Richard Wyatt shares practical ecommerce advertising advice, especially on Google Shopping, Performance Max and product-feed strategy. The examples show how to separate low-traffic products into Shopping campaigns and use Onefeed labels and tracking data to assess performance.
2.3K followersPerformance MaxBased on 11 posts · latest Aug 2026
Useful for hands-on Google Shopping and Performance Max campaign structures using Onefeed feed labels and product-level data.
Best for
An ecommerce brand running Google Shopping and Performance Max that wants feed segmentation and campaign splits.
Not for
A reach-led sponsored placement or influencer package with audience scale and defined deliverables.
Key takeaways
Performance Max misses products with no impression history, so run Shopping alongside it.
Use Onefeed custom labels for products with fewer than two sales and 1,000 impressions.
The example reports 50 to 60 percent more Shopping revenue, not a benchmark.
Check first: Ask for recent comparable engagements, scope, fees, and deliverables before committing.
Main ideas
Performance Max misses products with no impression history, so run Shopping alongside it · The product feed is the control layer, and new items need a scheduled test window · Read a performance drop from campaign-level signals before changing strategy
Richard Wyatt shares practical ecommerce advertising advice, with a focus in these posts on Google Shopping, Performance Max and the product data behind them.
They are also closely associated with Onefeed: Wyatt describes its feed-labeling, basket-tracking and campaign-analysis features.
The clearest example is a proposed way to use Shopping alongside Performance Max: isolate products with few sales and impressions, group stronger sellers separately, and return products to Performance Max after they demonstrate sales and a good return.
Wyatt says this approach produced 50–60% more Shopping revenue in the example they discuss; that is their reported result, not a general benchmark.
Follow their work for hands-on campaign structures and frank glimpses of the product and app-development problems behind ecommerce tools.
Wyatt connects campaign decisions to product-level evidence rather than treating Performance Max as a complete answer.
In their worked example, custom labels separate low-traffic products, top sellers and products supported by basket or checkout data; campaign splits then reflect sales, traffic and, when available, margin.
They also describe checking budget pacing, stock-outs, changes in product sales and Performance Max traffic mix when diagnosing a decline.
That combination makes their posts useful for understanding what to test in an account and what data the proposed test depends on.
The exceptions matter: their Magento post is a request for help completing app testing, while their Shopify authentication post is a brief account of debugging frustration, not a tutorial.
Together, the posts show practical product and implementation concerns.
Richard Wyatt's posts converge on one argument: ecommerce advertising performance is decided at the product and customer level, not by choosing the right campaign type.
Read more
The recurring positions are that Performance Max leaves low-visibility products behind, while a parallel Shopping structure built from feed labels recovers that revenue; product feeds control how items are routed between campaign types and give new products a test window; performance drops should be read from campaign-level signals before anyone changes strategy; and basket-level customer data — new versus existing buyers, lifetime value, repeat rates — makes those segmentation choices possible. Two further posts show the build side of the same work: Shopify is the platform the tooling is built on, while Magento is described as unusually painful to test against.
Performance Max misses products with no impression history, so run Shopping alongside it
Wyatt's central claim is that Performance Max is so focused on warm traffic that it is effectively blind to items that have not yet built up demand signals.
Read more
Products with almost no impressions stay that way, and the revenue they could contribute never appears in the PMax account.
The fix they propose is structural rather than a bidding tweak: use the feed to identify those items and give them their own Shopping campaigns, then let proven items return to Performance Max.
Decisions about which items belong in which campaign draw on sales, impression, conversion and, where available, margin data, plus basket-level tracking to catch products converting on site through other channels.
Wyatt describes building a custom label in Onefeed for items with fewer than two sales and fewer than 1,000 impressions in the past 30 days, then setting a Shopping campaign to show only those items — which, in their telling, can cover around 95% of a feed.
A second label marks top sellers, usually by 90-day conversions plus any best-seller list the client supplies.
The Shopping campaigns are split three ways: best sellers with low traffic on aggressive return targets, everything else on the client's target return, and further tiers by required return where the client shares margin data.
A final label lets items back into Performance Max only once they are selling at a good return.
Wyatt reports that the parallel structure produced an extra 50–60% revenue in the example shown.
The product feed is the control layer, and new items need a scheduled test window
A second recurring method runs through the same post: campaign eligibility is set in the feed rather than negotiated campaign by campaign.
Read more
Labels based on sales history, margin or product creation date determine which campaign type an item can enter, while separate campaigns absorb material that does not belong in the main structure. This makes segmentation repeatable as a catalog changes.
Wyatt suggests a Shopping campaign for newly added items, labelled by product creation date, because Performance Max often leaves new products with no exposure until they have sales.
Once those items start selling, they can be picked up by Performance Max.
They also recommend a low-bid fallback campaign for every client that collects the negative keywords added elsewhere — described as low priority and manual bidding, and framed as a recurring source of unexpected volume.
Read a performance drop from campaign-level signals before changing strategy
When an ecommerce account underperforms, Wyatt's approach is diagnostic first: pull a fixed set of campaign-level readings that point to a cause, rather than starting with a strategy rewrite.
Read more
Each signal maps to a different explanation for a dip.
Wyatt describes an addition to Onefeed that summarises a deep-dive analysis of Bing and Google campaigns gathered through the advertising APIs.
Calculated per campaign, it covers six outputs: items that were selling well over the previous 7 or 30 days and have dropped away; budget pacing, including whether caps are likely to be hit; spend and performance with negative and positive contributing factors plus suggested actions; out-of-stock product reporting; Performance Max traffic-mix shifts, named as a common cause of a sudden dip; and changes in impressions, clicks and costs.
The stated purpose is to replace several hours of pulling reports.
A top-level summary overview was still to be added, Wyatt notes, so the feature was not complete in the post's own account.
Customer-level tracking — new versus existing buyers and repeat purchase — sits underneath the feed decisions
Wyatt argues that most ecommerce businesses lose sight of customer lifetime value, the cost of acquiring a new customer, and how to bring existing customers back, because those numbers get buried in day-to-day operations.
Read more
The data layer that answers those questions also makes product-level segmentation possible: basket-level tracking records what was bought, who bought it, and whether the buyer was new or returning.
Onefeed's tracking, in Wyatt's account, captures the entire basket along with the customer ID and new or existing status as a standard part of the Shopify app.
The platform reports lifetime-value measures alongside lifetime and last-12-month comparisons, repeat rates, and the items that attract the most new customers.
The same basket-level data appears earlier in the Shopping method, where it flags products converting on site through other channels, including items that reached checkout.
Platform choice shapes the work: Shopify-first, with Magento treated as a known cost
Wyatt writes about the engineering behind the tooling, not just the strategy, and the platform determines how much friction that work carries.
Read more
Shopify is where the app is built and where their attention sits; Magento is described as difficult to work with and thinly used. There is a tension worth noting: they frame Shopify as displacing legacy ecommerce platforms while still running a Magento app project.
In one post, Wyatt describes spending two days working through every debugging tool available to get Shopify's GraphQL, OAuth and Remix authentication process working, and says plainly that the two days were lost to it. In another, after a developer let the team down on a Magento app, they publicly ask for access to a test store because Magento offers no sandbox or basic test account — unusual compared with other ecommerce platforms, they say, and openly framed as a big ask.
The two posts reflect a continuing position: the Magento work persists despite Wyatt's stated preference for Shopify and criticism of Magento's testing options.
This is the most complete worked example in the posts here.
Read more
To see how Richard Wyatt approaches recovering revenue that Performance Max leaves behind, start here: it explains what to do when PMax concentrates spend on products that already sell and leaves low-impression items with almost no traffic.
Wyatt lays out a routine for running Shopping and Performance Max in parallel using Onefeed custom labels.
The first label isolates items with fewer than two sales and fewer than 1,000 impressions in the past 30 days; a Shopping campaign then shows only those items.
Wyatt says this can cover around 95% of a feed.
A second label marks top sellers, typically by 90-day conversions plus any best-seller list the client supplies.
Basket-level tracking data flags products that convert on site through other channels, including items that reached the checkout.
The Shopping campaigns are split three ways: best sellers with low traffic on aggressive ROAS targets, everything else on target ROAS, and, where the client shares margin data, further tiers by required ROAS.
A final label lets items back into PMax only once they are selling at a good ROAS.
Wyatt reports that running Shopping in parallel produced an extra 50–60% revenue in the example shown, and argues those items were getting almost no PMax traffic.
Two further suggestions: a Shopping campaign for newly created items, labelled by creation date, and a low-bid fallback campaign that collects the negative keywords added elsewhere.
For brands planning a campaign
For a Shopping and Performance Max engagement, this offers a concrete starting hypothesis for feed segmentation and campaign splits: separate low-traffic items from PMax, tier them by sales evidence and margin, then re-admit proven items. Treat the 50–60% figure as the author's reported result, not a benchmark, and test the sales and impression thresholds against a specific account before adopting them.
It shows where to look first when an ecommerce account suddenly underperforms, and what the tooling Wyatt builds reports.
Read more
Read it if diagnosis matters to a Shopping or PMax engagement.
Wyatt describes an addition to the Onefeed platform: a summary of a deep-dive analysis of Bing and Google campaigns, gathered through the advertising APIs.
It lists six outputs, calculated per campaign: items that were selling well over the previous 7 or 30 days and have dropped away; budget pacing, including whether budget caps are likely to be hit; campaign spend and performance, with negative and positive contributing factors plus suggested actions; out-of-stock product reporting; channel performance changes in Performance Max, since a shift in traffic mix is named as a common cause of a dip; and general changes in impressions, clicks and costs.
Wyatt says the aim is to replace several hours of pulling reports when performance has fallen, and notes that a top-level summary overview was still to be added.
For brands planning a campaign
This is a usable dip-diagnosis checklist regardless of the tool: check for products that stopped selling, budget caps, stock-outs, traffic-mix shifts, and cost or click changes before rewriting strategy. If the Onefeed product is part of the pitch, ask which of the six outputs are live in the current version; the post says the summary overview is unfinished.
It explains the data layer behind the feed segmentation and diagnosis in the other posts, including the customer-level information Wyatt treats as standard and why new-versus-existing customer data matters to a repeat-purchase brief.
Read more
Wyatt argues that most ecommerce businesses lose sight of customer lifetime value, the cost of acquiring a new customer, and how to bring existing customers back, as those data points get buried in day-to-day operations.
In this account, Onefeed's tracking goes beyond conversion counting: it captures the entire basket along with the customer ID and whether the buyer is new or existing.
The post describes this as a standard part of the Shopify app.
The platform shows lifetime-value metrics compared with lifetime and the last 12 months, repeat rates, and the items that attract the most new customers.
For brands planning a campaign
If a brand's goal is repeat purchase or efficient acquisition rather than one-off feed optimisation, first check whether basket-level tracking and new-versus-existing customer flags are in place: the labelling and re-admission logic in the Shopping example depends on that data. Ask the client to confirm the tracking setup on its own store rather than assuming the app description applies.
It shows the build side of Wyatt's work rather than the strategy side, and explains what happens when the software side of an ecommerce engagement goes wrong.
Read more
Read it for context on the platforms Wyatt operates across and how the app work is resourced.
Wyatt says a developer let the team down on a Magento app, leaving final output testing incomplete, and asks whether anyone can provide a test store because Magento offers no sandbox or basic test account.
The post calls this unusual compared with other ecommerce platforms, describes Magento as difficult to work with and thinly used, and adds that Shopify is displacing legacy platforms.
The request is openly framed as a big one.
For brands planning a campaign
For a brand on Magento, app and integration work in that environment brings testing friction and dependence on outside developers, which is worth pricing into a project plan. For Shopify-based work, this provides background on Wyatt's focus rather than a campaign tactic.
A short, direct look at the engineering behind the Shopify tooling the rest of the work depends on.
Read more
Read it to judge how much custom development sits behind the feed and tracking features rather than assuming an off-the-shelf setup.
Wyatt says they spent two days working through every available debugging tool to get Shopify's GraphQL, OAuth and Remix authentication process working, and states plainly that those two days were lost to it. The post complains about the platform's authentication design rather than offering a technical walkthrough.
For brands planning a campaign
Custom Shopify integrations have limited direct impact on campaign planning, but authentication and app setup are real cost lines.
When buying feed tooling, ask whether it is a supported app or bespoke development.
Pair this with the Onefeed posts to distinguish product features from custom build work.
Separate low-traffic products from Performance Max
Wyatt describes using product-feed labels to move items with fewer than two sales and fewer than 1,000 impressions in 30 days into separate Shopping campaigns.
Read more
They then divide those products into groups such as best sellers and other low-traffic items, with different priorities and return targets. Products that start selling profitably can move back into Performance Max.
Use sales, impression, conversion, and, where available, margin data to create product groups. Wyatt also suggests using basket-level tracking to identify products gaining traction through other channels, and a separate low-bid fallback campaign for queries excluded from other campaigns.
For brands planning a campaign
For a brand assessing a Shopping and Performance Max strategy, this is a concrete proposal to test whether products with little visibility in Performance Max can gain incremental traffic and revenue through segmented Shopping campaigns.
Agree on product thresholds, budgets, and return targets first, then compare results with a suitable baseline.
Ask for account-specific evidence to assess the size of the claimed lift.
Wyatt says Performance Max may give newly added products little exposure before they have sales.
Read more
They suggest labeling products by creation date and running a Shopping campaign for new items until they begin selling, after which they may be eligible for Performance Max.
Apply a feed label based on product creation date and use it to control campaign eligibility.
For brands planning a campaign
This approach is relevant to brands with frequent launches or large catalogs where new products risk receiving little initial traffic. A campaign discussion could define the test period and success signals for moving products into other campaigns, then check whether the extra exposure produces useful sales rather than just impressions.
Diagnose ecommerce ad performance with campaign-level signals
In a post about a Onefeed feature, Wyatt lists signals for reviewing Google and Bing PPC campaigns: changes in previously strong products, budget pacing, positive and negative performance factors, out-of-stock products, shifts in Performance Max traffic mix, and changes in impressions, clicks, and costs.
Read more
They present the aim as finding causes of performance drops without manually searching through multiple reports.
For brands planning a campaign
Brands considering PPC reporting or optimization support can assess whether an audit connects these signals to specific decisions—for example, separating a product availability problem from a budget or traffic-mix change. Ask for evidence of the feature's accuracy and results.
Wyatt describes Onefeed tracking that records basket contents and customer identifiers, distinguishing new from existing customers.
Read more
They say the platform reports customer and repeat-purchase measures against lifetime and recent-period comparisons, including products associated with new customers.
For brands planning a campaign
This is worth evaluating when a brand wants to understand whether advertising contributes to customer acquisition or repeat purchases, not only immediate conversions. Confirm what consent, data handling, and attribution arrangements are required, and how the reported measures would inform campaign decisions.
Practical product-development and implementation updates
Two posts offer a different kind of evidence: Wyatt describes resolving a difficult Shopify authentication and development issue, and asks for access to a Magento test store to complete testing of an app.
Read more
These posts show hands-on discussion of ecommerce software implementation and testing, rather than a worked advertising case study.
For brands planning a campaign
Brands evaluating related technical work can use these posts as conversation starters about platform integrations, testing environments, and release readiness.
Wyatt's posts point to a working involvement with Onefeed: they describe using the platform to build product-feed labels and write in the first person about features "we've added" and a Shopify app "we've built".
Read more
Ask whether that involvement is ownership, employment or a client engagement. Their other brand references are platform mentions rather than commercial deals: building an app for Shopify and asking readers for help testing a Magento app after saying a developer let the work down.
Onefeed
Disclosed partnership
Wyatt uses Onefeed to create custom labels that split low-traffic and best-selling products between Shopping and Performance Max campaigns.
Read more
They say basket-level tracking and conversion data come through the platform's tags. They also write about the product in the first person — "we've now added a summary of the deep dive information that we gather through the APIs" and "a critical element of the Shopify app we've built" — indicating involvement with the company rather than simply using the software.
Richard Wyatt's posts are a working record of running ecommerce Google Ads, with the centre of gravity in the split between Performance Max and Shopping.
Read more
The fullest statement of the approach is the long walkthrough of using custom feed labels to move products that have almost no impressions out of Performance Max and into their own Shopping campaigns, then letting them back once they sell with a good return; start there.
Two posts describe the Onefeed features built around the same idea: basket-level tracking that separates new from existing customers and reports repeat purchase and lifetime measures, and a campaign-level analysis summary for working out why performance dropped before changing strategy.
Two shorter posts cover the build side — authentication work on a Shopify app and a public request for a Magento test store.
Readers who follow feed-level segmentation, Performance Max structure, or ecommerce ad measurement will find the most depth in the Shopping and Performance Max post.
The Onefeed posts are shorter feature descriptions rather than worked case studies; ask for worked case studies if those are central to your decision.
Davies writes short LinkedIn posts on diagnosing Google Ads accounts for ecommerce sites, including a method for deciding whether slower sales come from weaker market demand or from more aggressive competitors.
Read more
That is the same practical, account-level Google Ads territory Wyatt covers, approached from the reporting side rather than the feed side.
Davies reads account signals — daily impression share, brand versus non-brand clicks and revenue — and points readers to a third-party script for the brand decision. Wyatt starts further down, at the product feed, and splits items into campaigns by sales, impression and margin labels.
Tesselaar addresses the same problem Wyatt works on: getting Performance Max to compete for cold, non-branded traffic and profitable products rather than whatever converts most easily.
Read more
Both treat account and data structure, not bidding tweaks, as the explanation for why Performance Max results vary between accounts.
Tesselaar's structure sits at campaign and audience level — a dedicated branded search campaign, asset groups split by margin, CRM and lifetime-value lists, conversion value rules, and search themes used as a diagnostic — and reasons in profit terms such as break-even return and cost of new customer acquisition. Wyatt's equivalent move is product-level: feed labels decide which items may enter Shopping and when they return to Performance Max.
Gregory covers the same reader problem from the brand side — ecommerce advertisers whose results break down as spend grows — and like Wyatt argues that the fix is structural, checking margin, acquisition cost, lifetime value and funnel coverage before budgets increase.
Read more
Both write in short, method-shaped posts on LinkedIn.
Gregory's frame is cross-channel: Meta and Google overlapping, cold-audience funnel gaps, weekly creative testing and landing pages, plus an offer of free account audits. Wyatt stays inside Google, and where Gregory maps the funnel, Wyatt maps the product feed into campaign eligibility.
The only measured audience figure in the posts here is a LinkedIn following of 2,256, observed on 11 October 2026.
Read more
That is small for a reach-led campaign and typical of a specialist practitioner account, making the placement signal weak on scale and stronger on relevance.
The content suggests an audience of in-house ecommerce marketers, feed managers and agency performance staff working on Google Shopping and Performance Max, but that is an inference, not measured audience composition.
The content focuses on operational work: feed segmentation inside Onefeed, splitting low-impression and low-sales products into their own Shopping campaigns, and reading basket-level and repeat-purchase tracking.
A brand assessing Richard Wyatt for a Shopping and Performance Max engagement should view the audience as a peer practitioner network rather than a consumer or executive audience.
That suits a campaign seeking demonstrated method and a credible technical voice more than raw impressions.
The posts do not report audience geography, age, gender, employer, job title or seniority.
None of the five posts carries engagement figures such as reactions, comments or shares that would show which topics land.
Before committing budget, ask for recent results from comparable work—feed-segmentation engagements of similar depth, on the same ecommerce platforms and delivered in the same formats—and any audience survey or platform analytics covering geography and job function.
If the campaign depends on reaching senior decision-makers rather than practitioners, ask whether documented audience data speaks to that.
Richard Wyatt's five available posts show hands-on ecommerce advertising work and a software product, Onefeed, which they discuss building and using: feed labels, campaign segmentation, basket-level tracking, a Shopify app, and a campaign-analysis feature.
Read more
The posts do not describe a sponsorship or paid-partnership program, sponsored placements, audience packages, rates, deliverables, exclusivity terms, or an application process.
This points to a consulting or software-led relationship rather than a media buy.
For a campaign involving Google Shopping and Performance Max, contact them through the LinkedIn profile already listed and ask about rates, deliverables, exclusivity, and how to scope the engagement; no standard partnership route or deliverable is published.
An ecommerce brand running Google Shopping and Performance Max wants a specialist to structure feed segmentation and campaign splits so low-visibility products can earn their own traffic and revenue.
Read more
Reasons
The most complete published example tackles this exact problem. Wyatt describes Performance Max concentrating spend on products that already sell, leaving other items with almost no impressions. They lay out a routine to pull those products out: a Onefeed custom label for items with fewer than two sales and fewer than 1,000 impressions in the past 30 days, a Shopping campaign restricted to those items, and, according to the post, around 95% of the feed landing in that campaign.
The same piece outlines a campaign structure a brand can adopt or test: best sellers with low traffic on aggressive ROAS targets, everything else on target ROAS, and further tiers when the client supplies margin data. It also describes returning products to Performance Max only once they sell at a good ROAS, plus a separate Shopping campaign for newly created items labelled by creation date.
The material also describes a complementary diagnostic capability. The Onefeed deep-dive summary lists six per-campaign checks for diagnosing a performance drop: products that stopped selling over the previous 7 or 30 days, budget pacing, spend and performance factors with suggested actions, out-of-stock reporting, Performance Max traffic-mix shifts, and changes in impressions, clicks and costs. A Shopping and PMax engagement usually needs diagnosis as well as construction.
The material describes the tracking layer behind the feed work: basket-level data, customer ID, and new-versus-existing customer flags, alongside lifetime value, last-12-month comparisons, repeat rates and the products that attract the most new customers. Feed labels and re-admission rules depend on this customer-level data, so the material presents them as one system rather than separate services.
The audience evidence supports a method-and-technical brief rather than a reach brief: 2,256 LinkedIn followers. That is a small audience for reach, but consistent with a peer practitioner following of feed managers and in-house or agency performance staff — the people who would scrutinise this work.
Conditions
The example reports an extra 50–60% revenue from running Shopping in parallel. This is the author's outcome from one example, not an independently verified benchmark. Agree a measured baseline and test window in the account; treat the figure as a hypothesis, not a target.
The sales and impression thresholds are described as experimental. Set them against the client's catalog size and volume before launch.
There is no published partnership route, rate card or standard deliverable. Agree scope, fees and deliverables directly through the LinkedIn profile listed on this page. The material describes consulting and software-led work (Onefeed, the Shopify app, bespoke builds), not a bookable media package.
The posts show no client case studies, named accounts or audited results beyond the author's own screenshots and figures. Ask for recent comparable engagements.
Verdict
Strong fit. This is the brief in the available material where the published work maps directly onto the deliverable.
When to choose a specialist
A consumer or lifestyle brand wants a reach-led sponsored placement, with impressions, audience scale and a defined deliverable package booked like an influencer partnership.
Read more
Reasons
The posts describe no sponsorship or paid-partnership format, rate card, deliverables, exclusivity terms, audience package or application process. Ask whether a custom partnership is available; there is no standard creator package to book.
The only measured audience figure is 2,256 LinkedIn followers. The posts provide no individual post engagement counts, platform demographics covering geography, age or gender, or audience survey. Ask for audience data before pricing a reach-led campaign.
The published material is operational and technical: feed labelling, campaign priorities, ROAS tiers, basket-level tracking, authentication debugging and a request for a Magento test store. It offers no shopper-facing content, product storytelling or lifestyle material for a consumer brand placement.
The audience evidence points to in-house ecommerce marketers, feed managers and agency performance staff — a peer practitioner network, not a consumer awareness audience.
Also weak for
Not a fit for a campaign requiring verified enterprise-scale outcomes, a keynote or executive thought-leadership slot, or work outside Google and Bing PPC and the Onefeed/Shopify/Magento app stack. The posts show no case studies or named clients; ask for evidence of relevant outcomes and a dated work history.
Verdict
Poor fit. The posts point away from this brief on three grounds: no documented offer, no measured scale, and content aimed at practitioners rather than shoppers.
A worked campaign example
Angle
Run a two-phase parallel Shopping test built on feed labels, using the re-admission rule to protect Performance Max efficiency rather than replace it. The engagement question is narrow and answerable: how much revenue sits in products that PMax never shows, and at what return.
Brief
An ecommerce brand with a large catalog runs Performance Max only, sees weak coverage across most products, and wants to recover incremental revenue from items PMax is not showing.
Measures
Incremental revenue and return by campaign tier, measured against a pre-launch baseline period with the same catalog and budget.
Share of feed items moved out of Performance Max and share later re-admitted, tracked as a coverage measure rather than a performance claim.
Total account spend and return as a guardrail, weighing gains in the separated campaigns against PMax performance rather than reporting them in isolation.
Next step
Agree scope, fees and the test window directly through the LinkedIn profile on this page. Ask for recent comparable feed-segmentation engagements of similar depth. Treat the reported 50–60% uplift as the hypothesis to test, not the expected result. Ask for the client's margin data upfront, since the example's ROAS tiering depends on it.
Steps
Phase 1 — confirm the data layer before changing campaigns. Verify that basket-level tracking, customer ID, and new-versus-existing customer flags are live on the client's own store. Feed segmentation and re-admission logic depend on this, making it a prerequisite rather than a parallel workstream.
Build four custom labels in the feed tool: low traffic (items with under two sales and under 1,000 impressions in the past 30 days); top sellers (90-day conversions plus any best-seller list the client supplies); new items (based on product creation date); and re-admitted (products that start selling at the target return).
Phase 2 — launch Shopping campaigns against those labels, split three ways: best sellers with low traffic on aggressive ROAS targets; everything else on the client's target ROAS; and, where the client shares margin data, further tiers by required ROAS. Add a low-bid, low-priority fallback campaign to collect negative keywords added elsewhere.
Where the account regularly launches new products, run a separate Shopping campaign restricted to the new-items label to give those products exposure before PMax will show them.
Apply the re-admitted label only once products are selling at a good return, then let them back into Performance Max. This keeps the parallel campaign from cannibalising efficient spend.
Set a fallback diagnostic routine using six campaign-level checks: products that dropped away; budget pacing; spend and performance factors; out-of-stock items; Performance Max traffic-mix shifts; and changes in impressions, clicks and costs. This helps explain a dip during the test rather than guess at its cause.
Choose Martin Davies when the engagement starts with diagnosis rather than feed construction: a brand whose sales have slowed and which cannot tell whether the cause is weaker market demand, a competitor or its own panic changes.
Read more
The published advice is to run an impression share report by day over the past 30 days, or from before the dip: steady impression share points to lower demand in the market, while falling share points to competitor activity.
The strategy should change accordingly.
A second post addresses whether brand traffic should be excluded from Performance Max.
It describes a third-party brand script that reports brand versus non-brand clicks, conversions and share of brand revenue over time, so a brand can see whether brand traffic is absorbing too much of its PMax delivery and move it into search campaigns with a brand negative instead.
The tradeoff is a narrower mechanism.
Davies' posts cover diagnosis and the brand-traffic question, but they do not describe feed-level product labelling, campaign priority tiers or a re-admission rule for products that start selling.
If the need is specifically to recover revenue from items Performance Max never shows, someone else would still need to design the feed segmentation and campaign split; Wyatt's example works through that loop step by step.
Choose Robin Tesselaar when the decision hinges on whether the account is actually making money.
Read more
Two posts argue that most Google Ads accounts look profitable without being profitable: ROAS is revenue divided by ad spend, before cost of goods, shipping, returns and fulfilment, so a 4x ROAS campaign can lose money and a 2x campaign can be the most profitable one.
The material gives a break-even rule (target ROAS of 1 divided by gross margin, so 35% margin means break-even at 2.86x), names new customer acquisition cost as the metric that reflects reality, and lays out three ways ROAS misleads: it ignores margins, rewards warm returning buyers, and hides diminishing returns through blended averages.
A companion post sets out the account structure that makes Performance Max behave: a dedicated exact-match brand search campaign to force PMax onto cold traffic, asset groups split by margin, CRM and lifetime-value audiences fed in as signals, conversion value rules so a $200 sale is not treated like a $20 sale, and search themes used as a diagnostic rather than a growth lever.
Use this framing when a brand needs to test whether a Shopping-and-PMax restructure is producing genuine incremental profit or is being carried by repeat buyers.
The tradeoff is scope inside the platform rather than across the feed.
Tesselaar's posts say which structure and profit metrics to use and warn that reported gains may be illusory, but they do not cover identifying and pulling out the large share of catalog items that never receive impressions, or the feed-labelling tooling to do it.
If the core problem is product coverage in a large catalog, Wyatt's routine supplies the label thresholds and campaign tiers that these posts assume somebody else has handled.
Choose Josh Gregory when a brand needs more than Google Shopping: a brand scaling paid media past roughly £25–30K a month across Meta and Google and seeing ROAS decline, acquisition costs climb and results swing.
Read more
The posts describe where that money usually goes wrong: no funnel built for cold audiences once warm buyers run out, channels planned in isolation so Google and Meta overlap and waste clicks, and no lifetime-value or offer strategy to justify paying more for a first purchase.
They then set out the structure that supports scale: weekly creative testing cycles, margin and acquisition-cost checks before budgets rise, first-time-buyer offers, a mapped funnel from awareness to purchase, and landing pages built for cold clicks.
A second post explains the dilution that follows spending too quickly, with the algorithm broadening targeting and conversion rates falling.
It also offers a small number of free Google Ads audits over the following month, giving a brand an explicit route to an initial diagnostic.
The tradeoff is breadth against depth on the specific feed question.
Gregory's posts operate at the level of channels, creative, offers and funnel, and do not set out the label thresholds, campaign priority tiers or re-admission rule at the heart of Wyatt's engagement example.
Choose Wyatt for a narrow Shopping and Performance Max feed-segmentation brief; choose Gregory when the same brand also needs Meta, creative testing and offer strategy coordinated, accepting less published detail on the Shopping feed mechanics.
Five LinkedIn posts by Richard Wyatt cover a Google Shopping and Performance Max feed-segmentation walkthrough; two Onefeed product and feature announcements; a request for a Magento test store; and a post on Shopify authentication work.
The posts have no publication dates, so recency cannot be assessed.
No transcripts, longer articles, campaign data, or independent verification of reported results appear in the posts here.
The posts show no documented sponsorship, rate card, or paid-partnership format.
Whether the name refers to someone with books, shows, or earlier work remains unclear, so this may not be a complete name-search profile.
