Test distinct ideas and incrementality, not just asset count or platform score.
Shopify compares visitors who reached product pages; organic search still supplies much more traffic.
Nest reports Meta CVR +7.2% and CPA -3.7%; these are its own seven-day client-set trends, not market benchmarks.
Performance Marketing World reports that Nest has brought its AI-led Hummingbird creative-variation product, already available for Meta, to TikTok. The publisher exposes only an abstract, not an independent lift test or implementation detail.
Interesting if it reduces the time from signal to usable edit. Ask for holdout results, variant provenance and the human review loop before calling it an AI performance win.
Meta's 2026 Creative Edition recommends partnership ads, a mix of static, video and Reels with distinct messages, catalog video and Advantage+ variations. It says Advantage+ catalog ads had 16.4% lower CPA than static ads in a 2024 causal-inference analysis of 108,000 accounts, and format display options drove 23% more conversions in a separate 2025 announcement.
The useful bit is the production calendar: lock creator permissions and test automated edits before peak, not during it. These are Meta's own studies across selected products and cohorts, not a forecast for your account. Track incremental contribution, creative fatigue and margin by variant; 'more assets' is not a strategy if they repeat the same idea.
Creative Edition primary PDF: Meta's guide . Meta's own evidence is vendor-reported; the 2024 study and 2025 format claim are different analyses.
Pinterest's festive guide introduces a campaign optimisation score based on set-up, creative mix and signal strength. It suggests at least three formats per ad group and five or more distinct ads; Performance+ can remix assets. Pinterest says higher scores correlate with better results, not that raising a score causes them.
Score is a checklist, not an outcome. The platform rewards its own automation, broad audiences and stable budgets. Test genuinely different use cases and product sets, then judge on incremental sales and contribution rather than trying to win a green badge.
Marketing Week's Jonathan Knowles argues that CEOs with a finance mindset can over-optimise current margins and underfund future demand. He suggests pricing power, defined as raising price without losing demand or share, as a bridge between marketing and finance.
A sharp provocation, not proof that every cost-conscious CEO is wrong. Bring a measurement bridge: current payback and contribution alongside price elasticity, penetration and customer value over time. Brand arguments need a financial test, not a demand for faith.
Shopify says Q2 2026 AI-referred visitors who reached product pages converted about 80% better than organic-search visitors who reached product pages. More than half of AI referrals land directly on a product page, against about 20% for organic. AI sessions grew 197% year on year, but organic still sent more visits than all tracked AI platforms combined and grew 12%.
This is not 'GEO beats SEO by 80%' or a causal lift. It compares selected product-page visits with very different intent, category mix and attribution, without absolute session counts here. Fix feed accuracy, PDP clarity and real reviews, then inspect the channel by category and new-customer contribution. Do not move a material search budget on this stat.
Cross-check: Shopify's Q2 category analysis gives the same 80% product-page comparison and makes clear that organic remains far larger.
A 17-second vertical Nest Studio film follows a hike-to-swim outing. The creator shows a foldable cup and Seawhisper shorts, describing UPF 50 protection and quick-dry recycled ripstop for a trip that combines hiking and swimming. The benefit sits inside a real-use story rather than a catalogue list.
Good product demonstration, though the opening must still earn a cold viewer's first two seconds. Test the trip narrative against a direct problem/solution cut and check whether the claims remain readable with sound off.
ONS figures reported by InternetRetailing show UK retail sales volumes up 0.5% month on month in August and 0.9% across the three months to August. Online share rose to 28.8% from 28.4% in July.
A rebound from July is not permission to write a buoyant peak forecast. Look at category, promotion and inventory before extrapolating the national print.
The BRC-Opinium September measure put expectations for the economy at -34, from -28 in August; personal finances at -15, from -9; and expected retail spending at +5, from +8. These are forward sentiment balances, not observed sales.
Both this and August's sales rise can be true. Plan offers for value-conscious shoppers, but do not translate a confidence survey directly into a revenue haircut.
An IPA/Opinium survey of 2,000 UK adults in May found 49% said they would not cut back on anything this Christmas, while 45% expected more financial anxiety. Savings cues mattered: free delivery ranked 37%, discounts 24%, price match 20% and free returns 20% among retailer-choice factors.
Intent is not the till. Build a value story that protects margin: useful bundles, delivery certainty and trusted returns, rather than universal discounting dressed up as generosity.
Fraud vendor Ravelin's survey of more than 6,000 UK, French and German consumers found 37% of British respondents had engaged in, or planned to engage in, refund abuse. The article says 98% of those who tried it reported getting away with it; that is self-report, not an audited fraud-loss rate.
Separate genuine returns from false claims. A blanket clampdown can punish good customers and hide fulfilment faults. Measure claims by item, delivery evidence and repeat behaviour before changing the policy.
A Nosto-commissioned survey of 2,000 US and UK consumers found 59% often leave stores over poor onsite search; 40% say they are likely to try ChatGPT or Claude after irrelevant results. This is stated intent after a bad result, not 40% of all shoppers migrating.
Fix zero results, synonyms, filters and stock visibility before buying an 'agentic' search pitch. Search exits tied to query and basket value will tell you where the money leaks.
Retail Dive reports a SmartCustomer survey of nearly 1,200 US consumers: a third said they had regretted a purchase based on an AI recommendation, while 89% did not fully trust recommendations. These are survey responses, not measured return rates.
Discovery without confidence is a weak funnel. Product truth, proof and a painless human check still matter at the moment of purchase.
The linked BBC report describes an EU proposal for under-13s to be barred and 13- to 15-year-olds to use parent-linked limited accounts, with accounts from 15. The newsletter's shorthand 'under-15s' misses the tiers and its policy status.
Don't change targeting or legal copy on a proposed rule. Have the youth-audience exposure inventory ready and let legal track the actual text and implementation.
A Drum/iResearch survey of 300 marketers found 77% of agency leaders thought they delivered what brands wanted, against 23% of brand marketers. The article also says nearly 80% of brands expected brand strategy in-house by 2030, compared with 49% of agency leaders.
A useful prompt for the agency review: agree what decisions the partner owns, what evidence would change a plan and what the in-house team must supply. But the sample is small and split across regions; the gap is not your agency's score.
Jon Loomer documents a test 'Landing Page Experience' enhancement that uses AI to build a visual page from the advertiser's site. Meta says people tapping an ad may see it instead of the website; Loomer says under 5% of impressions are allocated to test features and the option can be switched off.
Audit the setting, especially regulated claims, price, analytics and consent. A better platform CVR can mask lost first-party data or a broken post-click experience. Test with controlled destination and actual downstream profit.
Social Media Today reports Shopify's planned Shop Pay checkout integration with Meta's Muse agent, while Amazon has blocked Muse bots under its automated-shopping rules. The article speculates on the effects of agentic price comparisons.
Distribution is becoming a platform negotiation, not merely a feed upload. Treat predictions of a wholesale commerce shake-up as speculation; verify actual adoption, order economics and who owns the customer.
Meta says from September 2026, advertisers no longer need matching Threads and Instagram usernames and can advertise from a native Threads profile without Instagram. Rollout is gradual.
Less account friction, not a reason to paste Instagram creative into a text-led feed. Check availability and run native creative against a holdout if the audience matters.
Search Engine Roundtable shows Merchant Center notices saying eligible Shopify stores matched to Merchant Center are included in checkout through Google AI Mode and Gemini by default; an opt-out is described in Shopify's Sales channels > Agentic. The report is based on merchant notifications and screenshots.
Check your own merchant setting, product eligibility, tax/shipping accuracy, attribution and customer data path before treating this as incremental sales. An automatically enabled channel still deserves a commercial owner.
TikTok's creative-pipeline guide advises at least six creative videos for a new Smart+ campaign, with fresh assets added as it runs. It suggests creator content, catalog assets and image-to-video; it cites a 56% higher CTR and up to 18% lower CPA for Carousel Collage versus traditional carousels, without enough test detail to generalise.
Six is a platform heuristic, not a magic learning threshold. Make six different hypotheses, not six crops of one ad. Separate hook, offer and product testing, and watch marginal spend per winning concept.
Social Media Today describes the score as largely aligned to Performance+ adoption, creative variants, broad targeting and steady budgets. Pinterest recommends three formats and five distinct ads within an ad group.
Cross-read with the Pinterest guide above: a higher score correlates with performance, and also encourages use of Pinterest's automated products. Judge against your own lift, not the score.
An independent technical post describes an OpenAI ad collector cookie and cross-site event requests on advertiser sites, with tests on a phone and a large pixel inventory. It also discusses consent and limits. This is one researcher's observed mechanism, not a full audit of every user's configuration.
The real operator question is consent, data-sharing terms and whether agent-led commerce can preserve customer trust. Don't assume either total anonymity or universal tracking from a single test; audit the actual pixels and privacy path.
ASOS's pre-close update showed low-single-digit Q4 GMV growth, but full-year GMV still fell 5%. Adjusted EBITDA rose more than 25%, in the upper half of £150m-£180m guidance; adjusted gross margin exceeded 50%. Active customers were 16.4m at year end, broadly flat on 16.5m at H1.
The recovery is credible, but do not turn one positive quarter into a fully repaired demand engine. Watch repeat buying, full-price sell-through and cash generation when the full results arrive.
Debenhams Group is considering concessions for Boohoo and PrettyLittleThing, following Karen Millen's return to physical space. Group GMV rose 1.8% in the half to 31 August; its marketplace was 38.9% of GMV, from 32.7% a year before.
Concessions are a distribution experiment, not a brand resurrection. The useful test is incremental reach and contribution after rents, returns and channel cannibalisation.
H&M's Q3 operating profit rose 23% to SEK6.04bn as margin moved from 8.6% to 10.6%, while sales rose 1% in local currencies. The gross-margin gain includes a roughly 1.6-point favourable tariff/import-cost effect.
Good cost discipline, but strip out the one-off before declaring structural margin progress. The demand side is still barely moving.
Retail Week says Boots will open a TikTok Shop storefront on 28 September with selected No7 skincare and beauty ranges. The accessible article stops after its opening section and gives no trading targets.
A strong brand can make live commerce feel less like a bargain bin. The test is creator fit and incremental customers, net of fees and returns, rather than launch-day views.
Tilray's post-acquisition campaign uses lines including 'A little less allegation' and 'No more cunning stunts. Just great beer.' City AM reports allegations about the prior workplace culture and the new owner's pledge to change practices.
Memorable, risky work. A promise about conduct is only as good as what employees and customers see next; a clever line can become a receipt against you.
London role across paid search, affiliates and social, with forecasting, cross-channel strategy and team leadership. The linked posting does not set out a verified salary here.
Potentially a sizeable remit, but clarify decision rights across brands and the performance budget before treating the title as scope.
London-area role reporting to the CMO, combining paid acquisition, affiliates, partnerships and influencer marketing. The posting calls out blended CAC, payback, contribution and creative production across markets.
The most integrated brief of the four. It explicitly joins creators and paid media rather than treating influencer as a side department; ask how much P&L control comes with it.
Malmesbury-based fixed-term role focused on direct-to-consumer growth and digital touchpoints in a global organisation. The posting describes a campus working model.
Strong brand and direct-channel learning, but a 12-month term and campus location are real constraints. Confirm scope and decision speed.
London role to own paid growth across regions and channels for a beauty and wellness platform, with forecasting and collaboration with design and influencer teams.
The marketplace/platform economics may be more interesting than the title. Ask how supply-side acquisition and consumer demand are measured separately.
Will says platform trends compare the seven days to the preceding Thursday. No sample size, spend weighting, vertical mix or baseline is supplied in the email. These are sender-reported internal figures, not independently verified.
A lower CPA beside lower AOV does not settle contribution. Check revenue per acquisition, discount depth and category mix before celebrating the Meta week.
The actionable bit is to secure creators now and test automated catalog and format changes before the peak window. The claimed 16.4% CPA reduction and 23% conversion gain are Meta-reported studies, not a ready-made forecast. The right experiment holds budget, audience and measurement steady while varying a real creative idea. If margin and incremental orders do not move, the extra variants are just more files.
Read the guide