Mark Tran shares practical advice on improving Google and Meta ads for ecommerce brands, with a focus on finding new customers and reducing wasted spend.
Mark Tran is useful for practical Google and Meta ads advice on finding new customers and cutting wasted spend.
Best for
Ad measurement or paid media software platform selling to DTC ecommerce operators.
Not for
Consumer brands seeking shopper facing content, direct purchases, and broad reach.
Key takeaways
Two Meta accounts counted returning customers as new, hiding repeat purchases behind strong reported results.
Branded search reported 25x ROAS, while Performance Max was closer to 5x.
For small Performance Max budgets, narrow the product feed before reviewing individual products.
Check first: Ask for impressions, engagement, profile visits, and audience demographics beyond the 863 follower figure.
Main ideas
Separate new-customer growth from repeat purchases · Look beneath blended ROAS for demand the brand did not create · Spend less broadly when the budget cannot support enough learning
Mark Tran shares practical advice on Google and Meta ads for ecommerce brands, especially how to tell whether campaigns are finding new customers or taking credit for sales that might have happened anyway. Their posts turn account data and platform changes into concrete decisions: whom to exclude from prospecting, where to trim spend, and when a setting matters for a particular product category.
Tran connects performance numbers to the business questions behind them.
In two Meta accounts, strong reported results obscured repeat purchases; they recommend separating new and returning customers and measuring prospecting by new-customer volume, even if reported ROAS falls (campaigns that looked like they were printing money).
In a Google Ads account, branded search reported 25x ROAS while Performance Max was closer to 5x.
They capped branded spend and redirected budget toward finding new buyers, challenging the idea that a flattering account average necessarily signals growth (the branded-search ROAS breakdown).
The same analytical approach informs their advice to narrow a Performance Max product feed when a small budget is spread too thin, then review individual products to find what deserves more or less spend (the $40-a-day Performance Max example).
They also distinguish between platform changes that matter for some advertisers and those that do not: a longer purchase-audience window may help brands with long replacement cycles, but matters less for products commonly repurchased within 90 days (who benefits from Meta’s 730-day purchase window).
For ecommerce marketers, the posts offer a useful habit: question what a metric counts, then connect campaign settings to the customer and business outcome they want.
Across these five posts, Mark Tran returns to one test: does an ad campaign create new demand, or merely make existing demand look like growth?
Read more
The examples lead to practical checks—separate new from returning customers, inspect performance below the account-wide average, and concentrate spend where it can produce useful evidence. They also argue that audience settings should reflect buying cycles and that brands can use AI creative while preparing for its labels.
Separate new-customer growth from repeat purchases
Tran argues that platforms can favor cheap conversions from people who would have bought anyway.
Read more
A strong reported ROAS is not proof that prospecting found new customers; separating new and returning buyers can make the numbers look worse while giving a clearer measure of acquisition.
In posts about a cricket store and a tennis store, Mark Tran says broad Meta campaigns without existing-customer exclusions counted many repeat purchases as “new.”
The proposed fix was to separate customer groups, exclude past buyers from prospecting, and leave retention to email and SMS.
They say reported ROAS fell, while new-customer volume became the more useful measure.
Look beneath blended ROAS for demand the brand did not create
Tran questions account-wide returns lifted by branded searches, which often come from people already seeking the brand.
Read more
The argument is to distinguish capturing existing intent from finding new buyers, then judge budgets with that difference visible.
For a bedding brand spending $10,000 a month on Google Ads, Tran says branded search used 12% of the budget and returned 25x, while Performance Max was closer to 5x. They capped branded spend at 5%, cut weak keywords, and moved freed budget to Meta prospecting; the post reports little revenue loss and a clearer view of account performance.
Spend less broadly when the budget cannot support enough learning
Tran’s small-budget advice is to reduce the number of products before expecting an algorithm to optimize well.
Read more
Product-level results, not a large feed or a reassuring dashboard, should guide what gets removed or more budget.
A client had more than 4,000 products and a $40 daily Performance Max budget.
Tran says even narrowing the set to fewer than 1,000 left too many products to learn from.
The next step was to remove high-cost products without revenue and fund unexpected winners and underbid products in a catch-all campaign.
The post also warns that seasonal Performance Max can lose to a plain Shopping feed after the seasonal window ends, so product and campaign choices need revisiting.
Match purchase-audience windows to how long customers take to buy again
Tran distinguishes purchase events from general site activity when explaining Meta’s 730-day retention window.
Read more
The change is not equally useful for every business: a longer purchase cycle can make a longer past-buyer exclusion valuable, while a short cycle may already fit within the old window.
Tran says the 730-day window applies to purchase events, while page views and add-to-carts remain capped at 180 days. They contrast skincare and supplements with furniture, appliances, and jewellery, and advise long-cycle brands to check whether acquisition exclusions cover buyers from 12 or 24 months earlier.
Prepare for AI labels without treating them as a ban on AI creative
Tran presents automatic labeling as a consumer-trust and production issue, not a reason to stop using AI-generated creative.
Read more
The practical response is to identify where generation or editing occurs, check live ads, and make sure creative and media teams know what to expect.
Tran says Meta may label ads using AI-generated backgrounds, images, or animation, including work from tools such as Photoshop generative fill, DALL-E, and Canva. The post distinguishes automatic labels from the stricter disclosure requirement they describe for political and social-issue ads, and recommends auditing the creative pipeline and checking live ads for the label.
The example shows how Tran audits an account and answers the first question a DTC brand should ask: how much of reported ROAS comes from real demand creation?
Read more
Mark Tran describes taking over a bedding brand spending $10k a month on Google Ads.
About 12% of the budget went to branded search — ads triggered when someone already types the brand name — which was running at 25x ROAS and lifting the blended account average.
Performance Max, built to find new buyers, was closer to 5x.
Tran capped branded search at 5% of spend and cut weak keywords, concentrating the rest on converting terms.
Branded ROAS rose to 64x on a fraction of the budget, with very little revenue lost.
The freed budget moved to Meta prospecting, and account ROAS began to reflect the actual business.
For brands planning a campaign
When auditing paid search, separate branded from non-branded and restate ROAS without branded search before judging performance. Cap branded spend as a small toll, not a growth lever, and move the savings into prospecting.
The post identifies a common measurement failure that makes Meta accounts look better than they are and offers a repeatable check for campaigns.
Read more
A cricket store and a tennis store ran the same Meta setup: one Advantage+ campaign, broad targeting, campaign budget optimisation and no exclusion for existing customers.
The month's reported ROAS, customer acquisition cost and ad performance looked strong, but a large share of the "new" sales came from existing customers buying again.
Tran says Meta optimises for the cheapest conversion, usually someone who was going to buy anyway, so the campaign recycled existing demand and reported it as growth.
They point to Cody Wittick's breakdown of a $71K version of the same pattern and Australian ad-spend audits estimating that up to 44% of digital spend is wasted or misattributed.
The fix: separate new and returning customers, exclude past buyers from prospecting, and put retention back into email and SMS.
Reported ROAS fell; new-customer volume became the metric that mattered.
For brands planning a campaign
Before crediting a prospecting campaign with growth, split new from returning purchasers and exclude past buyers from prospecting. Reported ROAS may drop when recycled demand is removed; judge the campaign by new-customer volume instead.
The post corrects a common misreading of Meta's data change and turns it into a concrete audience-setting decision for planning exclusions.
Read more
Meta extended its data retention window to 730 days, but Tran says the change applies only to purchase events; page views, add-to-carts and other engagement events still cap at 180 days.
So anyone building lookalikes or exclusions from general site activity is unaffected.
Someone who bought 14 months ago now stays in a purchase-based custom audience instead of dropping out at six months.
Tran says this makes little difference for skincare or supplements with a 90-day repurchase cycle, but matters for furniture, appliances, jewellery and other categories with a one- to two-year purchase cycle, where past buyers were falling out of exclusion audiences and being hit by prospecting campaigns.
Data capture is automatic; check the window on exclusion audiences and, for long-cycle products, extend it to 12 or 24 months.
For brands planning a campaign
Match exclusion-audience windows to the category's typical purchase cycle. For long-cycle products, extend past-buyer exclusions so acquisition budget stops reaching people who bought last year.
The post gives small-budget accounts a specific first move and shows why product-level analysis can beat dashboard-level judgement for leaner DTC spend.
Read more
A client with more than 4,000 SKUs gave Tran $40 a day for Performance Max.
They narrowed the feed to under 1,000 SKUs, then found a few days later that even that was too many: spread across a thousand products, none received enough data for the algorithm to learn from, leaving the account effectively running a thousand underfunded experiments.
Tran cut further, pulled brands with high cost and no revenue, gave more budget to a handful of unexpected outliers, and raised bids on winners that had been underbid in a catch-all campaign.
The "exclude these five, boost these five" pattern came from reading individual product performance rather than a dashboard.
They also note that seasonal themed Performance Max can beat a plain Shopping feed during its window, then fall behind once the season ends.
For brands planning a campaign
On a small daily budget, treat product-set size as the main lever: cut products until each gets enough data to learn from, then shift spend toward those showing actual revenue rather than adding more SKUs.
AI creative is now part of most ecommerce production pipelines.
Read more
This post explains what happens to those assets on Meta, a compliance question brands should settle before the label appears.
Meta now auto-applies an "AI info" label to ads that use background generation, image generation or animation.
It detects this through metadata written by its own tools and third-party tools such as Photoshop generative fill, DALL-E and Canva.
Tran stresses that this is automatic labelling, not a rule requiring brands to declare AI or risk rejection; the stricter disclose-or-be-rejected requirement applies only to political and social-issue ads.
Audit the creative pipeline and flag every step where AI generates or edits an image, check live ads through the three-dot menu to see whether the label appears, and brief designers and media buyers so nothing is a surprise.
Tools that write no metadata may slip through for now, but Tran expects the gap to close as Meta's own detection improves.
Google added its own AI ad labels on 9 July, with EU disclosure rules following on 2 August.
Tran supports the direction on consumer-trust grounds and sees AI creative accelerating, rather than as something to avoid.
For brands planning a campaign
Map which creative steps use AI generation or editing, and check live ads for the label so client expectations are set in advance. Treat AI labelling as a settled direction of travel, not a reason to move away from AI-assisted creative.
Auditing AI-generated creative before Meta labels it
Mark Tran explains that Meta now automatically adds an "AI info" label to ads on Facebook and Instagram when brands use background generation, image generation, or animation, including third-party tools such as Photoshop generative fill, DALL-E, and Canva that write metadata.
Read more
Automatic labeling is not a rule requiring brands to declare AI use or face rejection; the stricter disclose-or-be-rejected requirement applies only to political and social-issue ads.
The stated reason is consumer trust as photorealistic AI people become more common.
Tran says the direction is set: Google added its own AI ad labels on 9 July, and EU disclosure rules take effect on 2 August.
Audit the creative pipeline to flag every step where AI generates or edits an image. Check live ads by clicking the three dots to see whether the "AI info" label appears, and brief designers and media buyers so nobody is surprised.
Tools that do not write metadata may avoid labeling for now, but Meta also runs its own detection, and the gap is expected to close as the standard spreads.
For brands planning a campaign
For a brand running AI-generated visuals, map where AI touches the creative pipeline and decide whether the label changes creative choices. Treat automatic labeling as a compliance and consumer-trust consideration, not a bar to using AI creative.
Separating new customers from recycled demand in Meta campaigns
Across two ecommerce accounts, a cricket store and a tennis store, Mark Tran describes campaigns with the same setup: one Advantage+ campaign, broad targeting, Campaign Budget Optimisation, and no exclusions for existing customers.
Read more
Monthly reports showed strong ROAS, low customer acquisition cost, and strong ad performance, but deeper inspection found that a large share of "new" sales came from existing customers cycling back.
Meta, left to its own optimization, will aim for the cheapest conversion, which usually means someone who was going to buy anyway, Tran explains.
The campaigns were recycling existing demand and calling it growth.
The post cites Cody Wittick breaking down a $71K version of the same issue and Australian ad-spend audits estimating that up to 44% of digital spend is effectively wasted or misattributed.
Separate new from returning customers, exclude past buyers from prospecting, and move retention work back into email and SMS. Tran notes that reported ROAS dropped after the fix, which was accepted because new-customer volume became the metric that mattered.
For brands planning a campaign
For an acquisition brief, require new-customer reporting and past-buyer exclusions before trusting headline ROAS. Expect reported ROAS to fall when measurement is separated; treat that drop as clarity, not failure.
Mark Tran argues that Meta's 730-day retention update is being misreported.
Read more
The 730-day window applies to purchase events only; page views, add-to-carts, and other engagement events still cap at 180 days.
Exclusion audiences or lookalikes built on general site activity are unchanged.
Someone who bought 14 months ago is now in a purchase-based custom audience; before 18 May, they would have dropped off after six months.
Tran says this is not meaningful for most brands, since skincare or supplement buyers should repurchase within 90 days and a six-month window was already enough.
It matters for furniture, appliances, jewellery, and anything with a one-to-two-year purchase cycle, where past buyers were dropping off exclusion audiences and potentially getting hit by prospecting campaigns.
Check the actual window set on acquisition campaigns that exclude recent purchasers.
For long-cycle categories, extending it to 12 or 24 months stops acquisition budget going to people who already bought last year.
The data updated automatically; decide what to do with it.
For brands planning a campaign
Audit exclusion audiences against the brand's real purchase cycle before extending windows. A long-cycle brand can reduce wasted acquisition spend by excluding buyers it may previously have re-targeted by mistake; a short-cycle brand may see no material gain.
Mark Tran describes taking on a client with $40 a day to run Performance Max across more than 4,000 SKUs.
Read more
The instinct was to load everything in and let Google sort it out, but narrowing to under 1,000 SKUs was still too many.
A few days in, the data showed that spreading spend across a thousand products on a $40 daily budget meant none got enough data to learn from, so the algorithm never received a clear signal.
Tran calls it running a thousand underfunded experiments rather than one campaign.
The response was subtraction: narrow the product set, pull out brands with high cost and no revenue, give more budget to outlier products performing unexpectedly well, and lift winners that were sitting underbid in a catch-all campaign.
Read individual product performance rather than relying on a dashboard, then exclude losers and boost outliers and underbid winners.
Seasonal themed PMax can fly during its window and then fall behind a plain Shopping feed once the season ends, Tran warns.
Set-and-forget quietly bleeds spend.
For brands planning a campaign
On a small Performance Max budget, start with a concentrated product set and review individual product performance before adding more SKUs. If ads were spread across hundreds of products, running half as many may improve learning and reduce waste.
Mark Tran describes taking over a bedding brand spending $10k a month on Google Ads, with 12% of budget in branded search.
Read more
The previous agency reported a beautiful blended ROAS, but branded search was running at 25x, dragging the whole account average up.
Meanwhile, PMax, the campaign built to find new buyers, was closer to 5x.
Branded search catches people who were already looking for the brand and mostly would have bought anyway, Tran argues, so a huge branded number is not growth.
Capping branded at 5% of spend and cutting weak keywords more than doubled its ROAS to 64x on a fraction of the budget, with barely any revenue lost.
The freed budget moved to Meta to find people who had never heard of the brand, and account ROAS finally reflected new-demand performance.
Separate branded search from prospecting campaigns, cap branded spend at a small share, cut weak branded keywords, and concentrate what remains on converters.
For brands planning a campaign
A strong blended ROAS can hide weak new-demand performance. For a brand auditing wasted spend, break out branded search and PMax before making budget decisions, and consider redirecting branded budget to channels that create new demand.
Mark Tran’s posts show an ads practitioner at work.
Read more
They describe running Google and Meta campaigns for ecommerce clients, including two unnamed stores (a cricket store and a tennis store), an unnamed bedding brand spending $10k a month, and another client with more than 4,000 SKUs on a $40 daily budget.
The posts name Meta and Google as the ad platforms they work in, and explain platform mechanics and campaign practice rather than a sponsorship or commercial partnership.
Ask for client references or permission to name the businesses.
Meta
Mention
Mark Tran writes about how Meta Ads behaves: the 730-day retention window for purchase events, Advantage+ campaigns that recycle existing customers into apparent new-customer growth, and Meta’s automatic “AI info” labelling of AI-generated ad creative.
Read more
Their posts describe using and interpreting Meta’s advertising tools for client work. Ask whether any work with Meta involved sponsorship, employment or a disclosed partnership.
Mark Tran writes about running Google Ads for ecommerce clients: Performance Max spread across too many products on a $40-a-day budget, and a bedding brand’s branded search campaign inflating the account’s blended ROAS.
Read more
They work in and advise on Google Ads, and cover Google’s announced AI ad labelling. Ask whether any work with Google involved a client, employment or sponsorship relationship.
Explore Mark Tran's posts by theme: start with the argument that new-customer growth should be separated from repeat purchases, then follow the specific checks on Meta audience windows, branded-search budgets, and small-budget Performance Max.
Read more
The work is anchored in ecommerce accounts, with concrete numbers, product decisions, and audit steps rather than general advice.
Gregory also works on ecommerce paid media at scale, writing about why ROAS declines as budgets grow, funnel gaps for cold audiences, cross-channel overlap between Google and Meta, and the need to check margin and CAC before scaling.
Read more
Like Tran, they advise diagnosing the system behind the ads rather than changing bids or duplicating campaigns.
Gregory's posts present general scaling systems and offer free Google Ads audits.
Tran's posts focus on specific account interventions, with named budgets, product counts, and cuts.
Gregory covers broader funnel and LTV strategy, while Tran focuses more narrowly on measurement windows, branded-search leakage, and product-level Performance Max decisions.
Tesselaar builds on the argument that ROAS is not return, showing how it can hide margins, warm audiences, and diminishing returns.
Read more
Both explain that Google Ads accounts can look profitable while serving repeat buyers, and recommend separating branded from non-branded and using new-customer or profit metrics.
Tesselaar teaches metric frameworks such as POAS, nCAC, and break-even ROAS, along with a Performance Max structure checklist based on margin-split asset groups and CRM data. Tran tends to start with a named client account and show the specific budget or product cuts made.
Paxton also diagnoses ecommerce paid media from the margin side, telling the story of a home goods client whose ROAS looked fine while contribution margin was under 12%.
Read more
Both treat margin and new-customer economics as the real test, with step-by-step checks for tracking, creative, audience overlap, and channel fit.
Paxton's posts lay out process sequences, including a four-step sales-drop diagnostic and a four-change client turnaround.
Tran's posts tend to isolate one issue at a time, such as audience windows, AI labeling, or branded-search budget.
Paxton also covers post-purchase and offer changes, while Tran stays closer to campaign settings and product feeds.
Mark Tran's posts suggest a topic match, not a measured audience fit.
Read more
For a DTC brand campaign about paid ad performance and wasted spend, they write for ecommerce operators and performance marketers: Meta's 730-day retention window and exclusion audiences, campaigns that look profitable while recycling existing customers, Meta's automatic labelling of AI-generated ad creative, narrowing the SKU set on a small Performance Max budget, and branded search inflating account ROAS.
The people drawn to that material are peers and practitioners in ecommerce ad buying, so a sponsored placement reaches that peer audience rather than the brand's own shoppers.
A buyer planning a campaign to acquire new customers should treat this as reach into the paid-media conversation, not into a consumer market.
The one measured audience figure is 863 LinkedIn followers, observed 9 October 2026.
That's a small following, so reach depends on how far individual posts travel within the ecommerce paid-media niche rather than on audience scale.
The five posts were published between 19 May and 15 July 2026.
The posts do not report audience location, age, gender, employers, job roles, engagement counts, impressions, or audience composition.
Before committing spend, ask Mark Tran for recent results on posts comparable in topic, depth, and format to these examples — impressions, saves, comments, and profile visits — plus any audience survey or platform analytics documenting geography, age, gender, and job role.
Those figures, rather than the topic match alone, will show whether a placement puts the brand in front of ecommerce decision-makers or mainly other media buyers.
None of the five posts available for Mark Tran disclose a sponsorship offer, media kit, rate card, or defined partnership format.
Read more
The material is operator commentary on Google and Meta advertising rather than promotional content written for brand partners, and no post names a sponsor or advertises a paid collaboration.
Because there is no published sponsor policy, no disclosed sponsored example, and no verified commercial inquiry route, a brand cannot currently build a paid partnership plan from what is documented here.
The practical route to working with Mark Tran appears to be a direct professional conversation rather than a published package.
This is an absence of documented offers in the material at hand, not evidence that Mark Tran declines or accepts sponsorships.
Evidence that would change this section: a sponsor policy page, published deliverables or rates, an example of disclosed sponsored work, or a contact route specifically for commercial inquiries.
Good fit for a focused campaign. The posts support a narrowly targeted series.
Read more
Brief
An ad-measurement or paid-media software platform selling to DTC ecommerce operators wants a sponsored LinkedIn series showing brands how to find wasted spend in their Google and Meta accounts, with teaching grounded in real account mechanics rather than product features.
Reasons
The audience matches the buyer, not a consumer market. Mark Tran's posts attract ecommerce operators, media buyers and agency-side performance marketers. A platform selling audit and measurement tooling to those roles is speaking to the same people the content already reaches.
The posts demonstrate a repeatable audit method, not general opinion. Examples include freeing a bedding brand from a blended ROAS propped up by branded search, splitting new from returning purchasers in two Meta accounts, and cutting a Performance Max feed from under a thousand SKUs to what a $40 daily budget can actually learn from. A sponsor can point to each as a worked example.
The format suits a sponsored series. Each post is a short narrative with a numbered or signposted set of actions and a closing question to the reader. The structure supports product-adjacent teaching without turning into an advertorial.
There is a natural bridge to a paid-media product. Each post ends with an actionable decision, giving a measurement platform a credible place to offer a checklist, template or trial.
Limits
The only measured audience figure is 863 LinkedIn followers, observed 9 October 2026. Ask for impressions, engagement and profile-visit data to assess reach beyond the niche.
No sponsorship offer, rate card, media kit or sponsored example is documented. Agree the commercial shape directly with Mark Tran rather than selecting from a published package.
Treat the fit as a topic-and-practitioner match. The posts don't specify the audience's location, seniority or employers; ask for audience demographics before forecasting qualified-buyer reach.
When to choose a specialist
Poor fit.
Read more
Brief
A consumer skincare brand wants shopper-facing content that drives direct purchases from its own customers, including short-form video showing the product in use, plus broad reach beyond paid-media professionals.
Reasons
The audience is the trade, not the market. These posts are written for people who run ad accounts, not the skincare shoppers a consumer brand needs to reach.
The available posts show no product demonstrations, lifestyle content, taste-test or try-on formats, customer-facing hooks, or creative built to sell a physical product to a shopper.
The five items reviewed are text posts, so a brief that depends on short-form video performance cannot be assessed from this material. A video title would not substitute for a transcript.
Reach is small. With 863 followers as the only measured figure, a consumer brand buying for scale would be paying for niche professional reach rather than volume.
No sponsor policy, deliverables, rates or disclosed sponsored work are documented. Ask for these before building a brand partnerships plan.
What would change it
Engagement and impression data showing posts reaching well beyond the follower count, plus audience demographics that include consumer shoppers.
A published sponsorship package with defined formats, rates and a commercial inquiry route.
Actual consumer-facing or video examples, collected and reviewed in full before any fit judgment.
A worked campaign example
Angle
"Where Your ROAS Goes" — three teardowns using the method Tran already applies in their posts, each showing a brand where reported performance diverged from real demand.
Brief
The analytics platform wants a three-post sponsored LinkedIn series aimed at DTC ecommerce operators, with one downloadable resource and a demo or trial offer at the end.
Measurement
Qualified demo requests or trial signups attributed to the series, split by whether visitors arrived from a post or the checklist.
Comments, saves and profile visits from readers whose profiles list ecommerce or paid-media roles.
Checklist downloads are a weaker but still useful intent signal. If downloads run high and demo requests do not, the offer placement is the problem, not the audience.
Offer
Close the series with a direct route to a demo or free trial, placed after the third teardown rather than woven into every post.
Post plan
Post one: blended ROAS. Walk through the branded search case: a $10k monthly Google Ads account where 12% of budget sat in branded search running at 25x, while Performance Max, the campaign built to find new buyers, ran closer to 5x. Show the move: cap branded spend at a small share, cut weak keywords, and restate account performance without branded search.
Post two: new customers versus returning ones. Show the two-account pattern: one Advantage+ campaign, broad targeting, budget optimisation and no past-buyer exclusions produced strong reported ROAS, while a large share of “new” sales came from repeat buyers. Show the fix: separate new from returning customers, exclude past purchasers from prospecting and judge the campaign on new-customer volume. Add the audience-window decision: purchase events extend to 730 days, while engagement events stay at 180. That matters for furniture, appliances and jewellery, but barely for skincare or supplements.
Post three: small-budget Performance Max. Show the $40-a-day account with more than 4,000 SKUs, the cut to under a thousand, the finding that even that was too many and the subtraction that followed. End with product-level reading, not dashboard judgment, and the seasonal themed PMax trap.
Resource
One portable audit checklist with three columns — branded versus non-branded spend, new versus returning purchasers, and number of products receiving meaningful data — that readers can fill in against their own accounts without the sponsor’s tool. It should work as a standalone document, not as a lead form dressed up as a template.
Tradeoffs
Expect modest raw reach and judge the series on relevance rather than volume. The teaching format keeps the sponsor’s product out of most of the text; if the sponsor needs logo presence in every post, ask for a different approach.
A brand spending enough that scaling has become the problem, and looking for a structure-first programme rather than a single-account teardown.
Read more
Gregory's two posts describe where growth breaks once budgets cross roughly £25–30K a month on Meta or Google: cold audiences lack a funnel; Google and Meta are planned in isolation, causing overlap and wasted clicks; and there is no lifetime-value or offer strategy to justify paying more for a first purchase.
The fixes are programmatic rather than tactical: weekly creative-testing cycles, margin and customer-acquisition-cost checks before budgets rise, offers built for first-time buyers, funnel progression mapped from awareness to purchase, and landing pages made for cold clicks.
The second post also offers free Google Ads audits to ecommerce brands, giving brands that want a working session alongside the content a direct route.
These posts name categories of waste rather than showing one account rebuilt line by line, so readers get less of the inspectable mechanics in Mark Tran's branded-search and small-budget Performance Max write-ups. The closing audit offer also brings sponsor-adjacent content close to a service pitch, making it a better fit for a lead-generation brief than a pure teaching one.
A brand facing a measurement problem rather than a tactical one — reported ROAS looks healthy, but profit is thin.
Read more
Tesselaar works through the arithmetic: ROAS is revenue divided by spend, before cost of goods, shipping, returns and fulfilment, so a 4x campaign can lose money while a 2x campaign can be the most profitable.
Average ROAS hides diminishing returns, so marginal return on extra spend should determine where budget moves.
Profit-on-ad-spend sets break-even at one divided by gross margin, meaning a 35% margin breaks even at 2.86x.
A worked example moves an account from 39% to 51% new-customer share, with new-customer cost down 30%, while ROAS looked strong throughout.
The companion post sets out a five-part Performance Max structure: protect branded traffic with a dedicated exact-match campaign so a reported ROAS collapse is visible; split asset groups by margin; feed the algorithm CRM and lifetime-value audiences; use conversion value rules so a $20 sale and a $200 sale are not treated as the same signal; and read search themes as a diagnostic.
For a platform selling measurement or profit reporting, this is the closest match to the product.
The writing is dense and formula-led, closer to a working note than the short narrative with signposted actions that carries well in a LinkedIn series. Brands looking to see one account taken apart, rather than learn a framework for judging any account, will get less from these two posts.
A brand facing a sudden, unexplained drop in sales that wants a diagnostic order to follow before anyone touches campaigns.
Read more
Paxton's sequence is explicit: compare Shopify, GA4 and Meta to check for broken tracking, treating a gap above 10% as a reporting problem; compare the store's seven-day conversion rate with the prior 30 days to distinguish a site problem from an ads problem; look for a specific change in the ad accounts that lines up with when the numbers moved; then check seasonality and external context across the last 12 months.
Working through these four checks usually turns up one culprit.
The second post details a full client turnaround: a home goods brand doing $120K a month on Meta, with contribution margin under 12% and revenue flat for four months, was spending 65% of its budget retargeting the same 14-day window.
Cutting retargeting to 25%, adding a three-step post-purchase sequence, raising the free-shipping threshold by $14 and removing weak feed SKUs took the brand to $197K by month four and $280K by month nine, with margin at 23%.
For a brief built on before-and-after proof, that offers more concrete outcome evidence than the reviewed Mark Tran posts.
The material ranges across agency operations and broader growth work, with less depth on platform mechanics such as Meta's purchase-versus-engagement retention windows or how to size a product feed against a small daily Performance Max budget. The turnaround concerns a single unnamed client; ask for the starting baseline to assess how transferable the result is.
Mark Tran's five LinkedIn posts cover Meta's 730-day retention window for purchase events, a two-account Meta campaign where reported results hid repeat purchases, Meta's automatic labelling of AI-generated ad creative, a Performance Max account running on a $40 daily budget across thousands of products, and a bedding brand's branded search spend inside a blended ROAS figure.
The posts are undated, so readers cannot tell when these views emerged or how they developed.
The posts are text only — they include no video, transcripts, other channels, or longer-form work.
They include no engagement, impression or audience-demographic data, and clients appear without being named, so Meta and Google are the only named brands and the fit discussion is limited to what the posts show.
The only measured audience figure is 863 LinkedIn followers, observed on 9 October 2026.
No sponsorship offer, media kit or disclosed sponsored work appears in the posts, so sections on commercial terms, such as partnership formats and rates, remain open.
