Fabian Gmeindl works on conversion optimization for online shops at DRIP. Gmeindl's work focuses on using A/B test results to help teams choose what to test, including through Apex by DRIP, a testing tool Gmeindl describes in posts.
Fabian Gmeindl is most useful for B2B SaaS advice on test prioritization and conversion optimization.
Best for
A B2B SaaS company building experimentation or conversion analytics features.
Not for
Consumer brands buying sponsored Instagram posts for reach.
Key takeaways
Apex's Likelihood to Win ranks ideas; it is not a literal probability.
BLACKROLL delivery-time message results differed by placement above or beneath call to action.
KICKZ mobile slider indicators change lifted conversion rate by 2.28%.
Check first: Ask for audience data and sponsorship terms before assessing fit.
Main ideas
Use what other shops have tested to decide what to build next · A prediction orders your backlog; it does not replace the experiment · Where an element appears can matter more than whether it appears
Follow their work for practical test ideas, explanations of what the results show, and reminders to validate promising changes with your own customers.
Fabian Gmeindl connects the full testing process: generating hypotheses, using results from earlier experiments to prioritize them, and checking results in the shop itself.
They describe Apex’s “Likelihood to Win” as a score for ranking ideas based on supporting results, not a literal probability that a test will succeed.
Their case studies make the distinction tangible.
For BLACKROLL, they report different results for delivery-time information above the call to action and beneath it, alongside other reassurance points.
The conversion movement was directional; ask for direct measures of trust or hesitation.
Fabian Gmeindl argues that teams should choose tests using results from experiments other shops have already run, rather than committing design, development and traffic to whichever idea sounds most convincing in a meeting.
Read more
They make the case with Apex by DRIP, a testing tool they say draws on a database of 4.3 million tests from 151,000 shops collected since 2011 and scores and ranks test ideas before teams build them. Gmeindl also marks the limits: a score orders a backlog but does not replace running the experiment in your own shop, and placement and context can matter as much as whether a change exists at all.
Use what other shops have tested to decide what to build next
Gmeindl argues that the expensive part of experimentation is not the test itself but the research, design, build and checking that come before it—and that teams routinely spend that effort on ideas that only sound persuasive.
Read more
Apex by DRIP ranks candidate tests against results from comparable tests at other shops before that work starts.
Gmeindl describes the tool as drawing on a database of 4.3 million tests from 151,000 shops collected since 2011.
Each test idea receives a Likelihood to Win score with supporting results attached, helping teams order their backlog.
The goal is narrow: identify what to test next and what supports that choice.
In the launch post, Fabian Gmeindl frames the problem as a recurring puzzle: a change works for one shop and fails for another, and different reviews, offers or product-page layouts can each sound convincing in a meeting before anyone knows whether they work. The tool brings results from previous experiments into that decision, rather than relying on the strength of the argument.
A prediction orders your backlog; it does not replace the experiment
Gmeindl spells out the limits of the Likelihood to Win score: it is not a literal probability that a particular test will win, but a way to rank ideas within one shop.
Read more
The experiment still has to run, with real customers deciding the outcome.
They also distinguish evaluating an idea before it is built from changing how a live test runs after launch.
A whitepaper promoted by Gmeindl explains what a prediction cannot tell you and how to reject weak hypotheses before development.
They argue that new ideas without a track record should still have room in the backlog rather than being filtered out by the model.
Both published breakdowns distinguish between directional signals and statistically significant results.
In the KICKZ post, the conversion-rate movement is described as directional rather than statistically significant, while the revenue result is reported separately as statistically significant.
The BLACKROLL post makes the same distinction.
Explanations for why results occurred are labeled as hypotheses, not measured causes.
Where an element appears can matter more than whether it appears
Gmeindl makes a concrete case for testing placement.
Read more
Rather than asking whether reassurance such as delivery-time information belongs on a product page, they test where it sits relative to the call to action and other reassurance points.
Their takeaway: test where reassurance appears, not just whether it appears.
They also limit their interpretation to what the test measured: the comparison was between placements and did not directly measure trust or hesitation.
At BLACKROLL, delivery-time messaging above the call to action produced a small drop in ARPU.
The identical messaging below it, beside other reassurance points, increased ARPU and conversion rate on both mobile and desktop.
Small mobile details are a real revenue lever on high-traffic shops
Gmeindl argues that small mobile navigation details are worth testing because the changes are cheap and their measured effects can justify the work at a high-traffic shop.
Read more
They link the reasoning to cognitive ease, suggesting that easier image navigation lowers cognitive load so shoppers process visuals faster and decide sooner.
The test measured purchasing outcomes; the explanation is a hypothesis.
Gmeindl also reports device-level outcomes separately, a recurring practice rather than a one-off.
At KICKZ, clearer image-slider preview indicators on mobile product detail pages for men's sneakers were followed by a 2.28% conversion-rate lift and ARPU rising from EUR 3.89 to EUR 3.99. Gmeindl's closing line is to test the small stuff because it pays off big for high-traffic sites like sneaker shops.
Idea generation and prioritization are two separate steps
Gmeindl turns testing knowledge into a productized asset, then draws a line between generating and prioritizing ideas.
Read more
They describe a library of more than 45 A/B test ideas in Notion, each with a hypothesis, the psychological mechanism behind it, key takeaways and available win-rate predictions.
The ideas span product pages, navigation and checkout.
Finding an idea is the first step; deciding which deserves the team's time comes next.
The library feeds hypotheses, while the prediction tool helps prioritize them.
The listed categories cover social proof and expert endorsements, cross-sell, bundle and cart upsell, mobile search, size filters and quick add-to-cart, and price presentation, savings and guest checkout. Examples are attributed to Birkenstock, Puma, yFood and Caraway.
Start here for the question that drives Gmeindl's work, rather than a single test result.
Read more
The post explains what Apex by DRIP is and the problem it was built to solve.
Gmeindl opens with a recurring puzzle in conversion testing: a change that wins for one shop fails for another.
Reviews, offers or a new product-page layout can all sound convincing in a meeting, before anyone has researched, designed, built and checked them.
The answer is Apex by DRIP, built on a database of 4.3 million tests from 151,000 shops collected since 2011.
Each test idea gets a "Likelihood to Win" score with supporting evidence, used to rank ideas within a shop.
They make clear that the score is not a literal probability that a test will win: the experiment still has to run, and customers still decide.
The DRIP team can also build, launch and evaluate tests for teams that want execution help.
The goal is narrow and concrete: answer "What should we test next, and what supports that choice?"
For brands planning a campaign
The prioritization framework is the part a platform team could reproduce: rank candidate tests before development and attach a rationale to each score. The clear explanation of what the score means—and does not mean—is a useful standard for any analytics feature built on the same idea.
A compact example of Gmeindl's full loop: a pre-test prediction, the change itself and the measured outcome in one post.
Read more
The test changed image-slider preview indicators on mobile product detail pages for men's sneakers at KICKZ, making them clearer for swiping through product images and reducing clutter so key information appeared sooner.
Apex predicted a 67% likelihood to win, noting that 12% of the 25 closest market tests kept this kind of change, slightly above the typical 8%.
The reported result was a 2.28% conversion-rate lift, ARPU from EUR 3.89 to EUR 3.99 (+2.57%), and AOV up 0.08%, with the revenue result reported as statistically significant.
They distinguish the conversion movement from that claim, calling it a directional signal, and offer cognitive ease—less fumbling with previews on touchscreens—as a hypothesis for why it worked, not a measured cause.
For brands planning a campaign
A useful format precedent for presenting a prediction alongside an outcome. The post also distinguishes a statistically significant revenue figure from a directional conversion move instead of blending them.
A small, testable variable—and an example of why placement can matter more than whether a feature is present.
Read more
At BLACKROLL, the test compared delivery-time messaging above the call to action with the same messaging underneath it, beside other reassurance points.
Apex predicted a 45% likelihood to win, with 8% of the 25 closest market tests kept, around the typical rate for this kind of change.
Above the CTA, ARPU fell 0.22%.
Below the CTA, ARPU rose 1.95%, conversion rate 1.87%, and AOV 0.03%, with mobile ARPU up 2.24% and desktop ARPU up 2.37%.
They recommend testing where reassurance appears, not just whether it appears.
The test compared placements; it did not directly measure trust or hesitation.
For brands planning a campaign
This placement-level distinction is easy to lose in a feature that treats delivery messaging as an on/off setting. It makes a concrete case for keeping position as a variable in test planning and supporting analytics.
A look at how Gmeindl packages testing knowledge for practitioners and where they draw the line between idea generation and prioritization.
Read more
Gmeindl describes a Notion library of more than 45 A/B test ideas, each with a hypothesis, the psychological mechanism behind it, key takeaways and available win-rate predictions.
The ideas span product pages, navigation and checkout, covering social proof and expert endorsements, cross-sell, bundle and cart upsell, mobile search, size filters and quick add-to-cart, and price presentation, savings and guest checkout, with examples drawn from Birkenstock, Puma, yFood and Caraway.
The post draws a clear line: finding an idea is the first step; deciding which idea deserves the team's time comes next.
The library feeds hypotheses, and Apex prioritizes them.
For brands planning a campaign
Gmeindl pairs a library of ideas with prioritization. A platform could borrow that format to move users from browsing ideas to committing testing capacity.
The clearest account of what prediction is meant to do—and where Gmeindl says it stops.
Read more
Read it before judging the approach.
This post introduces a whitepaper on prediction-based experimentation and outlines its five parts: what "predictive" means, distinguishing evaluation of an idea before it is built from changes to how a test runs after launch; how results from comparable shops inform the next move and why context matters; what a prediction cannot tell you, including that a promising idea still needs testing in your own shop and that new ideas without a track record deserve room; which ideas to leave out, so weak hypotheses are rejected before development; and how Apex puts the approach into practice through test memory, a testing tool and managed execution, with predictions checked against real results. The post cites 4.3 million tests from 151,000 shops since 2011 and says a prediction helps prioritize, while the experiment still has to prove the result.
For brands planning a campaign
The limits section names a risk for cautious buyers: a prediction tool could narrow exploration.
Gmeindl's answer is to leave room for untested ideas.
Buyers should ask how a tool balances predictions against new hypotheses.
Stated limits of prediction: what a score can and cannot tell you4
Gated resources available by direct request3
Prediction-based experimentation: using past tests to decide what to build2
Case-study breakdowns built around one test and its numbers2
Small mobile detail changes as a revenue lever2
Placement and context, not just presence1
A test-idea library with hypotheses, psychology, and predicted win rates1
Prediction-based experimentation: using past tests to decide what to build
Fabian Gmeindl argues that teams should use data from tests other shops have already run to choose what to test, rather than committing design, development, and traffic to whichever idea sounds most convincing in a meeting.
Read more
Apex by DRIP draws on a database of 4.3 million tests from 151,000 shops collected since 2011 to score each test idea and produce a ranked backlog.
The score ranks ideas within a shop; it is not a literal probability that a given test will win.
The experiment still has to run in the shop, where real customers decide.
The DRIP team can also help build, check, launch, and evaluate tests for shops doing at least six figures a month in revenue.
Predictions draw on the closest comparable market tests and include the reasoning behind each score, producing a ranked idea with supporting evidence rather than a bare number.
For brands planning a campaign
For a B2B SaaS company, this is the clearest statement of how Gmeindl thinks about analytics: use historical test data to prioritize experiments, while keeping predictions separate from results. A partnership or advisory conversation can start with where that layer would sit in your product, how scores are explained to users, and how to handle ideas with no comparable history.
Stated limits of prediction: what a score can and cannot tell you
Across the launch post and whitepaper outline, Gmeindl marks the method's limits: a prediction helps you prioritize, but it does not replace running the test in your shop.
Read more
They distinguish evaluating an idea before it is built from changing how a live test runs after launch, and argue that new ideas without a track record deserve room in the backlog. One whitepaper section covers what a prediction cannot tell you; another explains how to reject weak hypotheses before development so resources go to stronger tests.
Gmeindl separates predictions from outcomes in every description. In the case studies, they label directional movements as directional rather than statistically significant.
For brands planning a campaign
If your platform adds predictive or benchmarking features, a credible advisor would push for clear labels explaining what a score means, visible reasoning, and an explicit path for novel ideas the model cannot score well. This checklist can guide feature copy and internal reviews of an analytics roadmap.
Case-study breakdowns built around one test and its numbers
Gmeindl publishes short breakdowns of individual tests run for named shops.
Read more
The KICKZ post covers clearer image-slider preview indicators on mobile product detail pages for men's sneakers, reporting a 2.28% conversion-rate lift, ARPU from EUR 3.89 to EUR 3.99, and a small AOV change; the revenue result was statistically significant.
The BLACKROLL post compares delivery-time messaging above the call to action with messaging underneath it beside other reassurance points.
The placement below the button delivered +1.95% ARPU and +1.87% conversion rate, while the placement above it had negative ARPU.
Each writeup gives the predicted likelihood, reasoning, measured outcome, and a hypothesis for why the result occurred.
Each breakdown states the predicted likelihood and measured business metrics, then labels the explanation of why the numbers moved as a hypothesis.
For brands planning a campaign
The format is directly reusable: one change, one shop, the numbers, and a stated reason.
A SaaS company could ask Gmeindl to adapt it for customer stories or benchmark reports.
The two examples show the level of detail they consider sufficient to make a result trustworthy.
The BLACKROLL test makes Gmeindl's case that where an element appears can matter more than whether it appears.
Read more
Delivery-time information performed differently depending on its position relative to the call to action and other reassurance points.
Their takeaway: test where reassurance appears, not only whether it appears.
The post notes that the test compared placements but did not directly measure trust or hesitation, so the interpretation stays within what the data supports.
The result comes from a single test at one shop. The conversion movement is directional, not statistically significant, and the above-the-button variant moved in the opposite direction. Ask for more evidence before treating placement as a general rule.
For brands planning a campaign
This is the kind of question a conversion-focused advisor brings to a product decision: not "should we add this element" but "where does it sit relative to the decision it is meant to support." For an analytics platform, it shows the value of testing variants of the same element rather than relying on on/off feature flags.
The KICKZ post argues that polishing navigation details on mobile pays off for high-traffic shops.
Read more
Easier image navigation is framed as reducing cognitive load so shoppers process visuals faster and decide sooner; the post cites cognitive ease as the explanation.
The same test reports mobile ARPU up 2.24% and desktop ARPU up 2.37% in the BLACKROLL case, showing how Gmeindl separates device-level outcomes.
Their closing line: test the small stuff, because it pays off big for high-traffic sites.
The reasoning cites a named psychological mechanism, while stating plainly that the test measured purchasing outcomes and the explanation for why they changed is a hypothesis.
For brands planning a campaign
For a platform selling testing to e-commerce teams, this argues for making small, cheap experiments first-class in the product rather than treating them as noise: they are cheap to build, and on high-traffic sites the measured revenue effect can justify the work. It also supports device-split reporting as a default view.
A test-idea library with hypotheses, psychology, and predicted win rates
Gmeindl offers a library of more than 45 A/B test ideas in a Notion document.
Read more
Each idea includes a hypothesis, the psychological mechanism behind it, key takeaways, and available win-rate predictions.
Categories include social proof and expert endorsements; cross-sell, bundle, and cart upsell ideas; mobile search, size filters, and quick add-to-cart; and price presentation, savings, and guest checkout.
Examples are attributed to Birkenstock, Puma, yFood, and Caraway.
The post presents the library as the idea-generation step and the prediction tool as the next step: prioritization.
For brands planning a campaign
This is a productized asset a brand can evaluate directly, paired with a division of labor Gmeindl makes explicit: generate hypotheses broadly, then rank them with evidence. A SaaS company could use the same two-step framing to explain how its analytics features help teams short on test ideas rather than traffic.
Two posts offer the whitepaper and Notion testing library by direct request, not through public download links, with the line "I'll send it over."
Read more
The Apex launch post points to a demo request link and sets a revenue threshold: at least six figures a month for shops that want to see how the tool could work for them.
Ask what happens after a request and whether the six-figure threshold is a standing qualification rule; it appears in only one post.
For brands planning a campaign
The posts use low-friction lead capture through direct messages and qualify the paid product by revenue. A brand planning a collaboration should expect this style of call to action and decide where a downloadable or gated asset would live if it wants to measure the audience.
Fabian Gmeindl's posts centre on one commercial relationship: DRIP, the company behind the Apex testing tool, which they promote and write about in the first person plural.
Read more
The same posts report named test results from individual online shops (KICKZ, BLACKROLL) as CRO breakdowns.
Their testing library also names Birkenstock, Puma, yFood and Caraway as example brands.
Ask what relationship, if any, Gmeindl or DRIP has with these shops and brands.
DRIP
Disclosed partnership
Gmeindl's posts present DRIP as the company behind Apex, the A/B testing tool, and they write on its behalf: "At DRIP, we've spent years running experiments for online shops", "we built our own testing tool, Apex by DRIP", and "Apex tells you which test wins before you build it, and runs your whole testing program in one place."
Read more
They offer a whitepaper and a demo request link for the product and say the DRIP team helps build, check, launch and evaluate tests. The posts use "we" and "our own testing tool" but do not state their title, ownership stake or employment terms; ask for details of their formal relationship with DRIP.
KICKZ appears as the named shop in a mobile CRO breakdown: optimizing image slider preview indicators on men's sneaker product detail pages, reported as +2.28% conversion rate and ARPU of €3.89 to €3.99, with the revenue result described as statistically significant.
Read more
The post says "we optimized" the change for KICKZ; ask whether KICKZ was a client or a record drawn from the test database.
BLACKROLL appears as the named shop in a delivery-messaging placement test, comparing delivery-time information above the call to action with the same message beneath it beside other reassurance points.
Read more
The below-CTA variant is reported at +1.87% conversion rate and +1.95% ARPU. Ask whether BLACKROLL was a client or partner.
Fabian Gmeindl writes about ecommerce experimentation from the perspective of deciding what to test, not just running tests.
Read more
Start with the two single-test breakdowns: clearer image-slider indicators on mobile product pages for a sneaker shop, and moving delivery-time messaging below the call to action for a supplements brand.
Each follows the same pattern: the change, the predicted likelihood, the measured numbers, and an explanation labeled as a hypothesis.
The launch post for the Apex by DRIP tool and the whitepaper outline explain the method behind those numbers: a database of 4.3 million tests from 151,000 shops, used to rank test ideas before anyone builds them.
The score helps teams order a backlog; it is not a probability that a test will win.
The library of 45-plus test ideas covers idea generation, organized by page area—product pages, navigation, checkout—with a hypothesis and psychological mechanism attached to each idea.
Read the case studies for practical detail, and the method posts for the argument about how teams should choose what to work on.
Khan tackles the same practical conversion problem from the design side: what a shopper needs answered before buying, and whether the page provides those answers.
Read more
Their supplement product-page rebuild follows the structure Gmeindl uses in a test breakdown—what changed, why, and what the numbers did afterward—with conversion moving from 1.8% to 5.8% and revenue from $18,000 to $58,000.
The Wuffes navigation post makes a similar argument about placement: organizing a store around customer problems such as itching or joint health can carry more of the sales conversation than a polished product name.
Both distinguish what a result proves from what it suggests; Khan says the Wuffes menu does not prove growth, but offers it as a hypothesis to test.
Khan reasons from design craft and single-page rebuilds, backed by client outcome figures.
Gmeindl's distinctive move is the prediction layer: ranking an idea against comparable tests before development starts.
For tactical page-design judgment, follow Khan; for arguments about how teams should choose what to test, follow Gmeindl.
A conversion optimization practitioner publishing for the same audience of ecommerce and DTC brands, with a similar habit of turning advice into specific, checkable claims.
Read more
Molyneux's supplement CRO post covers ground also found in Gmeindl's library: social proof at every decision point, mobile-first fixes quantified in revenue terms, checkout anxiety, and guided product quizzes.
Molyneux grounds the advice in more than 100 successful tests on supplement brands.
The Black Friday post makes a similar arithmetic argument about test volume: below 100,000 monthly visitors, a traditional A/B test will not reach sample size, so test landing pages on the ad platform instead and cut losing variants fast.
Molyneux's focus is narrower and more service-shaped: supplement brands, an agency offer with a booking link, and category checklists rather than single-test breakdowns with predicted versus measured results.
Gmeindl covers ecommerce across categories and centers on tooling—the prediction score, the idea library, and the database behind them.
Choose Molyneux for supplement products or a checklist to work through; choose Gmeindl to see how one test's numbers were interpreted.
Fabian Gmeindl’s LinkedIn posts cover e-commerce conversion testing, from product-page changes and test prioritization to Apex by DRIP.
Read more
For a conversion-optimization brief, request recent LinkedIn audience data for the buyer’s target roles and markets. For a strategic-advisor brief, ask for documented survey results or another source covering audience occupation, employer type and seniority; platform data alone may not verify those attributes.
The five reviewed LinkedIn posts document no sponsorship packages, brand-partnership formats, rates, exclusivity terms or partnership inquiry route.
Read more
They do document DRIP’s commercial offers: a whitepaper on prediction-based experimentation and a testing library of 45+ A/B test ideas, each with hypotheses, psychological mechanisms and predicted win rates, both available by direct message; and Apex by DRIP, a testing tool the posts describe as telling you which test wins before you build it.
Apex targets shops doing at least six figures a month in revenue, and the posts direct prospects to request a demo, not to contact DRIP about partnerships.
DRIP also offers managed execution: the team builds, checks, launches and evaluates tests for shops that want help running them.
For a partnership plan, ask Gmeindl directly about sponsorship availability, lead times, deliverables, approval steps and pricing.
A B2B SaaS company building experimentation or conversion-analytics features is a strong fit for a strategic-advisor engagement with Fabian Gmeindl.
Read more
All five posts address the problem such a product has to solve: deciding which tests deserve to be built.
They describe a database of 4.3 million tests from 151,000 shops collected since 2011 and a tool, Apex by DRIP, that scores test ideas and ranks a backlog.
More useful for an advisory engagement than the tool itself is how they discuss it.
They state that a score is not a literal probability that a test will win, that the experiment still has to run in the shop with real customers deciding, and that ideas with no comparable history still deserve room.
They distinguish evaluating an idea before it is built from changing a live test after launch, and devote a section of the whitepaper to what a prediction cannot tell you.
That discipline matters when shipping a predictive or benchmarking feature: clear labels on what a number means, visible reasoning behind it, and an answer for cases the model handles badly.
The published case studies are also a reusable content format.
Each names the change, the shop, the predicted likelihood, the measured business metrics, and a hypothesis about why the numbers moved.
The hypothesis is explicitly labeled as such, and directional movements are labeled directional rather than statistically significant.
A SaaS company can borrow that structure for its customer stories and benchmark reports, and ask Gmeindl to pressure-test its analytics roadmap against it.
When to choose a specialist
A consumer brand buying sponsored Instagram posts for reach is a poor fit based on these posts: they are practitioner-facing B2B material on e-commerce testing, not consumer lifestyle or product content, and none shows the kind of paid placement a reach-based sponsorship campaign would need.
Read more
The posts document no sponsorship packages, rates, exclusivity terms, deliverables, approval steps, or inquiry route.
They do document DRIP's commercial routes: a whitepaper and testing library offered by direct message, and Apex, a testing tool aimed at shops doing at least six figures a month in revenue, with a demo request link.
Neither is presented as creator sponsorship.
The posts also provide no audience-demographic information, so a buyer should ask for audience data and sponsorship terms before assessing fit.
The same weak fit applies to campaigns that depend on guaranteed reach or a published media kit; neither appears in the posts.
A worked campaign example
For a B2B SaaS company assessing Gmeindl as a strategic advisor for conversion optimization, one workable angle is a paid review of the platform's predictive scoring feature, followed by a co-produced benchmark piece.
Step one is an audit session in which they review how the product presents predictions, using their stated standards as a checklist.
Does the interface explain what the score means and does not mean?
Is supporting evidence visible next to the score, as their posts attach reasoning to each likelihood figure?
Is there a defined path for a novel idea the model cannot score well, so the feature does not quietly narrow what teams try?
Does the product distinguish evaluating an idea before build from changing a running test?
The output is a written set of findings and specific copy and interface changes.
Step two turns the audit into content.
Gmeindl writes a public breakdown in the format of the KICKZ and BLACKROLL posts: the change, the prediction, the measured outcome, and a hypothesis labeled as a hypothesis.
For a platform feature, that becomes a case study about what happened after a scoring feature was revised.
Step three is a joint benchmark or whitepaper using the platform's anonymized test data, following the structure of the prediction-based experimentation paper, including a section on limits.
Before scoping the work, establish whether an advisory relationship overlaps with a competitor or channel partner, since Apex by DRIP is a testing tool.
Also agree on the commercial shape: what Gmeindl is paid for, what they are accountable for, and whether the joint content is co-authored.
The posts do not describe how they structure advisory or paid work, so ask directly.
The immediate next step is a scoping call, with the audit checklist as the agenda.
A better choice when a platform's users are DTC and Shopify brands that need a working CRO program, especially smaller stores that cannot reach A/B test sample size.
Read more
Molyneux writes about testing landing pages directly on Meta instead of running an underpowered A/B test, shifting buyers to subscriptions without hurting overall conversion, keeping quiz funnels short, and placing trust signals where checkout doubt peaks.
There is also a concrete Black Friday sequence: acquire email subscribers at normal CPMs before Q4, lock the offer before design, build one landing page per offer and ad angle, and kill losing variants fast.
For a SaaS product selling into consumer commerce and seeking advice on what its customers do week to week, this is closer to the work than Gmeindl's platform-side material.
These posts offer consumer brands an agency playbook, not experimentation infrastructure.
They provide no method for scoring or ranking test ideas, benchmark database, or guidance on presenting predictions to users.
A platform building analytics features would gain tactical CRO insight, but not the product-level methodology Gmeindl describes.
A better choice when the engagement needs design-side conversion work rather than experimentation methodology.
Read more
Khan's posts cover page and navigation design: leading with the product benefit, answering purchase barriers such as convenience, taste, and cost up front, adding reasons to believe, and organizing a Shopify menu around customer problems such as itching, joint health, or anxiety rather than product names.
The redesign post reports conversion moving from 1.8% to 5.8% and revenue from $18,000 to $58,000.
This is the closer match for a platform whose customers struggle with page-level conversion, or a design-led redesign project that needs CRO judgment applied to layout and copy.
Khan's evidence centers on hands-on CRO design, not testing data.
The two posts report a single redesign and describe a navigation hypothesis without claiming it caused Wuffes' growth.
A platform looking to sharpen a predictive feature would get design judgment, but not evidence on test prioritization, scoring, or running an experimentation program.
A better choice only when the need is construction and real-estate project advisory, not conversion optimization.
Read more
Conzelman's posts cover a manufacturing facility build in Franklin County, construction risk, design defects, change orders, and building codes. A B2B SaaS company assessing a conversion-optimization advisor would not find CRO or experimentation evidence here.
The subject matter does not overlap. Nothing in the two posts addresses A/B testing, analytics features, or e-commerce conversion, so choosing Conzelman for a need involving A/B testing, analytics features, or e-commerce conversion would trade away all the relevant expertise.
This profile draws on five complete LinkedIn posts by Fabian Gmeindl: a library of more than 45 A/B test ideas with hypotheses and predicted win rates; a whitepaper outline on prediction-based experimentation; the launch post for Apex by DRIP; and two single-test breakdowns, one for KICKZ on mobile product pages and one for BLACKROLL on delivery-message placement.
No publication dates are given for these posts, so nothing here rests on a posting order or on change over time.
Account-level follower counts are from October 2026.
Two other creators, Harry Molyneux and Hayan Khan, provide comparisons; a third creator is not included because the subject matter does not overlap with e-commerce experimentation.
No Instagram posts are discussed here.
Material gaps: no publication dates, no audience-demographic data, no sponsorship packages, rates, or partnership contact details, no description of how paid advisory or consulting work is structured, and no posts beyond those listed, including no video, podcast, or long-form article material.
Where a fit judgment turns on the absence of published sponsorship formats, that reflects what Fabian Gmeindl's posts here show, not a claim that no such offer exists.
