Ronak Shah shares practical lessons on growing direct-to-consumer brands, including ways to use AI for creator sourcing, customer retention, and other growth tasks. His posts draw on experience scaling Obvi past $100 million in revenue.
Most useful for practical advisory on AI-assisted creator sourcing and repeat-purchase retention for DTC brands.
Best for
It fits a DTC brand past $10M seeking creator acquisition and repeat-order analysis.
Not for
This is not for paid creator posts, short video reviews, or reach buys.
Key takeaways
A rising revenue line can hide a customer base that never comes back.
Subscriptions manage only about a fifth of repeat customers.
Creator programs stall at twenty people because sourcing does not scale by scrolling.
Check first: Confirm data access, deliverables, and how success will be measured.
Main ideas
The prompt is the lesson: teaching AI use by handing over the exact instruction · A rising revenue line can hide a customer base that never comes back · Subscriptions manage only about a fifth of repeat customers
Ronak Shah is known for practical lessons on direct-to-consumer growth, drawing on posts about scaling Obvi past $100 million and using AI in everyday business work.
The supplied posts cover creator sourcing, repeat-purchase retention, marketing, and operational checks.
Shah’s advice is useful to follow because it often shows the problem, the numbers behind it, and a prompt or method to investigate it.
The supplied material does not establish a signature book, show, or earlier body of work.
Shah connects AI to specific business decisions rather than presenting it as a general cure-all.
In one example, a repeat-purchase analysis exposed a brand whose rising revenue masked weak customer retention; the proposed next step was to ask lapsed customers why they left before building a loyalty program.
In another, Shah explains how AI could help find suitable small creators and screen large volumes of their videos for risky claims.
The posts also extend beyond AI tips: a story about an unexpected box of candy after a partnership dinner makes Shah’s stated emphasis on listening and follow-through concrete.
Some examples describe work with named commercial partners, so readers can distinguish those posts from independent advice.
Together, the examples give readers practical ways to question growth numbers, identify bottlenecks, and judge whether a proposed fix addresses the actual problem.
Ronak Shah's recurring ideas are about where growth work actually gets stuck.
Read more
The posts argue that a rising revenue line can hide a one-and-done customer base, so the first move is cohorts rather than a loyalty program; that subscriptions manage only about a fifth of repeat customers; that creator programs stall around 20 people because finding hundreds of small accounts cannot be done by scrolling a feed; and that in regulated categories the limit on creator content is compliance review, not production.
Two further priorities run through the material: screen expensive decisions such as landing pages and financial models before the money goes out, and run a partner's product through your own brand before saying anything about it publicly.
Much of the output is a numbered daily series that publishes the exact prompt behind each fix, so a reader can rerun the exercise on their own data.
The prompt is the lesson: teaching AI use by handing over the exact instruction
In a recurring series ("one way I'm using AI every day"), each post takes a single operational problem and publishes the prompt used, so a reader can rerun it on their own inputs.
Read more
The argument built into the format is that the tool is not the interesting part; the named inputs and the required output are. Ronak Shah's posts specify what goes into the model and what should come back — full order history, a day's video transcripts, a product and contact target — instead of describing AI in general terms.
The sourcing post: researching contract manufacturers that fit a product, finding the right person to contact at each, and drafting ten personalised inquiry emails in one sitting, replacing what Shah describes as weeks or a month of stop-start work that keeps getting pushed to next week.
A rising revenue line can hide a customer base that never comes back
The argument is that top-line growth and a leaking customer base look identical until you pull the cohorts, so the first move on a retention problem is a repeat-purchase curve rather than a loyalty program.
Read more
New customers entering at the front door can mask the ones walking out the back, and a revenue dashboard will show that as growth.
A founder told Shah his brand was growing fast, with 0.2% of revenue coming from subscriptions.
The curve built from order history showed more than 80% of customers bought once and never returned, only 1.2% ever reached a fifth order, and average order value had slipped from $40 to $36.
Shah describes the growth as a treadmill, and notes the founder chose to survey lapsed customers before building anything.
Subscriptions manage only about a fifth of repeat customers
Shah argues that even the best subscription programs convert roughly 20% of repeat customers, leaving 80% who reorder on their own with no cadence and no one managing the relationship.
Read more
Treating subscriptions as the whole retention story therefore leaves most repeat revenue unmanaged. The direction Shah points to is predicted rather than scheduled commerce: using data to judge when a specific customer is ready to reorder instead of nudging everyone on the same cadence.
Work with rePete by Bold Commerce produced a free guide that Shah says breaks down how leading DTC brands reach the other 80%. The post highlights the claim that fewer than 20% of cancellations happen because a customer stopped wanting the product, and a case where Built Brands grew reorder revenue four times faster than subscription revenue in 60 days without cannibalising a single existing subscriber.
This is a stated change of position: Shah writes that they used to think subscriptions were the whole retention story, and now argues they are just one channel for one type of customer.
Screen a decision before the money goes out, not after a live test fails
Two posts make the same case for expensive, hard-to-reverse decisions: feed the model your own winners and losers and ask which bucket a new idea resembles before spend goes behind it.
Read more
For financial models, Shah separates safe errors from dangerous ones — a wrong assumption can be defended, but an internal contradiction across tabs tells a sharp investor the founder does not know their own numbers, and it is exactly the kind of error that stops being visible after the hundredth look.
A supplement brand's plain, article-style advertorial page acquired customers at $454, while its polished branded page ran $682 — a gap of $228 per customer on the same product and traffic, with media costs steady throughout.
A listicle page was killed after $5,000 of spend and CAC above $1,000.
In the other post, a check of a founder's pre-raise model found lifetime-value math that did not match the revenue build, a chunk of ad spend missing from the budget, and deal terms on one tab contradicting another.
Creator programs stall at twenty people because sourcing does not scale by scrolling
Shah argues the difficult part of a creator program is finding hundreds of small accounts that genuinely fit the product, which cannot be done by scrolling a feed — and that this is why most programs stall at around 20 people.
Read more
The method is to point AI at a category, pull nano and micro creators who already post about similar products, and score them on fit, real engagement, and whether they drive sales rather than on follower count.
In the case Shah recounts, a nine-figure beauty brand switched off paid advertising for 18 months after iOS 14 took its ROAS from 4x to under 1, spent that time building a base of nano and micro creators, and used that base as its engine when TikTok Shop launched in the US in 2023. The published prompt asks for creators with 1,000 to 50,000 followers who post about a category and returns handle, posting frequency on product content, real engagement rate, and a fit score, to build a seeding list of the 200 most likely to convert.
In regulated categories the bottleneck is clearing creator content, not making it
Shah argues creator programs are treated as a volume game, but every video makes claims, and in a regulated category one unsubstantiated line is a legal problem waiting to happen.
Read more
The practical cap on how much creator content can safely go live is review capacity, so that is where the automation belongs — not in producing more content.
A founder told Shah his program produces 30 to 50 videos a day and his team had no time to read them, so output either slowed to a crawl or risky claims slipped through. The published prompt transcribes each video and asks for every claim that could be a compliance risk in that category — health or medical claims, "clinically proven", "all natural", guarantees, anything unsubstantiated — quoting the exact line, explaining why it is a risk, and suggesting compliant wording.
Try a partner's product on your own brand first, and treat the criticism as the proof
Shah's stated standard for backing a company publicly is to run their own brand through it first, then report the unflattering parts.
Read more
The reasoning: a sales tool would flatter, so an analysis that scores your moat "Basic" and names where you are exposed is the reason to trust it. The same post argues about incentive design — a partner who funds the test and earns only a percentage of sales shares the downside, which Shah calls unusual in more than a decade of DTC.
ZyG, described as out of stealth with $118 million raised and the team that built ironSource behind it, took one of Obvi's hero products through a short intake and returned an AI-built market report: a sub-category sized at $3.6 billion and growing 14.5% a year, rated saturated, with nine direct competitors running 1,216 active Meta ads between just the top four; ad teardowns with customer praise and complaints pulled verbatim from Reddit, TikTok and Instagram; 37 selling points tagged exclusive, distinctive or common; and a moat audit that scored Obvi "Basic."
Shah describes the terms as a free report, a marketability test funded by $10,000 of ZyG's own media spend, and a percentage-of-sales model if ZyG takes the brand on.
The post does not state whether Shah is paid for the relationship.
Partnerships are judged on remembered details, not the pitch
Shah's conclusion from a partner visit is that what actually counts is paying attention to what someone cares about and then showing up — not the deck and not the follow-up email.
Read more
He applies that standard to how a partner treats a relationship, and by extension how he would judge one.
Nikolaus from Whop flew from California to the New Jersey suburbs for a dinner meant to be about a partnership, and the conversation turned to Halloween.
Shah described being the street's king-size candy house, a tradition tied to a family member's former job at Mars, and joked that a new job at Haleon ends it because you cannot hand out Advil to seven-year-olds.
A box of more than 200 Mars and Skittles bars arrived on the front step the next morning, not as part of a pitch.
This is the clearest answer to the question a newcomer is most likely to have: what does Ronak Shah actually do with AI?
Read more
It is the one piece here that hands over a working method rather than a conclusion, and the prompt can be run as written.
Ronak Shah tells the story of a nine-figure beauty brand that switched off paid acquisition after iOS 14 took its ROAS from 4x to under 1, then spent 18 months building a base of nano and micro creators.
When TikTok Shop launched in the US in 2023, that creator base was already in place and became the brand's main engine.
His point is that the difficult part is finding hundreds of small accounts that genuinely fit, which cannot be done by scrolling, and that this is why most creator programs stall at around 20 people.
He gives the prompt he uses: have AI find creators with 1k to 50k followers who post about a category, then return handle, posting frequency, real engagement rate, and a fit score for a seeding list of the 200 most likely to convert.
For brands planning a campaign
A brand with a small team can test this seeding approach on one category before committing budget, and the output is a shortlist that feeds directly into a paid pilot with a defined number of creators.
It shows how Ronak Shah reads a retention problem and what he checks before recommending a fix, which matters if you are being sold a loyalty program.
Read more
A founder whose brand was growing fast had 0.2% of revenue coming from subscriptions.
When AI built the repeat-purchase curve from order history, more than 80% of customers had bought once and never returned, only 1.2% ever reached a fifth order, and average order value had slipped from $40 to $36.
Shah describes the revenue growth as a treadmill: new customers replacing the ones leaving.
He notes the founder chose to survey lapsed customers before building a loyalty program.
The prompt asks AI to build the curve by order number and days between orders, then state whether growth is coming from returning customers or new ones.
For brands planning a campaign
For a retention engagement, the cohort curve is the first diagnostic to run, and it can redirect budget away from a loyalty build toward the reason customers leave.
It states Ronak Shah's current position on retention in one paragraph and shows the commercial relationships behind it, which helps a reader judge what kind of adviser he is.
Read more
Shah says the best subscription programs convert only about 20% of repeat customers, leaving 80% who reorder on their own with no cadence and no one managing the relationship.
He notes Obvi has scaled past $100 million in revenue, and that he worked with rePete by Bold Commerce on a published guide about reaching that other 80%.
The post lists what the guide covers: fewer than 20% of cancellations happen because the customer stopped wanting the product, a shift from scheduled to predicted commerce using AI to time reorder prompts, and a case where Built Brands grew reorder revenue four times faster than subscription revenue in 60 days without cannibalising existing subscribers.
For brands planning a campaign
A brand treating subscriptions as its whole retention story can use this to justify a second, unmanaged-reorder channel and to test reorder prediction against a control group.
It is a short, specific lesson about where money is actually lost in paid acquisition, with numbers attached rather than opinions.
Read more
A supplement brand's plain advertorial page acquired customers at $454, while its polished branded page ran $682, a gap of $228 per customer on the same product and traffic.
A listicle page was killed after passing $1,000 CAC having taken $5,000 of spend.
Shah's point is that media costs were steady throughout, so the page, not the ad account, explained the results, and that the expensive part is discovering this through live tests.
The prompt he gives feeds AI both winning and losing pages with their CAC and conversion, asks what the winners do differently in structure, opening, and objection handling, then asks which bucket a new concept resembles before spend goes behind it.
For brands planning a campaign
Creative and page concepts can be pre-screened against a brand's own winner/loser history before a test budget is committed, which lowers the cost of finding a losing page.
It is the most direct evidence that Ronak Shah works inside creator programs at volume, not just at the strategy level, and it names the bottleneck that slows them down.
Read more
A founder told Shah his creator program produces 30 to 50 videos a day, and the bottleneck was never making them but checking them before they could run.
In a regulated category a single unsubstantiated claim creates legal exposure, and the team had no time to read at that volume, so either output slowed or risky claims slipped through.
The system Shah describes transcribes every video and checks it against the claims the brand is not allowed to make, flagging the exact line, why it is a risk, and what to say instead.
The review that used to be the constraint now happens in minutes.
For brands planning a campaign
A brand scaling creator volume in a regulated category can add automated claim screening as a condition of scaling, which is often the objection that blocks the program in the first place.
It is a small, concrete example of the kind of operational task Ronak Shah removes with AI, useful for judging whether his method transfers to work a brand actually has.
Read more
Shah describes sourcing a manufacturer as a job that stalls launches for about a month: hunting directories, working out who makes the product, finding the right contact, and writing each inquiry so it does not read like a copy-paste blast.
Handing the whole task to AI produced research on manufacturers that fit the product, the right person to contact at each, and ten personalised inquiry emails in one sitting.
His framing is that the work which quietly delays a launch is often the work AI clears in an afternoon.
For brands planning a campaign
Product-launch delays caused by supplier sourcing are a separate, testable use case for the same AI-assist method, and one that does not touch brand or legal risk.
This is where the reader learns what Ronak Shah does commercially beyond posting tips, including a partner he states he works with, so it should be read alongside rather than instead of his own advice.
Read more
Shah states he had been working closely with ZyG for six weeks before posting, and that ZyG came out of stealth with $118 million raised and the team behind ironSource.
He also states Obvi has done over $100 million in revenue and has bought research products and hired agencies, and that he ran one of Obvi's hero products through the system to test it.
What came back, in his account, was an AI-built market report: a $3.6 billion sub-category growing 14.5% a year and rated saturated, nine direct competitors running 1,216 active Meta ads between the top four, competitor ad teardowns with verbatim customer comments from Reddit, TikTok, and Instagram, 37 selling points tagged exclusive, distinctive, or common, and a moat audit that scored Obvi "Basic."
He says the criticism is why he trusts it.
The commercial model he describes: the report is free, accepted brands get a marketability test funded by $10,000 of ZyG's own media spend, and if ZyG takes the brand on it runs the work end to end for a percentage of sales.
For brands planning a campaign
Competitive-teardown work that would normally consume weeks of agency time can be produced before a campaign is scoped, and a brand can ask what the provider earns and when before treating the audit as neutral.
It shows how Ronak Shah distinguishes dangerous problems from defensible ones, which is a useful filter for anyone getting numbers reviewed.
Read more
Shah ran a founder's financial model through AI before a raise and found three problems in minutes: lifetime value math that did not match the revenue build, a chunk of ad spend missing from the budget, and deal terms on one tab contradicting another.
He argues the dangerous thing in a model is not a wrong assumption, because an assumption can be defended, but an internal contradiction, because it signals to an investor that the founder does not know their own numbers.
His explanation for why these survive is repetition: after the hundredth look at the same spreadsheet, the errors stop being visible.
The prompt asks AI to check every tab for internal inconsistencies the way a sceptical investor would read it.
For brands planning a campaign
Before a raise or a large spend commitment, an automated consistency pass over the model is cheap insurance, and mismatches between LTV math and the revenue build are usually the first thing a sharp backer spots.
It is the most human piece in the set and the only one that shows how Ronak Shah judges a partner, which is the question a brand should be asking of him in reverse.
Read more
Nikolaus from Whop flew from California to New Jersey for a dinner that was meant to be about a partnership, and the conversation drifted to Halloween.
Shah describes being the street's king-size candy house, a tradition tied to a family member's former job at Mars, and joking that a new job at Haleon ends it because you cannot hand out Advil to seven-year-olds.
A box of over 200 Mars and Skittles bars arrived at the front step the next morning, not as part of a pitch.
Shah's conclusion is that the deal was not the deck or the follow-up email but remembering one small detail and acting on it within 48 hours, and that this is what makes him want to see how the partnership is handled.
For brands planning a campaign
When a brand evaluates creators or agencies, evidence of how they follow through on small details is a better signal than a proposal, and this post is Shah's own stated standard for the same behaviour.
A numbered AI tip series, each built on one real operational problem6
Documented partnerships and the incentive structures behind them3
Retention diagnosed through cohorts, not the revenue line2
Testing landing pages and financial models before money is committed2
AI applied to the operational bottlenecks that delay launches2
AI-assisted creator sourcing and seeding at scale1
A numbered AI tip series, each built on one real operational problem
The supplied posts include a recurring series of numbered AI tips (numbers 52 and 56 through 61 appear).
Read more
Each follows the same shape: one concrete problem from a founder or brand, the numbers behind it, the AI-assisted fix, and a copy-ready prompt. The problems are varied but always operational, covering a landing page that costs more per customer than a plainer version, a brand whose headline growth was hiding an 80% one-and-done customer base, 30 to 50 creator videos a day that nobody had time to compliance-check, and a financial model that contradicted itself across tabs.
Posts name the inputs they feed the model (order history, video transcripts, landing-page conversion and acquisition-cost data, every tab of a financial model) and quote the prompt in full so a reader can repeat the exercise with their own data.
For brands planning a campaign
This is teach-by-example content rather than a productized service. A brand can lift the prompts as internal templates today, and any paid engagement would likely be judged on whether the same hands-on pattern survives contact with the brand's actual data.
Ronak Shah argues that most brands' creator programs stall around 20 people because finding hundreds of small accounts that genuinely fit a product cannot be done by scrolling a feed.
Read more
The recommended approach is to point AI at a category, pull nano and micro creators (1,000 to 50,000 followers) who already post about similar products, and score them on fit, real engagement, and whether they drive sales rather than on follower count, producing a seeding list of the 200 most likely to convert.
The post supplies the prompt: find nano and micro creators in a follower band who post about a category and whose audience matches the customer, then return handle, posting frequency on product content, real engagement rate, and a fit score.
The argument rests on a recounted case, a nine-figure beauty brand that switched off paid advertising for 18 months, built a creator base, and used it as the engine when TikTok Shop launched in the US in 2023. No client name or performance data is given.
For brands planning a campaign
Directly relevant to a brand brief on creator-led acquisition. The focus is on the sourcing and scoring list itself, not on negotiating individual creator deals, which is where an advisory engagement would likely start.
Retention diagnosed through cohorts, not the revenue line
Two posts make the same case from different angles: a rising revenue line and a leaking customer base look identical until you pull the cohorts.
Read more
One founder's brand looked healthy on top-line growth, but only 0.2% of revenue came from subscriptions, more than 80% of customers bought once and never returned, only 1.2% reached a fifth order, and average order value had slipped from $40 to $36. The other post states that even the best subscription programs convert only about 20% of repeat customers, leaving the other 80% reordering with no cadence and no one managing the relationship.
The recommended diagnostic is to build a repeat-purchase curve from full order history (second, third, fourth, fifth order rates and days between each) and determine whether growth comes from returning customers or from new ones replacing lost ones. Recommended interventions are to survey churned customers before building a loyalty program and to move from scheduled commerce toward predicting when a specific customer is ready to reorder.
For brands planning a campaign
For a brand with a repeat-purchase brief, the approach is diagnostic first and program second: measure the cohort behaviour, identify whether the problem is loyalty or acquisition masking churn, then choose the intervention.
Testing landing pages and financial models before money is committed
Two posts apply AI to decisions that are expensive to get wrong.
Read more
One covers a supplement brand whose plain advertorial page beat its polished branded page by $228 per customer ($454 versus $682 acquisition cost, with a listicle page killed at over $1,000); the expensive part is not picking the wrong page but paying thousands in ad spend to find out. The other covers a pre-raise financial model audit that surfaced a lifetime-value calculation that did not match the revenue build, missing ad spend, and deal terms that contradicted each other across tabs.
For pages, feed the model past winners and losers with their acquisition costs, have it work out what winners do differently in structure, opening, and objection handling, then classify a new concept before spend. For models, ask for internal inconsistencies the way a skeptical investor would read them, listing every place the model contradicts itself.
For brands planning a campaign
Shows a habit of inserting a cheap review pass before an expensive commitment. A pre-launch review is a plausible deliverable in a paid engagement, particularly for a brand spending heavily on paid acquisition.
AI applied to the operational bottlenecks that delay launches
Two posts address work that stalls a launch rather than work that markets it.
Read more
One describes finding and contacting 10 contract manufacturers for a new product in an afternoon, where the model researched manufacturers that fit the product, found the right person at each, and drafted 10 personalised inquiry emails, collapsing a month of stop-start work. The other describes a founder whose creator program produced 30 to 50 videos a day, where the bottleneck was never making content but clearing it: every video was transcribed and checked against prohibited claims before going live, flagging the exact line, the risk, and suggested compliant wording.
Both posts give the AI a defined, repetitive task with a clear input (a product spec, a day's video transcripts) and a defined output (a contact list with drafted emails, a flagged claims list with replacement wording).
For brands planning a campaign
The compliance-review post is the more relevant one for a brand running a large creator program: it addresses the review capacity that limits how much creator content can safely go live, which is a practical constraint on scaling creator-led acquisition.
Documented partnerships and the incentive structures behind them
Ronak Shah writes publicly about the companies they work with and why.
Read more
The most detailed is ZyG, described as out of stealth with $118 million raised and the team that built ironSource behind it.
Shah says they ran their own brand, Obvi, through the system before posting: one hero product submitted through a short intake returned a full AI-generated market analysis covering sub-category size ($3.6 billion, growing 14.5% a year, rated saturated), competitor ad teardowns, selling points mapped against the field, and a moat audit that scored the brand Basic.
The post emphasises the incentive structure: a free analysis, a marketability test funded with the partner's own media spend, and an end-to-end model earning only a percentage of sales.
A separate post describes teaming up with rePete by Bold Commerce on a free resource guide about the repeat customers that subscription programs do not reach.
A third describes a dinner with Nikolaus from Whop, where a remembered detail from the conversation turned into a delivered package within 48 hours, which Shah frames as the real substance of partnership development.
The repeated pattern is to try a product on their own brand first, then describe it with the specific numbers and the reservations included (the moat audit that scored them Basic is presented as the reason to trust the analysis).
The posts do not state whether any of these relationships involve payment, and a partner hashtag is not evidence of a paid arrangement. The partnership details are the account Ronak Shah gives, not independently confirmed terms.
For brands planning a campaign
For a brand considering a paid advisory engagement, the record suggests an approach of testing before recommending and a stated preference for performance-aligned models. The Whop anecdote also signals that relationship management, not just analysis, is part of how Shah operates.
Ronak says he started working closely with ZyG about six weeks before the post and announces its launch out of stealth with $118 million raised.
Read more
He promotes its free ZyGnal market-analysis product and tags the post #zygpartner; the post does not spell out the commercial terms of the relationship.
Original disclosure
Six weeks ago, I started working closely with one of the most impressive teams and concepts I've seen, but I never talked about them publicly. Today I can: ZyG is out of stealth, with $118M raised and the team that built ironSource behind it.
Ronak says Obvi teamed up with rePete by Bold Commerce to reach the roughly 80% of repeat customers who reorder without a subscription, and that the two published a free resource guide together.
Read more
The post presents the guide as a joint asset built on data from Obvi and other DTC brands.
Original disclosure
So we teamed up with rePete by Bold Commerce to go after the other 80%, and just published a full resource guide breaking down how we - and other leading DTC brands - are doing it.
Ronak writes that a Whop representative flew from California to New Jersey to take him to dinner to discuss partnership.
Read more
He closes by saying he looks forward to seeing how Whop treats a partnership. The post describes a partnership conversation and an impressed reaction, not a signed or ongoing agreement.
Original disclosure
We were there to talk partnership, but we ended up talking about Halloween... If that's how Whop treats a dinner conversation, I can't wait to see how they treat a partnership.
Ronak Shah writes for people running consumer brands who want to see exactly where AI removes friction from growth work, rather than a general argument that AI matters.
Read more
The clearest entry point is the numbered AI-tip series: each post names one founder or brand problem, gives the numbers behind it, describes the fix, and publishes the prompt so a reader can rerun it on their own data.
If repeat purchase is your problem, start with the cohort posts, which argue that a rising revenue line and a leaking customer base look identical until you pull the repeat-purchase curve, and that subscriptions reach only about a fifth of repeat customers.
If creator-led acquisition is your problem, start with the creator-sourcing post, which says programs stall around twenty people because finding hundreds of small accounts cannot be done by scrolling a feed, then the compliance-review post, which treats review capacity rather than content volume as the real ceiling on how much creator content can safely go live.
The partner posts show a consistent habit of testing a company's product on their own brand before saying anything about it publicly.
The tags below sort the work by subject, specialty, format, and the audience it addresses so you can pick a track and read it in order.
Both write operator-level analysis of how demand reaches a marketplace buyer.
Read more
McGraw takes a signal that looks like a decline — product detail page views falling for the client portfolio around Prime Day — and reads it alongside improved conversion, orders, and ROAS, citing Adobe's finding that retail traffic from AI sources grew 89% year over year and converted 40% better, and Salesforce data on AI assistants in the buying journey.
The argument is that more of the research and comparison now happens before the shopper reaches a listing.
Shah makes the same move from the creator side, arguing that a creator can create demand on TikTok that then closes on Amazon, and that judging a channel only on revenue that closes inside it is becoming hard to defend.
McGraw's evidence comes from an agency's client portfolio and third-party studies of Amazon marketplace behaviour, and the posts stay on channel economics and attribution. Shah packages the response as a repeatable AI task with the prompt printed in the post, and ranges wider — retention cohorts, landing-page testing, compliance review, manufacturer sourcing — rather than staying on marketplace traffic.
Both work the same question: is a TikTok Shop programme acquiring new customers, or paying to re-buy ones the brand already owns?
Read more
Bjorseth sets out the CAC and lifetime-value math, cites a brand whose in-platform repeat rate climbed from 30% in January to 72% in August, argues that a million dollars of TikTok Shop revenue and a million dollars of Shopify revenue are not the same asset, and lists GMV Max's limits — no retargeting, no objective other than conversion, no way to exclude existing customers — as a reason ads keep hitting people who already subscribe elsewhere. Shah shares the concern with repeat-purchase behaviour and with moving profitable revenue between channels at a breakeven.
Platform mechanics are the centre for Bjorseth: ad controls, channel pricing, cohort metrics, often tied to an episode of their show. Shah treats TikTok Shop as one case inside a wider set of operational fixes, and publishes the prompt that produces the fix.
Both argue that retention should be fixed before more acquisition spend, and both argue it from having run the consumer side.
Read more
Waiche writes that they would not spend another dollar on acquisition before fixing retention, and describes building paid membership at Adore Me, where members paid $39.95 a month and received the same amount in store credit to spend as they chose, then carrying that model to Shopify as Subscribfy. Shah reaches the same territory from the data side, arguing that retention is what makes growth less dependent on acquisition in the first place and that even the best subscription programs convert only about 20% of repeat customers.
Waiche's recommendation is a specific commercial lever, paid membership and store credit, and the posts read as a vendor's case drawn from years of merchant data. Shah starts with a cohort diagnostic — the repeat-purchase curve built from full order history — before any intervention is chosen, and puts the prompt in the post so a reader can run the same exercise.
For a DTC brand considering advisory work on AI-assisted creator sourcing or repeat-purchase retention, Ronak Shah’s LinkedIn posts show relevant subject matter: one describes finding and scoring nano- and micro-creators, and another discusses using AI to analyze repeat-order patterns.
Read more
These examples show what his content covers, not who his audience is or how it is composed.
The available dated evidence does not establish audience demographics or professional roles.
Before deciding on a paid engagement, ask for recent audience survey results or another documented source addressing the buyer’s target roles and customer segments.
Ronak Shah has not published a sponsorship policy, media kit, or rate card, so there is no documented package a brand can simply buy.
Read more
The posts show two kinds of commercial collaboration.
Ronak promoted ZyG, a market-research platform that launched out of stealth with $118 million raised, in a founder-voice post carrying the #zygpartner hashtag and a link in the first comment; before posting, Ronak ran a hero Obvi product through the platform's short intake and reported the results that came back days later, including an unflattering score that labelled Obvi's moat "Basic."
Ronak also co-produced a free gated guide with rePete by Bold Commerce on repeat purchasing beyond subscriptions, using figures such as Built Brands growing reorder revenue four times faster than subscription revenue in 60 days.
Both examples are teaching-first and tied to a named product or service; neither discloses terms, deliverables, exclusivity, or compensation, and a partner hashtag alone does not establish payment.
One post about a dinner with Whop shows how partnership conversations reach Ronak — relationship-led, with the partner acting on a small personal detail within two days — rather than through a published intake form.
Treat any format as something to negotiate directly, and expect to give Ronak enough access to a real product, account, or data set that he can say something specific rather than generic.
Partner post built on a tested product
Buyer requirement
Expect to provide real product access and a few days of hands-on use before Ronak posts.
The ZyG post describes submitting an Obvi product through a short intake, waiting a few days for the AI-generated market report, and then publishing findings that included the critical parts — a saturated sub-category, nine direct competitors, and a "Basic" moat score — alongside the pitch.
No post length, review process, link placement, hashtag convention, or exclusivity term is documented, so the brand has to propose them, and it should be ready for the write-up to include weaknesses as well as strengths.
Co-produced resource guide
Buyer requirement
Bring a partner platform, a landing page to host the download, and permission to publish other brands' performance data.
The rePete by Bold Commerce guide is free to the reader and positions the partner's technology alongside Ronak's own Obvi numbers and third-party results such as Built Brands' 4x reorder growth.
The asset name-checks the partner in the body of Ronak's post and drives to a gated link, so the partner supplies the hosting and the supporting case studies while Ronak supplies the audience and the operator-side narrative.
Strong fit for a fixed-scope advisory engagement. Two conditions apply: the brand defines the deliverables and the data access before the engagement starts, and it buys analysis and method transfer rather than content or audience reach.
Read more
Brief
A direct-to-consumer brand past $10M in revenue wants two things: more customers from creator-led acquisition, and more repeat orders from customers it already has. The team has order history and paid-media data, but its creator program is a handful of names, and it is deciding whether to pay an adviser to fix both.
Reasons
Both halves of the brief are addressed in published work. For acquisition, the published method is to find nano and micro creators (1,000 to 50,000 followers) who already post about the category, then return each one's handle, how often they post product content, their real engagement rate, and a fit score, cut down to a seeding list of 200. For retention, the published method is to build a repeat-purchase curve from full order history: what share of customers place a second, third, fourth and fifth order, and how many days pass between each.
The retention work starts with a diagnosis rather than a product pitch. In the example Ronak Shah describes, a fast-growing brand looked healthy on revenue while 0.2% of revenue came from subscriptions, more than 80% of customers bought once and never returned, only 1.2% reached a fifth order, and average order value had slipped from $40 to $36. The recommendation there was to survey lapsed customers before building a loyalty program, which is the order of work a brand with this brief should want.
The method is handed over rather than kept behind the engagement. The posts quote the prompts in full and name the inputs, so a brand's own team can rerun the creator search and the cohort analysis after the adviser is gone.
There is operator context behind the advice. Ronak Shah states that Obvi has passed $100M in revenue, and describes co-producing a free guide with rePete by Bold Commerce about the repeat customers subscription programs do not reach, including a cited case where reorder revenue grew four times faster than subscription revenue in 60 days without cannibalising subscribers.
They test before recommending, and report the unflattering result. Before posting about the market-research platform ZyG, they ran a hero Obvi product through it and published the finding that the moat audit scored the brand "Basic." That is the behaviour a brand wants when it is paying for a real answer.
They work inside creator programs at volume, not only at strategy level. One post describes a program producing 30 to 50 videos a day where the bottleneck was checking each video against prohibited claims before it could run, and the fix was to transcribe and screen every video, flagging the exact line, the risk, and replacement wording.
Limits
No package, rate, scope, or contract is published. The brand has to write the brief and the deliverables itself.
The named cases are recounted, not published client results with an independent measurement. Agree in advance how success will be measured.
The methods need real data access: order-level history for the cohort work, and product and category detail for the creator search. Without that, the output will be generic.
This is analysis and advisory work. Nothing supplied shows Ronak Shah producing creator video or appearing on camera, so do not buy it expecting content.
No audience composition evidence was supplied, so this is not a reach buy.
When to choose a specialist
Poor fit as written. This is not a judgement of the advice quality, which is a separate question. The buy depends on things the available evidence does not support.
Read more
Brief
A consumer brand wants a paid creator post plus a short video review, priced on reach: a guaranteed number of impressions against women aged 25 to 40 in the United States, posted from the creator's Instagram and TikTok accounts.
Reasons
No video or audio material was supplied. Everything reviewed is written LinkedIn posts, and nothing shows on-camera work or short-form video production.
The Instagram account appears in the record with a follower figure of about 5,200 and no post content at all. LinkedIn is the only channel with content behind it, at about 41,800 followers. A buy spread across Instagram and TikTok has no working evidence underneath it.
There is no audience composition evidence. A follower count is not a demographic match, and nothing supplied describes who follows these accounts or what they buy.
There is no media kit, rate card, or sponsorship policy, which is the basic paperwork a reach or impressions buy is priced against.
The method Ronak Shah publishes argues against this kind of purchase. It says to score creators on fit, real engagement, and whether they actually drive sales, not on follower count. Buying this person to sell follower-count reach asks them to do the thing their own material tells brands to stop doing.
The commercial examples on record are a partner post about a product they tested and use, and a co-produced resource guide with a named partner. Neither is an unrelated paid promotion, so there is no evidence they sell that kind of placement.
What would change it
A documented audience survey or platform demographic data for the accounts in the brief.
Video samples on the channels the campaign needs, so the format can actually be judged.
Published formats, terms, and rates. If the brand instead wants a written breakdown of its category and a scored shortlist of creators, the same person becomes a reasonable fit for a different brief.
A worked campaign example
Angle
Start with the retention picture, not the creator list. Use the repeat-purchase curve to establish whether the problem is acquisition, retention, or both, then build the creator seeding list against that answer. Measure the seeded creators on whether their customers reorder, not just on first purchase.
Brief
The growth-stage DTC brand from the first example, running a 90-day paid advisory on creator-led acquisition and repeat purchase.
Deliverables
A cohort model of repeat purchasing that the brand's team can rerun monthly.
A scored list of 200 nano and micro creators with the reason each one was included.
A written seeding pilot brief with defined success measures and a holdout group.
A compliance-screen template for creator video claims, if the category needs one.
How to judge it
Agree the measures before the engagement starts. Sensible ones: second-order rate for the creator-acquired cohort against a holdout, cost per customer from seeding against paid media, and whether the brand's team can run the curve and the creator list on its own by month two. If none of those can be measured with the data on hand, the engagement should not run.
What this example is not: Ronak Shah has not published this package, these steps, or these measures. The plan is assembled from methods described in the published work, and any real engagement has to be negotiated directly.
Steps
Detail
24 months of order history with order number, order date, order value, and acquisition channel; the product catalogue and a clear definition of the category; and the paid-media page history with acquisition cost and conversion rate for each page.
Title
Hand over the data
Detail
What share of customers place a second, third, fourth and fifth order, and how many days pass between each. The output states plainly whether revenue growth is coming from customers returning or from new customers replacing the ones leaving.
Title
Build the repeat-purchase curve
Detail
Compare reorder behaviour for customers who arrived through creator posts against those who arrived through paid media. This decides whether the next dollar goes to more seeding volume or to better reorder timing.
Title
Split the curve by acquisition source
Detail
Nano and micro creators, 1,000 to 50,000 followers, who already post about the category. Each gets a handle, posting frequency on product content, real engagement rate, and a fit score. Cut to 200 names, then seed product to a first wave.
Title
Build the seeding list
Detail
Before the seeding landing page and the creator brief go live, review them against the brand's own winning and losing pages with their acquisition costs, and ask which bucket the new concept resembles. The published benchmark is a plain advertorial at $454 per customer beating a polished branded page at $682, so this step is about catching a losing concept before the test budget is spent.
Title
Pre-screen the launch assets
Detail
Transcribe the first wave of creator videos and check each one against the claims the brand cannot make, flagging the exact line, why it is a risk, and what to say instead. This is the step that lets creator volume scale without legal exposure.
Title
Add a compliance screen if the category is regulated
Detail
Leave the prompts, the cohort model, the scored creator list, and the pilot's success measures with the brand's own team, so the curve and the list can be rerun in month two without the adviser.
Choose Jonathan McGraw when the brand's real blocker is proving that creator and TikTok Shop spend pays off across the whole business, not building the creator list itself.
Read more
The published work is cross-channel measurement: portfolio data from Prime Day where product-detail-page views fell while conversion, orders and ROAS improved, an Adobe figure that AI-sourced traffic to US retail sites rose 89% year on year and converted 40% better, and a SmartScout finding that the top 10,000 Amazon brands with a TikTok Shop presence grew Amazon revenue 5.2 times faster than those without one.
The framing question is how much of the purchase decision was won before the shopper reached the product page, which is the question a finance lead or board asks when a channel gets renewed or cut.
Use this when the deliverable is a defensible cross-channel read on whether creator investment is working.
The evidence on record is a strategic frame for reading the halo, not a handover method.
There is no published creator scoring list, no repeat-purchase curve or cohort model, and nothing on reorder prediction, so the retention half of the brief is not covered by what is available.
A brand that wants its own team running the analysis after the engagement gets less reusable tooling than it would from an adviser who publishes the prompts and inputs.
Stronger on attribution and cross-channel strategy, weaker on method transfer and retention mechanics.
Choose Jake Bjorseth when the brand is already committed to TikTok Shop and the problem sits inside the channel: ad dollars re-buying customers the brand already owns, and the limits of the platform's main ad product at higher spend.
Read more
The published argument is that the in-platform repeat rate should be driven low for net-new acquisition, with one named brand climbing from a 30% repeat rate in January to 72% in August, meaning most ad spend stopped acquiring new customers and started re-buying ones who would have bought elsewhere.
The companion breakdown of GMV Max lists the constraints plainly: no retargeting, no objective other than conversion, no way to exclude existing customers, and no way to run specific creatives at specific targets.
It then gives concrete mitigations: differentiate assortment between channels and price TikTok Shop higher, cut 5-10% of ad budget to contract creators around new SKUs and run reach ads to bring net-new consumers into the algorithm, and retire creatives before fatigue sets in.
Use this when the next 90 days are about making TikTok Shop itself more profitable and less cannibalistic.
The material is channel-specific operator advice, drawn from an episode summary and an ad-product teardown, not a full method the brand can rerun unaided.
There is no published creator sourcing pipeline for nano and micro creators, no general retention diagnosis, and no cohort model, so a brand that needs both halves of the brief still has to source the acquisition and repeat-purchase work elsewhere.
The vantage is agency and platform-side, so expect a TikTok-Shop-scoped engagement rather than neutral method transfer across channels.
Choose Samy Waiche when the brand has already decided retention is the constraint and wants a membership or subscription product built and run, especially in a category where repeat delivery of the same item does not fit how customers shop.
Read more
The published case is the Adore Me model: members paid $39.95 a month, received $39.95 in store credit, and chose what to buy and when, which was tested across millions of transactions before becoming Subscribfy, an all-in-one retention platform.
The public position is that improving retention changes the economics of growth, and that paid membership with store credit, member pricing and shipping benefits gives customers a reason to keep choosing the brand beyond the first purchase.
Use this when the deliverable is a working membership programme with defined member benefits, not an analysis the team runs itself.
The retention answer arrives inside a platform, so the engagement is a vendor relationship rather than neutral advice, and the brand should expect scope shaped around the product rather than a handover of a rerunnable cohort model.
On the acquisition half of the brief — finding, scoring and seeding nano and micro creators — the published material says nothing, so that work still has to be bought or built separately.
Less useful where the brand wants to keep the retention build in-house from the start.
This profile is based on the full text of nine LinkedIn posts from Ronak Shah's account, read as the primary evidence for identity, recurring ideas, brand relationships, and campaign fit.
Written posts are the evidence; supplied follower counts, titles, and snippets are not treated as proof of reach or content.
No video, audio, transcript, or Instagram post content was supplied.
Most posts have no reliable publication date, and the supplied material does not establish a signature book, show, or earlier body of work; both remain gaps rather than claims.
Recent engagement
Across their last 7 LinkedIn posts, the median post got 51 reactions and 18 comments. Checked 2026-11-02.
