Bid strategies align to a single goal per campaign (tROAS vs. tCPA vs. Max Conv), never mixed within a funnel stage. Conflicting goals confuse Smart Bidding and stall the learning phase. Geo, ad-schedule, and device structures reflect real conversion data, not defaults. Default 'all locations / all devices' hides that one segment subsidizes another. RSAs carry 15 distinct headlines / 4 descriptions with pinned brand/offer positions where required. Thin or duplicative assets cap Ad Strength and limit auction coverage.
Core flows are live: welcome, browse-abandon, cart-abandon, checkout-abandon, post-purchase, winback, sunset. Missing flows are the single biggest silent revenue leak in most DTC accounts. The list is segmented by RFM or engagement tier, not blasted to 'everyone.' Sending to unengaged contacts drags domain reputation and inbox placement. A dedicated sending subdomain is authenticated (SPF, DKIM, DMARC) separate from transactional. Sharing the corporate root domain risks reputation damage from marketing sends. SMS and email are orchestrated so one trigger doesn't double-hit a contact in the window. Uncoordinated channels feel like spam and drive opt-outs. Suppression logic auto-removes sunset non-buyers and hard bounces. Bloated lists inflate cost and depress every rate metric. Zero-party data capture (quiz, preference center) feeds segmentation beyond email+consent. Without preference data, personalization ceilings out fast. Flow vs. campaign revenue split is tracked and healthy (flows ~25-40% of email revenue). Over-reliance on campaign blasts means automated money is left on the table. Consent and compliance (double opt-in where required, SMS TCPA consent, honored unsubscribes) are documented. A single compliance miss can trigger fines and deliverability blacklisting.
Site architecture is flat and logical (money pages within 3 clicks of home) with a clear URL taxonomy. Deep, messy architecture buries pages and wastes crawl budget. Internal linking follows a hub-and-spoke topic-cluster model with pillar and support pages. Orphaned or randomly linked pages never accrue authority. One canonical URL pattern is enforced (trailing slash, www, protocol) with correct canonicals sitewide. Duplicate URL variants split link equity and confuse indexing. XML sitemap is clean (only indexable 200s) and submitted; robots.txt doesn't block key assets. Bloated or dirty sitemaps waste crawl budget and slow discovery. Structured data (Organization, Product, FAQ, LocalBusiness, Breadcrumb) is valid and matches visible content. Missing or invalid schema forfeits rich results and entity clarity. Templates enforce one H1, logical heading order, and unique title/meta patterns per page type. Template-level heading errors multiply across every page. Pagination, faceted navigation, and URL parameters are controlled to prevent index bloat. Uncontrolled facets can spawn thousands of thin duplicate URLs. International/geo targeting (hreflang or geo pages) is correctly implemented where relevant. Wrong hreflang serves the wrong market and cannibalizes rankings.
T Tracking Can you trust every number you're about to optimize against?
48 checks A Analyze Are you reading the data at the depth that reveals the real levers?
48 checks O Optimize Are you systematically compounding wins and cutting waste?
48 checks Flow timing, order, and branching are tuned from step-level data. Right message, wrong delay still loses the sale. Subject-line, preview, and send-time tests run with real statistical significance. One-off 'winners' without significance ship noise as strategy. Segmentation is tightened so cadence matches engagement tier. Matching cadence to engagement lifts revenue and protects deliverability at once. Personalization and dynamic content (recs, replenishment) are deployed in flows. Generic sends underperform 1:1 relevance by wide margins. A sunset/re-engagement policy is enforced to protect sender reputation. Keeping dead contacts to 'grow the list' wrecks inbox placement for everyone. Discount strategy is optimized toward margin (tiered, thresholded, or value-add). Flat sitewide codes erode margin and train discount-waiting. Deliverability is actively maintained (warmup, hygiene, complaint response). Reputation is earned slowly and lost fast; maintenance is continuous. SMS is optimized for timing, length, and MMS use against opt-out rate. SMS punishes missteps instantly with opt-outs you can't recover.
A prioritized experiment backlog is worked continuously, not sporadically. Cadence, not one big redesign, is what compounds CVR over time. Tests run to significance with pre-declared MDE and sample size. Peeking and early stops manufacture false winners. Winning variants are shipped and re-tested for durability. Wins can decay or interact; verification protects the gains. Mobile-specific friction (tap targets, sticky CTA, autofill) is optimized first. Mobile is most traffic and most friction - the highest-ROI surface. Checkout is streamlined (guest checkout, express wallets, fewer fields). Every removed step and field measurably lifts completion. Message match and above-the-fold value prop are optimized per traffic source. The first screen decides most bounces; source-matched copy wins them. Trust and honest-urgency elements are tested at decision points. Reviews, guarantees, and real scarcity move the fence-sitters. Core Web Vitals are optimized against the CVR-correlated metric (INP/LCP). Speed is a direct, compounding CVR and ranking lever.
Title/meta rewrites ship for high-impression, low-CTR queries. Fastest organic revenue lever that needs no new rankings. Decaying and thin content is refreshed or consolidated on a prioritized schedule. Refreshing a decayed page often beats writing a new one. Internal links are added from authority pages to striking-distance targets. A few internal links can push page-2 keywords onto page 1. On-page is optimized for intent and entities, not keyword density. Modern ranking rewards topical completeness, not repetition. Cannibalizing URLs are consolidated or re-targeted. Merging competing pages concentrates signals and lifts rank. Technical debt (redirect chains, broken links, CWV) is remediated on cadence. Compounding tech debt caps everything else you do. Content is optimized for generative-engine citation (AI Overviews, AI Mode, ChatGPT/Perplexity) with clear answers, named expertise, verifiable data, and schema. In 2026, being the cited source (GEO) is often the real win, not the blue link. New content targets validated gaps with a clear ranking thesis, not volume for its own sake. Publishing without a gap and thesis dilutes site focus.
S Scale Can you add spend and reach without breaking efficiency?
48 checks