Gil David shares practical advice for direct-to-consumer ecommerce brands on paid advertising, marketing measurement and Shopify analytics, drawing on client examples and agency work.
Most useful for explaining paid ad and Shopify measurement changes and what DTC brands can check next.
Best for
Fits DTC Shopify brands needing trusted analytics for conversion rate, scaling, and counting changes.
Not for
This does not fit creator sponsorship, product placement, rate cards, exclusivity terms, or broad consumer awareness campaigns.
Key takeaways
Shopify bot filtering from 21 September can lower conversion rate without changing orders or revenue.
Meta attribution rank correlation helps order campaigns for scaling or cutting, not exact ROAS.
Meta defaults are an ongoing opt-out job; automated AI creative can be off-brand or outdated.
Check first: Ask whether any creator sponsorship format, rate card, or partnership terms exist, or only agency work.
Main ideas
A jump in reported performance is usually a counting change, not new buyers · Choose your attribution setting deliberately instead of inheriting the default · Meta's defaults are an ongoing opt-out job, not a one-time setting
Gil David shares practical commentary on paid advertising and measurement for direct-to-consumer ecommerce brands.
The posts show a recurring focus on separating real business performance from changes in platform reporting: for example, explaining how Shopify’s bot filtering can alter conversion-rate calculations without changing orders or revenue, and how Meta attribution settings can affect campaign comparisons.
David also examines platform defaults and wider market changes, including automated ad creative and election spending that can push up ad costs.
Follow David for plain-language explanations of what changed, what the numbers do and do not show, and what advertisers can check next.
David connects platform news to decisions advertisers actually make.
Rather than treating a higher reported ROAS or conversion rate as proof of improvement, the posts ask whether the measurement changed, whether the result appears in store orders and revenue, and whether an outside event explains the shift.
The Shopify bot-filtering post turns that reasoning into concrete steps: use 21 September as a new comparison point and focus on orders, revenue and new customers when judging performance.
The Meta attribution post explains that rank correlation helps order campaigns for scaling or cutting; it does not make reported ROAS accurate to the penny.
There are exceptions to the measurement-and-operations focus: the AI creative post warns about off-brand or outdated automated ads, while the short hiring and throwback posts offer less analytical guidance.
The supplied material supports a profile of David’s LinkedIn commentary, but does not establish signature books, shows, or earlier work beyond these posts.
Gil David writes about paid social and ecommerce analytics for direct-to-consumer brands, mostly from the position of an agency operator running client accounts.
Read more
The idea that runs through nearly everything he publishes is that the number a platform shows you is a measurement decision rather than money in the bank.
Platform changes, attribution defaults and outside events all move reported performance without moving demand, so his habit is to check what changed, when it changed, and what the business actually earned before reacting.
He applies the same standard to AI: the interesting question is not whether it is fast but who approved the output, who is liable for it, and whether it saved anyone real time.
A jump in reported performance is usually a counting change, not new buyers
Gil David's central argument is that reporting is what brands scale on: cost caps, ROAS targets and year-on-year comparisons are all set against metrics that platforms can quietly change.
Read more
When Meta or Shopify alter how something is counted, the headline number can move sharply while the orders in the bank stay the same.
His method is consistent: take a baseline of the affected metrics before the switch, record the switch date in reporting, stop treating the periods either side as like for like, and cross-check one platform's numbers against another source.
He is not against the changes.
On the Shopify bot filter he says it is the first platform update in a while that made data cleaner rather than murkier.
Shopify began filtering bot sessions out of Analytics by default on 21 September.
Orders and revenue are untouched, but sessions, conversion rate, add-to-cart rate and bounce rate are affected.
One client store showed Singapore at 5,200 sessions, up from one, all direct traffic, zero orders, for a market the store does not ship to, and Shopify flagged roughly another 5,000 Singapore sessions that month as bots, so the filter caught about half.
In the same spirit, Meta's on-device conversion measurement, rolling out from 12 November and switched on by default for datasets already sending events, uses details like an email address at checkout to claim purchases that never previously reached Meta, which can lift reported ROAS two weeks before Black Friday.
He advises marking the date and asking how those on-device conversions are deduplicated against pixel and CAPI events, a question Meta had not answered.
The two posts treat platform measurement changes as a recurring category rather than one-off news: he frames the Meta and Shopify updates as things to build into reporting hygiene, and says he cleans up double-counted purchase events routinely.
Choose your attribution setting deliberately instead of inheriting the default
Gil David argues that attribution settings should be picked for how well they rank campaigns, because scaling and cutting decisions are about ordering, not about a ROAS figure being accurate to the penny.
Read more
The figures he discusses come from geo holdout tests run by Haus, whose team he names as Eric Miller and colleagues, so the numbers are their research rather than his own.
He highlights two findings: adding 1-day view to 7-day click makes the match to measured lift worse, so a very common default is less reliable than plain 7-day click, and incremental attribution comes out slightly ahead of both.
He also pushes back on the view that view-through conversions are entirely fabricated, while still objecting to stacking every view conversion on top of clicks.
Rank correlation against measured lift: incremental attribution 0.79, 7-day click 0.73, 7-day click plus 1-day view 0.68, 1-day view 0.40, where 1.0 is a perfect match.
He explains that 0.40 is not zero, so some view-through conversions reflect real incremental purchases, but treating each one as caused by the ad and adding them to clicks overstates the effect.
He states the limits himself: Haus calls it an early read, it covers new-customer conversions only, and the ranges for incremental attribution and 7-day click overlap.
Meta's defaults are an ongoing opt-out job, not a one-time setting
Rather than treating automated ad enhancements as free upside, Gil David's working rule is to switch them all off and then re-enable one at a time to test whether it adds anything incremental.
Read more
The reasoning is about who controls what gets tested: if the platform injects copy rewrites, music, generated backgrounds or automatic recommendations, it becomes harder to say whether a change worked.
He notes that some settings only appear in advanced previews and can creep back when an ad is duplicated, which is why he treats this as an ongoing habit rather than a checklist that stays done.
He is not uniformly negative: automatic events should be left on where a CAPI setup already sends the needed data and nothing is double-counting, since more good-quality data for the platform is rarely a bad thing.
A list of settings his team switches off: Advantage+ creative enhancements such as copy rewrites, music and generated backgrounds; account-level creative feature testing, so enhancements turned off elsewhere do not return, including on duplicated ads; and auto-applied performance recommendations, so Meta cannot pause or combine ad sets unattended during peak trading.
Placement exclusions are flagged as one that cannot be switched off, with up to 5% of budget spendable on placements a brand has excluded.
The trigger for the post was Meta switching product browsing on for existing ads unless advertisers opted out, which could send a tap to a Facebook or Instagram product page instead of the advertiser's own site, and which Meta supported with a 3% conversion figure that did not specify purchases.
When costs rise, look outside the ad account before blaming the ads
Gil David repeatedly points out that auction costs are set by demand from other advertisers, not only by a brand's own creative and targeting.
Read more
A CPM spike can come from reserved inventory deals, an election or a platform change landing before peak, and his diagnostic order is to check the calendar and the wider market before rebuilding campaigns.
On reserved deals he adds the second-order question: if large brands buy inventory in advance, what happens to everyone else bidding on the same day?
He is explicit that the answer is not knowable from outside Meta.
Meta's Ads Super Burst lets large advertisers reserve a window of three to 24 hours on a single day at a price agreed before other bids are placed, with delivery promised within 5% of forecast reach, frequency capped at one impression per person, and category exclusivity available on some US dates.
He frames it as a television upfront compressed into a day and asks what happens to auction CPMs on dates like Black Friday, when several big brands may have locked up tens of millions of people before the day starts.
On a separate occasion he attributes a 25% plus rise in US CPMs to midterm elections the following week, citing a New York Times figure of $3.9bn in election ad spend, and closes with the instruction to always look outside your ad accounts for context.
AI in advertising is an approval and liability problem before it is a speed story
Gil David writes about AI as something done to advertisers rather than by them.
Read more
His concern is that platforms can generate or edit creative without the advertiser asking, which puts brand presentation and responsibility in the hands of a system nobody reviewed.
He connects that to platform terms that make advertisers responsible for what their AI produces, whether or not they knew about or approved it.
Separately, he argues against the assumption that AI saves him time: tasks get done faster, but the freed hours fill with more work, and reviewing and filtering AI output is more exhausting and more mundane than doing some of it himself.
He makes a point of doing some work manually to avoid letting critical thinking go stale.
In one week he describes the National Trust apologising for obviously AI-generated TikTok creatives that ran from its account without its knowledge and pausing advertising on the platform; a client of his own where TikTok added ten AI variants to existing ads, including irrelevant stickers and promotions that had ended months earlier, with no notification; and Just Eat being criticised for automated images added to menu items that lacked photos. He characterises the outputs as what a toddler might imagine a chip muffin looks like, and describes the client experience as like Meta's Related Media feature but worse.
Judge the work by what it did for the client's bottom line
The posts show a consistent priority: output is only interesting if it moved money.
Read more
This shapes how he hires, what he considers good work and how he recommends brands evaluate their own channels. He criticises reporting on conversion rate when orders, revenue and new customers were always the numbers that mattered, and describes email and paid social work in terms of what it changed for the client, not what was delivered.
A hiring post for an email marketer at his agency asks for someone who has run Klaviyo for real brands with large lists sending four or more times a week, can build flows beyond the basic four or five, A/B tests continually, and can show something they built that made money, noting that analysing what campaigns did for the client's bottom line is the part a lot of people skip. An older client story describes stepping in on camera to help a client win a competition.
This is the clearest single answer to what Gil David understands about Shopify measurement.
Read more
It explains what Shopify actually changed, which metrics moved, and how to avoid misreading a conversion-rate jump as a win.
Shopify began filtering bot sessions out of Analytics by default on 21 September.
Gil David works through what that does and does not touch: orders and revenue are untouched, while sessions, conversion rate, add-to-cart rate and bounce rate are affected.
A client store showed Singapore at 5,200 sessions, up from one, all direct traffic and zero orders, for a market the store does not ship to; Shopify flagged roughly 5,000 more Singapore sessions that month as bots, so the filter caught about half.
Ashburn, Virginia appeared across multiple accounts because of its data centres, though real people live there too.
The practical consequences are spelled out: treat 21 September as a new starting line, compare the last ten days of September with the preceding ten to estimate how much bot traffic was inflating the numbers, and note that the mid-month start means September reports stitch two counting methods together.
The recommendation is to judge Meta and Google performance on orders, revenue and new customers rather than Shopify conversion rate.
For brands planning a campaign
Any campaign decision that used Shopify conversion rate as the success metric needs re-baselining. Ask for a before/after split at 21 September, switch scaling judgements to orders, revenue and new customers, and treat late-September conversion-rate jumps as a measurement artefact until proven otherwise.
It answers two questions at once: what Meta changed in reporting, and how a practitioner keeps peak-season comparisons honest when the change lands mid-quarter.
Read more
From 12 November, Meta starts rolling out a pixel feature that measures conversions on the customer's own device, using details they have already supplied, such as an email address at checkout, to claim purchases that previously never came back to Meta.
It is switched on by default for every dataset already sending events, with an opt-out in Events Manager, and it rolls out gradually, so an account may switch over on a day nobody can pinpoint.
Gil David's argument is that reporting is what decisions are made on: CPA and ROAS are what cost caps and ROAS targets were set against, and what this Black Friday will be compared with last year's.
The instructions are concrete: take a baseline now of purchases, ROAS and Event Match Quality; mark 12 November in reporting and stop treating before and after as like for like; reconcile Meta's purchase count against Shopify because Meta has not published how on-device conversions are deduplicated against pixel and CAPI events; check cookie consent for UK and EU stores, since the feature stores and reads information on the customer's device.
The post weighs leaving it on during peak against switching it off until after Cyber Monday for comparable numbers, and repeats that blended MER and incremental revenue are the numbers that matter.
For brands planning a campaign
Build a pre-change baseline and a dated change log before peak trading starts. If the goal is a clean year-on-year comparison, decide in advance whether the feature runs during the peak window, and reconcile platform-reported purchases against the store's own order data.
It shows how Gil David handles attribution claims: what the numbers are, what they mean in plain language, and where the evidence is thin.
Read more
A newcomer gets the difference between a setting that is directionally right and one that is not.
Haus ran geo holdout tests on Meta campaigns and compared how closely each attribution setting matched measured lift.
Rank correlation scores: Incremental Attribution 0.79, 7-day click 0.73, 7-day click plus 1-day view 0.68, 1-day view 0.40, where 1.0 is a perfect match.
Two findings stand out.
Adding 1-day view to 7-day click makes the score worse, so the default combination many brands report on is less reliable than plain 7-day click.
The gap between Incremental Attribution and 7-day click is small, but Incremental Attribution still comes out ahead.
Gil David explains rank correlation directly: it measures whether campaigns are put in the right order from best to worst, not whether a ROAS figure is accurate to the penny, which is why it matters for scaling or cutting.
The post also pushes back on the view that view-through conversions are entirely fabricated, since 0.40 is not zero, while objecting to stacking view conversions on top of clicks.
Caveats are stated: Haus calls it an early read, it covers new-customer conversions only, and the IA and 7-day click ranges overlap.
For brands planning a campaign
Choose the reporting setting deliberately instead of inheriting the default. Use an incremental-style setting for scaling and cutting decisions, and if a brand reports on 7-day click and 1-day view, check whether that combination is distorting the ranking of campaigns before acting on it.
This is the most concrete look at what the day-to-day work involves: a named list of account settings reviewed for clients, with the reasoning for each.
Read more
Readers who want to know whether this person operates accounts or only writes about them should start here.
Meta switched product browsing on for existing ads unless advertisers opted out, meaning a tap on an ad could send someone to a Facebook or Instagram product page instead of the advertiser's site.
Meta said businesses saw 3% more conversions on average, without specifying that these were purchases.
The one-click opt-out had a short window, after which it had to be done ad by ad in creative settings.
Gil David then lists the default settings their team switches off: Advantage+ creative enhancements, including copy rewrites, music and generated backgrounds, checked in advanced preview as well because some only switch off there; account-level creative feature testing, so enhancements turned off elsewhere do not creep back, including on duplicated ads; performance recommendations on auto-apply, so Meta cannot pause or combine ad sets unattended in peak week.
Placement exclusions are flagged as a setting that cannot be switched off, with up to 5% of budget spendable on excluded placements.
Automatic events is judged on its merits.
The stated principle is to switch everything off, then re-enable one thing at a time and test whether it is incremental.
For brands planning a campaign
Treat this as a pre-peak account hygiene checklist. Review auto-applied creative enhancements, duplicated ads and auto-apply recommendations before budget scales, and account for the placement-exclusion allowance when placement reports look inconsistent.
It demonstrates a habit worth learning: reading a platform announcement for its second-order effect on the advertisers who are not buying it.
Read more
Ads Super Burst lets large brands reserve a window of three to 24 hours on a single day in advance, at a price agreed before other bids are placed.
Meta promises delivery within 5% of forecast reach, frequency is fixed at one, US advertisers can buy category exclusivity on certain major dates, and Meta's own figure for early bookings is around 34 million people reached per booking.
Gil David frames it as a television upfront compressed into a day and asks where that inventory comes from and what happens to auction CPMs on the same dates, using Black Friday as the example: if several large brands have locked up tens of millions of people before the day starts, every other advertiser is bidding for what remains.
First Impression Ads, still in alpha, are noted as removing the first ad seen when Facebook is opened from the auction as well.
The post is explicit that nobody outside Meta knows how much inventory these deals carve out.
For brands planning a campaign
On dates known to carry reserved bookings, compare CPMs against the same days last year before assuming a change came from creative or targeting. Treat platform inventory changes as a cost input in peak-season budget planning.
A short, direct example of the diagnostic reflex that runs through the longer posts: look outside the ad account before concluding the ads stopped working.
Read more
US CPMs were up more than 25%, and the post rejects the two obvious explanations, an early peak-season tax or ads suddenly failing and needing a flood of new creative.
The cause given is midterm elections the following week, with a New York Times figure of $3.9bn being spent on election advertising, not all of it on Meta and Google.
The closing instruction is the point of the piece: always look outside your ad accounts for context.
For brands planning a campaign
Before cutting or rebuilding campaigns in response to a CPM spike, check whether an external demand event explains it. Build an external-events calendar into performance reviews so auction-cost changes are not misread as creative failure.
It answers what happens when a platform's AI edits a brand's ads without the brand asking, and it draws on a first-hand client example rather than a news summary.
Read more
Three failures are described in one week.
The National Trust apologised for AI-generated TikTok creatives that ran from its account without its knowledge, and paused advertising on the platform.
Gil David reports the same thing happening to a client: TikTok took existing ads and added ten variants with varying amounts of AI, including irrelevant stickers and promotions that had ended months earlier, with no notification or warning, described as being like Related Media on Meta but worse.
Just Eat was criticised for automated images added to menu items that lacked photos.
On top of that, Meta expanded its ad terms so advertisers are held responsible for anything their AI does, whether or not they knew about or approved it.
The closing line is that graphic designers, creatives and marketers are safe for a while longer.
For brands planning a campaign
When a platform generates or edits ad creative automatically, brand review and approval need to be a documented step, because platform terms may place responsibility on the advertiser regardless. Check platform-generated variants against live offers and ended promotions before a campaign runs.
Platform changes that move reported numbers, not revenue2
AI-generated creative and ad operations2
Agency work with DTC Shopify brands2
Which attribution setting can be trusted1
Meta ad defaults and auto-applied enhancements1
Platform changes that move reported numbers, not revenue
Gil David tracks platform updates whose main effect is on reported metrics rather than on real demand.
Read more
He explains that Meta's on-device conversion measurement, rolling out from 12 November and switched on by default for every dataset already sending events (opt-out in Events Manager), can claim purchases that never previously reached Meta, and that Shopify began filtering bot sessions out of Analytics by default on 21 September — changing sessions, conversion rate, add-to-cart rate and bounce rate, but leaving orders and revenue untouched. His argument is that reporting is what brands scale, set cost caps and ROAS targets against, so a jump in reported performance is often a counting change rather than a new buyer.
Take a baseline of purchases, ROAS and event match quality before the change, mark the switch date in reporting, and stop treating the periods either side as like for like; cross-check Meta's purchase count against Shopify for double counting.
He allows that leaving Meta's on-device measurement on may give the platform better data to work with, and treats the Shopify bot filter as a rare platform change that makes data cleaner rather than murkier.
For brands planning a campaign
A brand that wants decision-grade numbers should expect Gil David to advise a pre-change baseline and a dated read after the switch, rather than acting on a headline ROAS or CVR movement.
Gil David credits Haus geo holdout testing on Meta campaigns for rank-correlation figures: incremental attribution 0.79, 7-day click 0.73, 7-day click plus 1-day view 0.68, and 1-day view 0.40, where 1.0 is a perfect match to measured lift.
Read more
He reads the gap between incremental attribution and plain 7-day click as small, the addition of 1-day view as making the common default worse, and 1-day view as containing some real incrementality rather than none — while objecting to stacking every view conversion on top of clicks. He notes the read is early, covers new-customer conversions only, and that the ranges overlap.
For brands planning a campaign
A brand reporting and optimising on the default 7-day click plus 1-day view should expect Gil David to question the basis on which campaigns are scaled or cut.
Gil David treats Meta's default settings as an ongoing opt-out job.
Read more
He covers 'product browsing' being switched on for existing ads unless advertisers opt out, Advantage+ creative enhancements such as copy rewrites, music and generated backgrounds, account-level creative feature testing, performance recommendations on auto-apply, placement exclusions that can still take up to 5% of budget, and automatic events. His stated principle is to switch everything off and turn features back on one at a time to test whether they add anything incremental, checking duplicated ads because enhancements can creep back.
An opt-out checklist run before peak trading, including advanced previews and post-duplication checks, because some settings only appear in those places.
He would leave automatic events on where a CAPI setup already sends the needed data and no double counting appears, since better-quality data for Meta is rarely a bad thing.
For brands planning a campaign
A brand that wants control over what is tested should expect Gil David to enforce explicit opt-outs and to verify platform-recommended defaults rather than accept them.
Gil David explains Meta's Ads Super Burst — reserved windows of 3 to 24 hours at a price agreed before bidding, frequency capped at one impression per person, and category exclusivity on some dates in the US — as an upfront that takes inventory out of the auction, then raises the knock-on question of what happens to auction CPMs on booked-out days such as Black Friday.
Read more
He makes the same outside-the-account point about US CPMs rising more than 25% in the run-up to midterm elections, citing $3.9bn in ad spend from The New York Times, and repeatedly flags platform changes landing in the fortnight before Black Friday.
Compare client CPMs on dates known to be booked out with the same dates a year earlier, and look for spend events outside the ad account before reacting to a cost increase.
For brands planning a campaign
A brand heading into peak should expect Gil David to separate genuine performance change from seasonal and auction-driven cost movement, and to plan around fixed-date inventory deals.
Gil David's position on AI in advertising is operational rather than promotional.
Read more
He describes TikTok automatically adding ten AI variants to a client's ads, including irrelevant stickers and promotions that had ended months earlier, and notes the National Trust paused advertising after AI creatives ran from its account; he also covers Just Eat's automated menu images and Meta extending its ad terms to make advertisers responsible for what their AI produces, whether or not it was approved. Separately, he argues AI is not saving him time: quicker individual tasks are filled with more work, and reviewing and filtering output is more exhausting and mundane than doing some tasks himself.
For brands planning a campaign
A brand considering AI-generated creative or platform auto-enhancements should expect Gil David to focus on approval and review controls, and on the human cost of filtering output, rather than on speed claims.
The posts show Gil David running client work at High Voltage Digital: Klaviyo email and SMS, mostly Attentive, for DTC Shopify brands across Europe, the US and Canada, with demand growing in email.
Read more
A hiring post for an email marketer sets out the standard he applies — flows and campaigns beyond the basic four or five, sending four or more times a week to large lists, A/B testing, and showing work that made money — and an older client story describes stepping in on camera to help a client win a competition.
For brands planning a campaign
A brand evaluating Gil David should treat him as an agency operator whose public commentary comes out of client retainers, and who measures work by what it did for the client's bottom line.
Gil David writes short, frequent commentary on how advertising and ecommerce platforms change what they report, and what that means for the brands running on them.
Read more
The fastest way into the work is to pick the platform you use: the posts on Meta's on-device conversion measurement and Shopify's bot-session filter explain the switch, the date it took effect, and the reporting habit that keeps the numbers comparable.
From there, the attribution post is the most technical, and the posts on Q4 auction pressure and AI-generated creative are the easiest entry points if you care about cost or brand control rather than measurement.
Read them as one argument in different settings: check what changed before you react to a number, and judge the work by what it earned a client rather than what a dashboard shows.
Readers who want more than platform updates should look at the comparison creators below, who cover ecommerce growth strategy, creative strategy and email marketing respectively.
Sandham writes for the same audience of Shopify and ecommerce operators, and shares Gil David's habit of arguing from a specific client situation rather than from general advice.
Read more
Both treat ecommerce growth as something you work at systematically: Gil David through reporting hygiene and platform settings, Sandham through compounding small gains in traffic, conversion rate, average order value and repeat purchase, which he sets out as a slide from his own sales process. Both also post about the agency world around the work, including time spent with Shopify and with other agencies.
Sandham's thinking sits on the growth-model side, showing how modest improvements across the funnel compound, while Gil David's sits further down in the ad account and analytics layer, dealing with what a platform counts, which attribution setting to trust and which defaults to switch off. If you want a framework for where revenue comes from, read Sandham; if you want to know whether the number you are looking at is real, read Gil David.
Raspé covers the same DTC ecommerce ground from event takeaways, and reaches some of the same conclusions about creative.
Read more
He describes a creative strategy sequence of persona, message, angle, format and execution, talks about testing volume and diversity, and says he will put more energy into low-cost volume through AI creation.
That overlaps with Gil David's interest in who controls what gets tested and in what AI is actually adding.
Both write for brands trying to make better campaign choices rather than for a general audience.
Raspé concentrates on creative strategy and conversion rate optimisation, and reports what other people said at events.
Gil David concentrates on measurement, attribution and platform policy, and reports what a change did inside client accounts.
Raspé is a better follow for how to make and test creative; Gil David is a better follow for whether the results of that testing can be believed.
Rono works on ecommerce email marketing and sets out the same building blocks Gil David's agency hires for: abandoned cart and welcome flows, segmentation, and turning one-time buyers into repeat customers, with Klaviyo pulling product details from Shopify.
Read more
Both connect email and paid channels back to revenue rather than to opens or clicks, and both write specifically for ecommerce store owners.
Rono's posts are practical how-to and service promotion, walking through individual flows and inviting brands to get in touch.
Gil David's email material appears mainly as a hiring standard and as evidence of how his agency judges channel work, while most of his writing analyses platform changes that affect paid social and analytics.
Read Rono for flow-by-flow email guidance; read Gil David for what happens to the numbers once the flow is running.
For a DTC brand evaluating Gil David for a Shopify analytics accuracy brief, the supplied evidence is enough to check content fit with measurement-minded ecommerce operators, not to confirm the audience’s job titles, seniority, or location.
Read more
The only measured audience figure supplied is 9,384 LinkedIn followers observed 11 October 2026.
The ten LinkedIn posts reviewed are written for people running or buying paid social and ecommerce measurement: agency staff, DTC operators, and marketers dealing with Meta and Shopify reporting.
The strongest fit signal for this brief is the post on Shopify bot filtering, which discusses the 21 September default bot-session filter, inflated sessions, depressed conversion rate, and why orders and revenue are better checks; a hiring post for an email marketer shows the agency and client-service context.
That is content appeal, not measured audience composition.
No platform demographics export or audience survey is supplied, so geography, age, gender, employer, occupation, and job seniority cannot be assessed.
Next step: ask Gil David for recent post-level engagement or audience survey data for Shopify analytics and measurement topics, ideally split by DTC ecommerce decision-makers and by country, before treating this as a role-fit audience match.
Gil David's posts do not describe any sponsorship, brand-partnership or creator-collaboration formats.
Read more
The commercial material that does appear is agency-side: High Voltage Digital, the performance marketing agency Gil David represents in these posts, advertises open service capacity for direct-to-consumer brands and recruits staff for its own client work rather than offering sponsored placements.
Several posts close by inviting DTC brands to get in touch about agency services ahead of peak trading, and one asks for one more ambitious DTC brand heading into 2027.
That is a pitch for client work, not a documented partnership package, and the two should not be treated as interchangeable.
No rate card, deliverable list, exclusivity terms, disclosure terms, content-approval process or booking lead time appears in the reviewed material.
This records what the reviewed posts do not document; it is not a statement that Gil David never accepts sponsorships or brand partnerships.
A buyer wanting a creator-side deal has nothing verified here to plan against and would need to confirm directly whether any such offer exists, on what terms, and through which channel.
Brief: a DTC Shopify brand wants analytics it can trust — it wants to know why conversion rate moved, which numbers to scale on, and how much of a recent jump was a counting change rather than a real result.
Read more
This is the closest match in the reviewed material.
Gil David's post on the Shopify bot-session filter does the work the brief describes: it names the date (21 September), separates the metrics the filter touches — sessions, conversion rate, add-to-cart rate and bounce rate — from orders and revenue, which it leaves alone, and gives a re-baselining method.
Treat 21 September as a new starting line, compare the last ten days of September with the ten before to estimate how much bot traffic was inflating the numbers, and remember that the mid-month start stitches two counting methods into one September report.
He shows a client store where Singapore showed 5,200 sessions, up from one, all direct traffic and zero orders, for a market the store does not ship to, and notes Shopify flagged roughly 5,000 more Singapore sessions that month as bots.
The practical conclusion is stated plainly: judge Meta and Google on orders, revenue and new customers rather than Shopify conversion rate.
The supporting material points the same way.
The post on Haus geo-holdout testing covers which Meta attribution setting puts campaigns in the right order, with rank correlations of 0.79 for incremental attribution, 0.73 for 7-day click, 0.68 for 7-day click plus 1-day view and 0.40 for 1-day view, and states the caveats — an early read, new-customer conversions only, overlapping ranges.
The post on Meta's on-device conversion measurement covers a change rolling out from 12 November that can lift reported purchases without new revenue, and sets out a baseline-and-reconcile process.
The account-settings post is a checklist of defaults to switch off or verify before peak, including automatic events.
The fit is not one clever answer but a consistent method: establish a baseline, mark change dates in reporting, cross-check platform numbers against the store's own data, and separate cost movement caused by outside events from performance.
One limit applies.
This is commentary drawn from published posts, not evidence that an audit of this kind has been run for a client, and no client-side documentation was supplied.
When to choose a specialist
Brief: a brand wants a paid creator sponsorship — sponsored posts or product placement with a rate card, a deliverable list, exclusivity and disclosure terms, and a booking lead time — or a broad consumer awareness campaign justified by audience demographics.
Read more
Nothing in the reviewed material supports either.
Across all ten posts there is no sponsorship policy, media kit, rate card or partnership inquiry route.
The commercial material that does appear is agency-side: pitches for High Voltage Digital's client capacity before peak trading, and a hiring post for an email marketer.
A brand shopping for creator-side inventory has nothing here to plan or budget against.
That is a statement about documented evidence, not proof that no such offer exists; the sensible next step is to ask directly whether one does, on what terms, and through which channel.
The audience side is weaker still for this brief.
The only measured figure supplied is a LinkedIn follower count of 9,384 observed 11 October 2026.
There is no demographic, geographic or engagement breakdown, so a plan that assumes a particular consumer audience, age band or market cannot be evidenced from this packet.
The posts are written for marketers and operators — agency staff, DTC teams, people working in ad accounts — and the subject matter is platform mechanics and measurement, not consumer lifestyle content.
A brief for entertainment-led creative production, or for reach-based awareness buying on the strength of audience data, is a poor match on the material available.
A worked campaign example
Worked angle for the Shopify analytics brief: a pre-peak measurement reset, delivered as a one-page plan the brand can run itself.
1. Re-baseline the store. Treat 21 September as the new starting line, compare the last ten days of September with the ten before it to estimate how much of any conversion-rate jump was bot traffic dropping out of the maths, and note in the report that September carries two counting methods. Record bot-heavy locations such as the Singapore and Ashburn patterns so the same traffic is recognised next month.
2. Change the decision metrics. Move scaling and reporting onto orders, revenue and new customers. Keep conversion rate as a diagnostic, not a target, so a landing-page change is not credited for a measurement change.
3. Audit the attribution setting. If reporting runs on 7-day click plus 1-day view, flag that this combination ranked worse than plain 7-day click against measured lift, and that incremental attribution ranked highest of the four tested. Frame the change around the decision it affects — scaling or cutting campaigns — since the test measures ordering, not the accuracy of a ROAS figure to the penny. Carry the caveats with the recommendation.
4. Build a change log with named dates. From 12 November, Meta's on-device conversion measurement rolls out by default for datasets already sending events, with an opt-out in Events Manager. Take a baseline now of purchases, ROAS and event match quality; decide in advance whether the feature runs through peak trading or waits until December for comparable numbers; reconcile Meta's purchase count against Shopify once it is live, since no deduplication method has been published; and for UK or EU stores, check whether cookie consent covers it. Blended MER and incremental revenue are the numbers that survive this kind of change.
5. Add an outside-the-account calendar. Mark dates carrying reserved inventory deals and other demand events such as elections, so a CPM rise is not misread as failing creative. Compare CPMs on booked-out dates with the same dates a year earlier.
Deliverable: a short document naming each metric, its cut-off date, who reconciles it, and the single question the report should answer for the next buying decision. The angle is a measurement reset before peak, not a new campaign.
A better choice when the brand already trusts its numbers and the open question is where growth comes from.
Read more
The supplied material shows a compounding model for ecommerce: a bit more traffic, a slightly higher conversion rate, a stronger average order value, and one extra repeat purchase, presented as a slide used in sales conversations.
It also shows the working-network side of the same world, with a week spent alongside the Shopify team in London, at eCommerce Expo and with agency partners.
If the brief is growth planning or introductions to the Shopify-side ecosystem rather than data integrity, this is the more useful starting point.
The tradeoff is that neither post addresses measurement.
There is no material on bot-session filtering, attribution settings, or reconciling platform numbers against store numbers, and the conversion-rate lever is treated as a thing to improve rather than a number to check.
For the stated brief, which is to understand why conversion rate moved and which figure to scale on, there is nothing here a brand could apply, and a team taking the growth-math angle without first fixing the baseline risks optimising against a number that changed for counting reasons.
A better choice when the brief moves from whether the data is right to what should change on the site and in creative.
Read more
The supplied posts set out priorities from two events: UX and conversion-rate work still being underoptimised from product page to checkout, creative strategy after Andromeda and GEM framed as persona, message, angle, format, execution, offer and subscription design as a unit-economics lever, and early-stage AI and GEO opportunity.
A second post reports an event on DTC creative strategy and a shift toward testing creative at higher volume, including low-cost AI-created volume.
If the task is deciding what to test next rather than diagnosing a metrics change, this material is closer to the decision.
The advice is directional, not an audit method.
It says to focus on website experience before optimising anything else, but it does not show how to separate a real conversion-rate improvement from a measurement change, and it gives no dates, no reconciliation steps, and no checks for platform reporting shifts.
A brand whose conversion rate jumped because bot sessions were filtered out could read this material and credit its new landing page instead, which is the opposite of the brief.
A better choice when the task is narrower and executional: building the cart and checkout abandonment flow.
Read more
The supplied material goes into specifics a brand can act on, with Klaviyo pulling cart item name, price, image and quantity from Shopify, offers written to point at one action, and a template reused rather than over-designed.
A second post pitches email marketing services around strategic flows, segmentation, and turning one-time buyers into repeat customers.
For a brand whose analytics problem is really a lifecycle-marketing gap, this is the more direct route.
It is channel execution, not measurement.
Nothing in the two posts covers data integrity, reporting changes, or attribution, so a brand that engaged on this basis expecting more trustworthy analytics would get email flow work instead.
If flows are underperforming because the underlying conversion or revenue numbers are unreliable, this material does not diagnose that, and the posts are a service pitch and a how-to rather than evidence of a measurement review.
Based on ten LinkedIn posts by Gil David, read in full, covering Meta advertising and measurement, Shopify analytics, AI-generated creative, auction costs and agency work.
The page summarizes recurring ideas and examples in those posts and compares them with selected ecommerce creators.
No video, podcast, newsletter or long-form work was included; audience demographics and engagement data were not available.
Recent engagement
Across their last 9 LinkedIn posts, the median post got 12 reactions and 11 comments. Checked 2026-11-02.
