Juliana J.

Google Analytics, Google Ads and an A/B testing tool measure different things, collect data at different moments, treat identity differently and apply their own unseen attribution and deduplication rules, so their figures were never going to agree.

• In Jackson’s view, chasing parity outsources confidence to the data instead of building judgement about what the numbers mean.
• The people most insistent on matching numbers are often the ones who trust the data least.
• Jackson credits Simo Ahava with the framing they now use: it is fine for data to be different as long as it is consistently different.
• The productive question is whether the relationship between two sources has changed, and why—not why they do not match.

In one cross-team conversation, a dashboard says one thing, Google Ads another and an A/B testing tool a third, while a stakeholder wants a single figure for a board deck. Jackson notes that Google Ads has reasons to report more conversions than GA: its attribution model affects whether spending continues.

The pairing of GA and Google Ads as an example comes from past work: Jackson says they avoid Google Analytics day to day and draw on cross-team conversations they have been part of over the years.

Read original

Tracking usually gets blamed first, but Jackson argues that nobody had formally agreed what a new customer is before anything was built.

• Instead, teams inherited the definition used by whichever system did the measuring, typically a session-level, cookie-scoped, device-scoped default shipped by the ad or analytics platform.
• Different teams have genuinely competing interests in the answer.
• Jackson breaks “new” into at least four distinct ideas, including new to the brand and new to the category.
• The fix is to write down a business definition before building audiences.

Someone who has bought Nike shoes for decades through Amazon, retail stores and other channels, but buys directly from Nike.com for the first time, gets flagged as a new customer, enters an acquisition flow, may receive a first-purchase welcome discount and can later appear in retargeting aimed at cold prospects.

Read original

Jackson argues that this assumes every step of the pipeline belongs to the same company.

• In a franchise model, the franchisor owns the brand and funds national marketing, each franchisee is a separate legal entity that holds customer records, and a third vendor owns the software holding the sale.
• Access to a sales record becomes a commercial product with its own pricing.
• The gap survives every new measurement technology because the data is not missing: it sits complete and accurate in thousands of systems owned by parties who have weighed what sharing it gets them and mostly concluded it is not enough.

An Autotrader white paper built on Clarivoy’s analysis of 875,000 automotive sales, which Jackson cites as showing that 92% of vehicle sales remain untraceable in dealership CRM systems, with only 8% tied back to a lead metric such as a form submission. Jackson also cites CDK’s agreement to pay $630 million in January 2025 to settle claims from 243 software vendors over access to dealership data, while admitting no wrongdoing.

A LinkedIn post first framed the topic as a change to how readers should approach offline integration posts, noting that the tutorials assume the whole pipeline sits inside one company. The newsletter edition then worked through the argument in detail, using dealerships and franchises.

Read original · Original 2 · Original 3

Jackson argues that a brand’s buyers and the way money physically reaches the company should determine its channel and measurement approach, not the newest tactic.

• The industry treats these established businesses as edge cases while generalising from a narrow slice of venture-backed commerce.
• What looks like slowness in large consumer goods companies, Jackson argues, is a highly evolved form of risk management: mistakes in product formulation or supply chain carry consequences across dozens of jurisdictions.

Jackson describes Cosnova, whose Essence and Catrice brands are social-first because their audience lives on TikTok and Instagram; a Beiersdorf speaker who noted that a large part of Nivea’s German base still watches linear television and learns about new products there; and a Unilever Foods speaker who treated attribution for two-euro grocery products as the physics of the business rather than a dashboard problem.

Read original

Their standard is effectiveness rather than efficiency: if you cannot point to a decision that changed, you have spent a lot of money learning to send emails faster.

• Jackson argues that distribution was never the problem: analytics data has never been more distributed.
• Putting an agent on top of a dashboard changes nothing; a stakeholder who ignores the dashboard will ignore the agent too.
• Jackson’s constructive claim is that the creative asset is the last digital touchpoint a brand still controls, so it is worth testing before launch.

On YouTube, where the creative carries the targeting, pre-flight testing can break down an asset, score it against creative effectiveness principles and the brand’s historical data, then give the creative team specific changes to make rather than a general impression. Jackson also notes that CRM data remains deeply overlooked.

Read original · Original 2 · Original 3

The strategic shift, in Jackson’s view, is moving from using data for reporting to using it to power advertising.

• The browser session is fading as a reliable source of commercial value as generative search handles intent before anyone reaches a site and more transactions happen server to server.
• With automation inside ad platforms, the quality of the data fed to them is the main lever a team still controls.

Jackson explains that Ads Data Hub is a privacy-safe clean room running on Google Cloud.

• It lets a company query user-level Google ad campaign data and join it against its own first-party data without exposing personal information, making questions such as the average time lag between a YouTube ad view and a later Search conversion answerable.
• Jackson’s measurement black-box example: YouTube spend influences Search conversions, but a last-click model gives Search all the credit.
• A live session with Stape CEO Denis Golubovskyi covers feeding ad algorithms real profit margins instead of raw revenue and deploying tracking for new channels such as ChatGPT ads in days.

The newsletter edition draws on Jackson’s recent move into the media industry and role as Cloud Director, Data Science at Jellyfish, a change they describe as reorienting them from day-to-day web analytics toward cloud data and applied AI.

Read original · Original 2

The guide separates three jobs often lumped together as data analysis: engineering and modelling; business intelligence dashboarding; and the business analyst’s work of interrogating prepared data, enriching it with context and turning it into a narrative about the current state, opportunity and recommendations.

• Jackson places themself in the third group and insists the work begins with business objectives and market context before anyone opens a dashboard.
• They also note that their process is shaped by their background in business, product and conversion rate optimization, and credit Marco Giordano with enriching the search approach in the piece.

A fictional direct-to-consumer eco-friendly skincare brand selling one-off products and a subscription serves as the example. The walkthrough follows the chain from framing business objectives to strategic recommendations, with market context covering fragmented journeys, AI-generated answers and discovery through Reddit and TikTok.

Read original

Technical skill alone, they argue, doesn’t show how a business makes money or how to contribute value; the farther you are from the source of revenue, the harder it is to prove the value of your work. Jackson credits positioning, messaging, product launches, narrative, go-to-market principles and competitive intelligence with improving the questions they ask, and calls product marketing the best extra skill set they’ve invested in.

Jackson says they repeat this reminder every year and attributes most of their career successes to the product marketing skills they built, rather than technical analytics skills alone.

Read original

It answers a newcomer’s question: what does a practical analysis process look like before and after the dashboard?

Jackson writes as a business analyst, not a data engineer or BI developer, and walks through a fictional DTC skincare brand that sells one-off products and a subscription.

• The guide starts before the dashboard: define business objectives, understand the market, interrogate the data, then translate findings into current state, opportunity and recommendations.
• The context section covers fragmented journeys, AI-generated answers, Reddit and TikTok discovery, and E-E-A-T as reasons to become the source rather than just rank.
• Jackson also credits Marco Giordano for enriching the SEO approach.

For brands planning a campaign

For a data storytelling workshop, this is a ready-made case for teaching students to frame questions, add context and turn analysis into a strategic narrative rather than a dashboard walkthrough.

Reporting from d3con in Hamburg, the piece contrasts brand-side panels from Beiersdorf, Cosnova, Spotify, Netflix, Google, Deutsche Bahn and Unilever.

• Cosnova's Essence and Catrice are social-first because their audience is social-first.
• Nivea still reaches a 40+ German buyer through linear television.
• Unilever Foods treats attribution for two-euro grocery products as business physics, not a dashboard problem.
• Jackson argues these are not edge cases but the dominant reality outside a US venture-backed bubble.

For brands planning a campaign

A workshop can use this contrast to teach why measurement design must follow distribution and buying behavior, not just the latest channel tactic.

It shifts the question from 'which number is right?' to 'has the relationship between them changed, and why?'

Jackson explains that tools measure different things, collect data at different moments, handle identity differently, and apply their own attribution and deduplication logic.

• They quote Simo Ahava: it is okay for data to be different as long as it is consistently different.
• The practical trap is treating numbers as comparable because a dashboard puts sessions, clicks and revenue in the same grid.
• Jackson notes that privacy changes, consent mode, ITP and ad blockers leave much traffic invisible or modelled.

For brands planning a campaign

For a data storytelling workshop, this offers a strong lesson in explaining measurement uncertainty to stakeholders without defending one number or drowning them in caveats.

Jackson shows that the money and customer record sit in different companies: the franchisor funds national marketing, the franchisee owns the customer relationship, and a third-party DMS vendor holds the software record.

• In US automotive, Jackson cites an Autotrader white paper based on Clarivoy's analysis of 875,000 sales: 92% of vehicle sales remain untraceable in dealership CRM systems, meaning only 8% tie back to a lead metric.
• Jackson also notes that CDK agreed to pay $630 million in January 2025 to settle claims from 243 software vendors over data access.
• The tracking problem is a commercial negotiation across company boundaries and vendor paywalls.

For brands planning a campaign

A workshop on offline conversion measurement can use this case to map data ownership, incentives and contracts before drawing architecture diagrams.

Jackson describes campaigns that look fine on CPA but later reveal that a meaningful share of 'new' customers already had purchase history.

• The cause: nobody agreed on what new customer meant.
• The ad platform, CRM and finance each define purchase history differently.
• Jackson's Nike example makes it tangible: a long-time Nike buyer who bought through Amazon, retail stores and other channels but visits Nike.com for the first time gets flagged as new by the site's systems.
• Jackson breaks 'new' into at least four possible meanings, including new to brand and new to category.

For brands planning a campaign

For a data storytelling workshop, this is a concrete exercise in aligning business definitions before building audiences or reporting acquisition results.

Jackson writes about moving into media and taking a role at Jellyfish as Cloud Director, Data Science.

• They describe GA4-to-BigQuery as the non-negotiable first step, but say the real work happens elsewhere in Google Cloud.
• They explain Ads Data Hub as a privacy-safe clean room for joining Google ad data with first-party data without exposing PII, and point to Vertex AI Model Garden and Hugging Face for solving business problems.
• The measurement black-box example: YouTube spend influences Search conversions, but last-click gives Search all the credit.

For brands planning a campaign

An advanced workshop can use this to show how measurement, cloud data and AI fit together in enterprise marketing, and why reporting is not the same as powering decisions.

It offers a useful starting point for newcomers looking to understand how Jackson thinks about career value.

Jackson's yearly reminder to CRO, digital analytics, CRM, data science, dev and engineering people: take a product marketing course.

• Technical skills alone aren't enough to understand how a business works or how to bring value to it.
• The further you are from how a company makes money, the harder it is to prove the value of your work.
• They credit product marketing skills for most of their career successes.

For brands planning a campaign

A workshop can open with this argument: data storytelling lands only when the analyst understands positioning, messaging, go-to-market and commercial context.

In a post about an Analytics Unscripted podcast episode, Jackson argues that many AI mandates are driven by fear of missing out—and that the overlooked question is whether the work is effective, not merely efficient.

• If you can't point to a decision that changed, you've spent a lot of money learning to send emails faster.
• They say analytics data has never been more distributed, and putting an agent on top of a dashboard changes nothing: stakeholders who ignore the dashboard will ignore the agent, too.
• In their view, the creative asset is the last digital touchpoint the brand still controls.
• On YouTube, the creative carries the targeting, which is why pre-flight testing can decompose an asset, score it against creative effectiveness principles and historical performance, and give the creative team specific changes.

For brands planning a campaign

A workshop on creative measurement can use this to move from dashboard critique to decision integration and pre-flight creative testing.

Jackson announces a live conversation with Denis Golubovskyi, CEO of Stape, on architecting a pipeline that reflects business reality rather than just browser behavior; separating useful signal from noise; why server-side deserves a bigger role than a compliance fix; feeding ad algorithms real profit margins instead of raw revenue; and deploying tracking for new channels like ChatGPT ads in days rather than waiting for native integrations.

For brands planning a campaign

A workshop can use this agenda to teach why server-side measurement, margin data and fast channel coverage are now part of the analyst's decision toolkit.

The guide distinguishes three jobs often grouped under data analysis: engineering and modeling, BI dashboarding, and business analysis, which interrogates prepared data, adds context, and builds a narrative covering the current state, opportunity, and recommendations. Jackson identifies with the third role and walks through the full process using a fictional DTC eco-friendly skincare brand, starting with business objectives before opening a dashboard.

Starts with core business objectives, then works through one example end to end instead of listing techniques in the abstract.

For brands planning a campaign

Structure the workshop around one client case, from business question to recommendation, and tailor it to which of the three analysis roles participants hold.

Examples: Source 1

Jackson argues that the tools measure different things, collect data at different moments, handle identity differently, and apply their own unseen attribution and deduplication logic, making a match the wrong target.

• They credit Simo Ahava with the framing they now use: it is acceptable for data to differ as long as it differs consistently.
• The productive question, they argue, is not "why don't these match" but "has the relationship between them changed, and why?"

Shifts the discussion from reconciling numbers to tracking whether the gap between two sources stays stable over time—and treating a change in that gap as the signal.

Jackson says they avoid working in Google Analytics day to day, so the examples draw on past use cases and cross-team conversations rather than current hands-on practice.

For brands planning a campaign

Suits sessions on measurement governance and stakeholder expectations for teams pressed to produce one defensible number.

Examples: Source 1

Jackson argues that the industry treats these as edge cases while generalizing from a narrow slice of venture-backed commerce. What looks like slowness in large FMCG organizations, they say, is a highly evolved form of risk management.

Draws on conference panel accounts from brand-side practitioners, not vendor or agency framing.

For brands planning a campaign

Relevant to audiences in established retail, FMCG, or traditional marketing organizations where the DTC measurement playbook does not fit the buying process.

Examples: Source 1

In dealerships and franchises, it does not.

• Jackson outlines the split: a franchisor owns the brand and national marketing fund; each franchisee is a separate legal entity holding customer records; and a third vendor owns the software holding the sale, the dealer management system.
• CDK and Reynolds & Reynolds have long dominated that market.
• Jackson cites a figure from an Autotrader white paper based on Clarivoy's analysis of 875,000 automotive sales: 92% of vehicle sales remain untraceable in dealership CRM systems.
• They also cite a $630 million January 2025 CDK settlement with 243 software vendors over access to dealership data.
• Their point: the attribution gap persists despite new measurement technology because the data is not missing; it sits in thousands of systems owned by parties weighing what they gain from sharing it.

Maps who owns each step of the pipeline and what each party gains or loses, treating integration as a commercial negotiation rather than a technical task.

For brands planning a campaign

Useful for sessions on data ownership and partnership in distributed organizations, and for audiences in automotive, franchise, or multi-entity businesses.

Examples: Source 1 · Source 2 · Source 3

Teams usually blame tracking first.

• Jackson argues that no one formally agreed on the meaning of "new customer" before building anything, so the definition came from whichever system was measuring—typically a session-level, cookie-scoped, device-scoped default shipped by the ad or analytics platform.
• They break "new" into at least four concepts, including new to the brand and new to the category, and offer a Nike example: a decades-long customer buying directly for the first time gets counted as new.

Unpacks conflicting team incentives—acquisition wants a wide definition, retention a narrow one, and finance whichever reports cleanest to CAC—and shows how the definition ends up unowned.

For brands planning a campaign

A strong fit for teams debating acquisition and retention reporting, or sessions on writing business definitions before building audiences.

Examples: Source 1

They argue the more interesting work happens further into Google Cloud: Ads Data Hub as a privacy-safe clean room for joining Google ad campaign data with first-party data without exposing personally identifiable information, and Vertex AI Model Garden and Hugging Face for building solutions to specific problems.

• A separate LinkedIn post about a live session with Stape's CEO covers architecting a pipeline that reflects business reality rather than browser behavior, why server-side deserves a bigger role than a compliance fix, feeding ad algorithms real profit margins instead of raw revenue, and deploying tracking for new channels such as ChatGPT ads in days rather than waiting for a native integration.
• Across both, the framing is that the browser session is fading as a reliable source of commercial value.

Positions the warehouse-and-dashboard layer as the foundation, not the destination, and shifts the discussion to server-side pipelines and the quality of the signal sent to automated ad platforms.

For brands planning a campaign

Best suited to technical paid media, data, and infrastructure audiences with a warehouse who are deciding what to build on top of it.

Examples: Source 1 · Source 2

Their test is whether anything is effective rather than merely efficient: if you cannot point to a decision that changed, you have spent a lot of money learning to send emails faster.

• They apply the same lens to analytics tooling, arguing that the problem was never distribution and that putting an agent on a dashboard changes nothing because a stakeholder who ignores the dashboard will ignore the agent.
• Their constructive claim is that the creative asset is the last digital touchpoint a brand still controls.
• Pre-flight testing—decomposing the asset, scoring it against creative effectiveness principles and against how those elements performed in the brand's own historical data—gives creative teams something specific to change.

Asks what decision changed as the test of any AI or tooling investment, then redirects effort toward testing and improving the creative asset itself.

For brands planning a campaign

Fits an event or workshop on AI adoption with an evidence-first, skeptical angle, particularly for brand and creative teams rather than infrastructure teams.

Examples: Source 1 · Source 2 · Source 3

The argument is that technical skill alone does not show how a business works or how to bring value to it; the further you are from how a company makes money, the harder it is to prove the value of your work. They credit positioning, messaging, product launches, narrative, go-to-market principles, and competitive intelligence with improving the questions they ask, and describe product marketing as the best extra skill set they have invested in.

Frames commercial literacy as a complement to technical skill rather than a career change, valuing it for the quality of questions it produces.

For brands planning a campaign

Useful for workshops aimed at analytics, data, or engineering teams that need to connect their work to commercial outcomes and make recommendations land with business stakeholders.

Examples: Source 1

The post describes this as Jackson's current role at the time it was written.

View example

Original disclosure

This newsletter edition is produced in partnership with Stape.

View example

The post does not describe a sponsor or other commercial arrangement.

View example

Their posts on branded search and SaaS SEO execution overlap with Jackson's interest in how measurement and business signals work together.

Dyulgerova's posts are short LinkedIn arguments about E-E-A-T, branded search, and slow execution in SaaS marketing; Jackson writes longer newsletter essays analysing measurement strategy and business-model constraints.

Read original · Original 2

Ali's posts focus on implementation for ecommerce and Meta/Google campaigns, including market-specific payment behavior; Jackson focuses on measurement definitions and why cross-company data flows break down.

Read original · Original 2

Both explain attribution and tracking discrepancies to non-specialist readers.

Roy's posts are step-by-step diagnostic checklists for tracking and attribution issues in Meta CAPI, Shopify, and consent setups; Jackson takes a broader view of business definitions, ownership, and decision quality.

Read original · Original 2

Dyulgerova's posts argue that branded search is an underrated trust signal and that struggling SaaS businesses usually lose to faster competitors because too many departments review a single blog post or a strategy takes months to approve.

• That gives a workshop a clear operational spine: how to cut review queues, ship SEO and content work faster, and treat branded search as a cross-channel outcome, not just a keyword metric.
• It is the stronger choice when the buyer wants a blunt execution and accountability session with explicit actions for marketing teams.

The tradeoff is that Juliana J.'s two LinkedIn posts offer neither a teaching method nor a data storytelling case.

• They don’t explain how to run a full workshop, present a worked analytics narrative, or provide training materials for a mixed audience of analysts and business stakeholders.
• Dyulgerova’s advice is directional and SEO-specific; it doesn’t show how to turn conflicting dashboards into a current-state, opportunity and recommendation story, the core of the Juliana J. brief.
• A training company choosing Dyulgerova would need to build the data storytelling curriculum itself or accept a narrower SEO execution workshop.

Examples: Source 1 · Source 2

Ali's Bonkers.corner case explains why a drop model that worked in India did not transfer to Pakistan because 65-70% of ecommerce transactions were cash on delivery, so scarcity lost force when customers had nothing at stake.

• Ali's first-party data post gives a three-step practical sequence for Meta and Google: upload customer lists as custom audiences, connect CRM data to ad platforms where possible, and exclude existing customers from cold campaigns.
• For a training company whose clients need immediate paid media actions rather than a narrative workshop, this is a more direct operational fit.

The tradeoff is that the posts are short tactical LinkedIn items, not a facilitated workshop.

• They don’t show a complete teaching method, a long-form analysis, or a data storytelling framework that moves from question to recommendation.
• Ali’s material is strongest on market behavior and ad platform data inputs; it doesn’t address the measurement disagreements, business definitions and cross-company data ownership problems that give Juliana J.’s workshop its range.
• A company choosing Ali would gain specific paid media tactics but still need another facilitator or curriculum for the storytelling and decision-writing half of a campaign aimed at CFOs.

Examples: Source 1 · Source 2

Roy’s posts lay out a concrete investigation for Meta Ads reporting more purchases than Shopify or GA4: compare attribution windows and click-through versus view-through credit, check Meta Pixel and Conversions API deduplication with shared event IDs, audit duplicate or test purchase events, inspect the payment checkout journey, and review consent and measurement limitations.

• Roy separates tracking discrepancies from attribution discrepancies and gives an audit order: tracking, deduplication, data quality, consent, attribution.
• That’s more precise than a general data storytelling workshop when the room is made up of tracking specialists, media buyers and analytics engineers.

The tradeoff is that the posts focus on platform troubleshooting, not story-led training for a mixed business audience.

• Roy’s posts don’t show a workshop format, a narrative teaching method, or how to translate an audit into a recommendation for finance or commercial stakeholders.
• They focus on Meta, Shopify, GA4 and consent, leaving out the business model, data ownership and cross-team definition problems that Juliana J.’s workshop covers.
• A training company would get a sharper diagnostic checklist but need to add the decision-making and presentation layers itself.

Examples: Source 1 · Source 2