Abdo Mazloum

Platform attribution leaves out discounts and cancellations, returns and refunds, fulfilment and payment fees, cash tied up during payback, and the lower returns that come with added spend.

• The measurement order is net sales, contribution margin, allowable CAC, marginal return, then controlled scale; only after that is the decision to kill, hold or scale.
• Before increasing spend, they check whether stock, cash flow, fulfilment and creative can support the extra demand.

One post shows the picture Abdo Mazloum is arguing against: Meta reporting 4.0X, Google 3.5X and email 2.2X when all three may claim the same order—a strong dashboard masking flat growth. The answer is four layers, each with a different job: attribution for daily decisions, lift tests for actual impact, MMM for budget allocation and finance data for real profit.

The earlier post argues against scaling an ad account on platform data alone; the later version adds lift tests and MMM to the profit-metric sequence.

Read original · Original 2 · Original 3 · Original 4

Mazloum separates the jobs: SEO gets products found; product data tells platforms what a product is, who it’s for and whether it can be bought.

• Both need to draw on the same source of truth.
• The four parts are identity—consistent product names, categories and variants; accuracy—matching prices, availability, descriptions and feeds; structure that search engines and commerce platforms can process; and experience—content that helps someone decide.

Mazloum describes a mismatch brands can easily miss: the title says one thing, the feed another; the price has changed, the variant structure is unclear and inventory status is outdated.

• The store may still appear in search, but turning visibility into qualified traffic and sales is much harder.
• The closing line is deliberately unglamorous: get found, be understood, make the purchase easy.

Read original · Original 2

AI systems don’t evaluate a store the way shoppers do; they rely on structured product data, product schema markup, GTINs, merchant feeds, FAQs, customer reviews, and trusted brand mentions and citations.

• The shift is not from SEO to AI but from rankings to recommendations.
• A store not built to be understood by machines reduces its chances of being discovered.

Mazloum starts with questions real shoppers ask AI—what’s the best protein powder for women, which skincare brand is worth the price, the best standing desk for a home office—then lists the signals that determine whether a brand is recommendable.

• A separate post describes a scoring framework to assess whether AI mentions a brand, understands its products, finds pages that answer shopper questions and surfaces products through the feed.
• The aim is to rank fixes by impact before AI-driven discovery gets more competitive.

Read original · Original 2 · Original 3

They split the work into three groups: problems already costing revenue, such as feed errors, unclear product pages, hidden reviews and confusing policies; conversion blockers, such as checkout friction, disconnected reporting, inconsistent pricing and poor customer-service handoffs; and future infrastructure, such as agent-ready catalogues, secure checkout, fraud controls, reliable commerce data and AI channel measurement. Two filters determine priority: the impact on revenue today and the importance of the work as AI-led buying grows.

Mazloum describes a practical sequence for leadership, not a single audit: fix what is easy and costly, repair what blocks conversion, and prepare for what future channels will require. The work is mapped into next sprint, next quarter and a longer-term roadmap.

Read original

They put unit economics first—real CAC, margin, payback, returns and MER target—then signal infrastructure such as CAPI, server-side tracking, enhanced conversions and pixel audits, followed by the creative system, with enough hooks, angles, variants and fresh creative.

• Bidding and scaling come last.
• Starting with the budget is why spend rises while margin thins.

Mazloum lists the inputs that matter now beyond ROAS: creative sameness, signal quality, post-click friction, placement dependence, fatigue patterns and incrementality. They also cite taxonomy-based creative testing, funnel-stage creative mapping and scaling rules without resets.

Read original · Original 2

What matters is unglamorous and stable: understand customer problems, create content around real buying intent, answer questions clearly, build trust faster than competitors, and show up where attention is moving.

The post is built on humour rather than data, including the imagined Slack panic — "Do we have a GEO strategy???" — answered with "Brother, we barely had a strategy," and the observation that a customer is still googling something like the best CRM that will not make them want to fight their laptop.

Read original

When Meta expanded its use of business-shared activity into Feed personalisation and Meta AI responses, they noted that Meta had not said better tracking would increase organic reach or drive Meta AI recommendations—only that shared activity would influence more than ads.

• The recommendation was to fix consent, deduplication and value accuracy rather than chase the personalisation headline.
• In the AI-shopping post, they likewise make a conditional claim—a store not built to be understood by machines reduces its chances of discovery—and offer a checklist to test readiness, not a guarantee.

For the July rollout, Mazloum specified a five-point audit: pixel and CAPI deduplication so the same purchase isn’t recorded twice; purchase values and currencies matched to Shopify revenue data; event coverage so key events fire at the right stage; consistent permitted customer parameters for event matching; and documented consent and governance covering what is shared and how preferences are respected.

Read original · Original 2

Start here for the clearest demonstration that measurement plumbing is the core craft, not a side topic.

Mazloum explains that Meta is expanding how it uses activity businesses already share — website visits, product views, purchases, app activity, offline and CRM events — so that data now also informs Feed personalisation and Meta AI responses, not just ad delivery.

• The post gives a five-point audit for the July rollout: pixel and CAPI deduplication so the same purchase is not counted twice; purchase values and currencies matched to Shopify revenue; event coverage at the right funnel stage; consistent, permitted customer parameters for event matching; and documented consent and governance.
• The piece draws a clear boundary: they state that Meta has not said better tracking will increase organic reach or drive recommendations, only that shared activity will influence more than ads.
• The recommendation is to fix consent, deduplication and value accuracy rather than chase the personalisation headline.

For brands planning a campaign

A direct sample of diagnostic work for a Shopify DTC brand seeking a tracking-quality engagement. It shows the shape of the work: a named checklist covering deduplication, currency matching and consent governance, with a deliberately narrow claim about what better data does and does not buy you.

Read it to see the AI-readability lens applied to a concrete Shopify change.

On 28 May 2026, Shopify allowed merchants to customise /agents.md, /llms.txt and /llms-full.txt through Liquid theme templates.

• Mazloum reframes that as a storefront becoming readable by AI systems trying to work out what a shop sells, who each product is for, which collections matter, and how shipping and returns work.
• The post lists what those systems need, then lays out an audit path: product titles, collection structure, product-page copy, schema and structured data, Merchant Center feeds, reviews and product context, shipping and return clarity, and AI-readable theme data.
• Its argument: the visible page and the machine-readable layer are separate jobs, and the second is infrastructure.

For brands planning a campaign

Useful when scoping AI-commerce visibility work for a Shopify store: it names specific surfaces — agent files, feeds, schema and collection structure — rather than promising generic AI optimisation. It also offers a marker of how current Mazloum's platform knowledge is.

Open it for the underlying thesis in one short piece.

The post starts with a mismatch a brand can easily miss: a product page ranks well while platforms reading it get conflicting signals — the title says one thing, the feed another, the price has changed, the variant structure is unclear, or inventory status is stale.

• Mazloum separates the two jobs: SEO gets products found; product data lets platforms understand what the product is, who it is for, and whether it can be bought.
• The framework has four parts — identity (consistent names, categories, variants), accuracy (matching prices, availability, descriptions and feeds), structure (information search engines and commerce platforms can process), and experience (content that helps someone decide).
• The closing line is deliberately unglamorous: get found, be understood, make the purchase easy.

For brands planning a campaign

Test this argument against your catalogue before hiring. It gives brands a way to check whether a feed problem is real — compare title, feed and price consistency — and frames the deliverable as one source of truth rather than two separate workstreams.

Read it to judge whether Mazloum's AI-visibility framing is practical or hand-wavy.

Mazloum opens with the symptom a brand would notice first: e-commerce SEO performing worse than it used to.

• The post gives concrete shopper questions — best protein powder for women, which skincare brand is worth the price, best standing desk for a home office — and notes that AI increasingly answers them.
• The signals cited are structured product data, product schema markup, GTINs, merchant feeds, FAQs, customer reviews, and trusted brand mentions and citations.
• The reframe is the substance: the shift is not from SEO to AI but from rankings to recommendations.
• The claim is conditional: a store not built to be understood by machines reduces its chances of being discovered.
• The post offers a checklist to test readiness.

For brands planning a campaign

Hand this to a client who still thinks of AI visibility as a buzzword.

• It turns the pitch into a diagnostic: does the store have schema, GTINs, clean feeds, visible FAQs and reviews?
• It also shows the terms Mazloum uses when proposing AI-commerce visibility work.

Start here to see whether there is a repeatable method behind the opinions.

The post describes a scoring approach for whether AI systems mention a brand, understand its products, find pages that answer shopper questions, and surface products through the feed.

• Its purpose is prioritisation: benchmark current performance and rank fixes by impact before AI-driven product discovery gets more competitive.
• The post describes the framework rather than reproducing it.
• Ask Mazloum for the benchmark numbers, thresholds and scoring sheet to assess the specific cutoffs they apply.

For brands planning a campaign

Ask for the underlying scoring sheet when you engage. The post shows a structured, priority-oriented audit, but does not include the numbers a client would be measured against.

Read it to see the distinction between a campaign that looks profitable and a business that keeps the money.

Mazloum argues that platform ROAS shows only part of the picture because it leaves out discounts and cancellations, returns and refunds, fulfilment and payment fees, cash tied up during payback, and the lower returns that come with added spend.

• The measurement sequence is clear: net sales, contribution margin, allowable CAC, marginal return, controlled scale.
• Only then does the decision become kill, hold or scale — and only after checking that stock, cash flow, fulfilment and creative can support the extra demand.
• A campaign can report healthy ROAS while the business keeps very little.

For brands planning a campaign

Useful for setting expectations on a consulting engagement: this is the language to expect when the conversation moves from tracking to spend. Mazloum will ask for post-sale cost data, so the work depends on cooperation between marketing and finance.

Review the breadth and self-serve style of the frameworks before paying for an engagement.

This distribution post promotes a bundle of 40 free frameworks.

• Its contents list includes Google Ads scaling blueprints; Meta, Google and TikTok comparison guides; ROAS recovery checklists; CAC reduction frameworks; paid media account structures; AI commerce readiness systems; SEO versus AEO versus GEO breakdowns; product feed performance fixes; email profitability playbooks; and creative testing frameworks.
• The material is packaged as repeatable checklists and systems, with a lead-generation mechanic — comment, connect, seven-day window — to collect an audience.

For brands planning a campaign

A practical way to sample the work before hiring: the free bundle covers feed fixes, AI-commerce readiness and measurement recovery, closely matching a brand's need for those services. Ask for evidence of results before treating the bundle as proof of performance.

The posts repeatedly put measurement in this order: net sales, contribution margin, allowable CAC, marginal return, then controlled scale. They support this with four layers: attribution for daily decisions, lift tests for actual impact, MMM for budget allocation, and finance data for real profit.

Numbered step sequences and a KILL/HOLD/SCALE decision frame. Before increasing spend, Abdo Mazloum asks whether stock, cash flow, fulfilment, and creative can support the added demand.

For brands planning a campaign

In a tracking-quality engagement, expect discussion of post-sale costs, incremental measurement, and allowable CAC—not just pixel or CAPI health.

Examples: Source 1 · Source 2 · Source 3 · Source 4

It notes that Meta has not said better tracking will lift organic reach. The accompanying audit list covers pixel and CAPI deduplication, purchase values and currencies matched to Shopify revenue data, event coverage, event matching, and consent and governance.

For brands planning a campaign

For a campaign focused on tracking quality, a brand can test whether Abdo Mazloum can audit deduplication, currency and value accuracy, event matching, and documented consent—not just recommend new dashboards.

Examples: Source 1 · Source 2

The proposed source of truth has four parts: identity, with consistent product names, categories, and variants; accuracy, with matching prices, availability, descriptions, and feeds; structure that search engines and commerce platforms can process; and experience, with useful content for a purchase decision.

For brands planning a campaign

Frames feed accuracy and product data as a discovery problem, helping brands assess whether this creator works on the data layer behind visibility, not just page copy.

Examples: Source 1 · Source 2

Signals discussed include structured product data, product schema markup, GTINs, merchant feeds, FAQs, customer reviews, and trusted brand mentions. A Shopify post covers merchants customising /agents.md, /llms.txt, and /llms-full.txt through Liquid theme templates, along with the store information AI systems need to read: what is sold, who each product is for, which collections matter, and how shipping and returns work.

Recommends auditing product titles, collection structure, PDP copy, schema and structured data, Merchant Center feeds, reviews and product context, shipping and return clarity, and AI-readable theme data.

For brands planning a campaign

Directly matches a brief on AI-commerce visibility. A brand can check whether the work moves from naming the shift to specifying fixes and measurement.

Examples: Source 1 · Source 2 · Source 3 · Source 4

It separates quick wins already costing revenue, such as feed errors, unclear product pages, hidden reviews, and confusing policies; conversion blockers such as checkout friction, disconnected reporting, inconsistent pricing, and poor service handoffs; and future infrastructure such as agent-ready catalogues, secure checkout, fraud controls, reliable commerce data, and AI channel measurement. Two filters determine priority: how much the issue affects revenue today, and how important it becomes as AI-led buying grows.

Pushes back on treating everything as a transformation project, sequencing the work into next sprint, next quarter, and longer-term roadmap.

For brands planning a campaign

A brand can ask for a sequenced plan rather than a blanket audit, and see how Abdo Mazloum separates urgent fixes from future build.

Examples: Source 1

A separate post orders the stack as unit economics, signal infrastructure, creative system, then bidding and scaling. It argues most brands start at the last layer, raising spend while thinning margins.

For brands planning a campaign

Shows the diagnostic order behind Abdo Mazloum's paid-media advice and how they rank creative and signal work against budget changes—useful when comparing them with creators who lead with bidding tactics.

Examples: Source 1 · Source 2

What Abdo Mazloum says actually matters: understanding customer problems, creating content around real buying intent, answering questions clearly, building trust faster than competitors, and showing up where attention is moving.

Uses humour, analogy, and direct address to the reader, followed by a short imperative list rather than data.

For brands planning a campaign

Sets expectations for tone: Abdo Mazloum is likely to resist new-acronym framing and reframe a campaign around customer questions, buying intent, and trust.

Examples: Source 1

Examples include commenting "PAID" for a paid-media cheat sheet covering input layers, creative testing, placement strategy, signal quality, KPI scorecards, and hook templates; commenting "40" for a bundle of 40 frameworks covering Google Ads scaling blueprints, CAC reduction, paid media account structures, product feed fixes, email profitability, and AI commerce readiness; and a benchmark scoring framework for whether AI mentions a brand and understands its products across ChatGPT, Perplexity, and Gemini.

Numbered and arrowed post structures, comment-and-repost prompts, and "Follow Abdo Mazloum for..." closers.

For brands planning a campaign

Brands can use the engagement routes described in the posts to request underlying frameworks or direct contact; the distributed assets, rather than the post text, are the offer.

Examples: Source 1 · Source 2 · Source 3

Both treat machine-readable product information as the deciding factor in whether a brand gets recommended, and both close with practical steps such as fixing feeds and strengthening schema.

Shivers focuses on the protocol standards themselves—Google's Universal Commerce Protocol and OpenAI and Stripe's Agentic Commerce Protocol—explaining how each was designed and what it changes about the discovery-to-purchase flow. Mazloum places the same shift within a broader e-commerce growth stack, tying AI visibility to measurement, prioritisation and ad spend rather than protocol architecture.

Read original · Original 2

The posts share a broad focus on making stores easier to find, use and convert, and both address operators with numbered, actionable advice.

Shukla's posts offer general best-practice lists for store owners—one-page checkout, guest checkout, mobile-first design, clean URLs—without measurement or profit context. Mazloum ties each recommendation to data quality, feed accuracy or a measurement decision, arguing for sequencing and priority rather than listing options in parallel.

Read original · Original 2

Both work through the Shopify-versus-Meta purchase-count mismatch, pixel and CAPI deduplication, event IDs, purchase values and currencies, and consent. Both distinguish tracking problems from attribution problems before recommending a fix.

Roy goes deeper into implementation and platform plumbing—GA4, GTM, Consent Mode V2, EU privacy requirements and external payment flows such as PayPal and Klarna—writing as a tracking and analytics specialist. Mazloum treats tracking health as one layer in a paid-media and profit system, alongside unit economics, creative and scaling decisions.

Read original · Original 2

The posts explain the OpenAI and Stripe Agentic Commerce Protocol and Google's Universal Commerce Protocol: how agents consume product feeds, compare items, build carts and complete checkout through delegated payments while the retailer remains merchant of record.

• They spell out the retailer's steps: supply OpenAI with a structured product feed, implement ACP checkout endpoints, support delegated payments, keep pricing, inventory and fulfilment consistent, and treat the Merchant Center Next feed, local inventory feed and structured data as determinants of eligibility for AI Mode and agent-led checkout.
• They also identify a checklist most brands lack: which feed fields an agent reads before recommending a product.

This is narrower than a campaign aimed at tracking quality and product-data accuracy, with AI visibility as one output.

• The posts focus on agentic-commerce standards and feed readiness; they do not cover pixel and CAPI deduplication, purchase-value reconciliation, consent and governance, or the shift from platform ROAS to contribution margin and allowable CAC.
• Nor do they test how pages or feeds perform against search and product-data inconsistencies.
• For an engagement that combines tracking quality and product-data accuracy with AI visibility, Tom Shivers covers the AI-commerce half in more depth than Abdo Mazloum, but leaves measurement unaddressed.
• Both posts are dated June 2026, making the protocol detail recent but the coverage only two posts wide.

Examples: Source 1 · Source 2

The posts examine the same investigation Mazloum's tracking post points to, at greater length and with the discrepancy as their sole focus.

• They distinguish tracking problems from attribution problems, then cover attribution windows and click-through versus view-through credit; Pixel and CAPI deduplication using consistent event IDs verified in Events Manager; duplicate purchase events from refreshes or unreliable triggers; non-customer orders; incorrect event revenue or currency; payment redirects from PayPal and Klarna; and consent-state limitations for EU traffic.
• The distinction helps scope the work: fixing duplicate events will not resolve attribution differences, and changing attribution settings will not fix broken purchase tracking.

The posts do not address measurement.

• They omit product feeds, catalogue identity and accuracy, search and AI visibility, and the financial test that should precede a budget increase.
• They also skip merchant-of-record protocol work and storefront readability for AI systems.
• This alternative is stronger for a single diagnostic of reported purchase counts, but weaker if the engagement also needs to reconcile product data, improve what AI systems can read, or connect the fix to margin and payback.
• The assessment rests on two posts from September and October 2026; ask for client names and campaign outcomes before relying on it for those decisions.

Examples: Source 1 · Source 2

The posts offer a ten-point checklist for frictionless checkout (one-page checkout, guest checkout, payment-method range, auto-fill, mobile-friendly flow, transparent pricing, security indicators, clear error messages, BNPL and distraction-free layout) and a paired checklist for SEO-friendly store design (site structure, mobile-first, page speed, clean URLs, navigation, product-page design, headings, internal links, image alt text and trust signals). Each point is plainly stated and can be turned into a design requirement without further translation.

The tradeoff is depth and specificity.

• The recommendations are generic and uniformly positive: every item is recommended, but none is prioritised, scored or tied to evidence from a working store.
• The content offers a checklist, not a decision about what to fix first.
• It has no measurement layer—no attribution, tracking quality, feed accuracy, consent or financial context—and no AI or agentic-commerce visibility.
• That misses the campaign's need for a sequenced backlog justified by revenue impact and future channel importance.
• Both posts date from January and February 2026, earlier than the other alternatives reviewed here, and the audience is beginner-level rather than operators spending at scale.

Examples: Source 1 · Source 2