1. The Setup
This report covers a simulated Instagram and TikTok in-feed test for Wholey, evaluating a single creative, "The anatomy of a Wholey bowl 🫶", across three audience groups in the German market: Convenience-Oriented Young Professionals, Active Lifestyle / Functional-Nutrition Adults and Mainstream Supermarket Family Shoppers. Across 41 rounds and 300 agents, the simulation captured not only who clicked and converted, but how agents reacted, what they objected to, and which elements of the creative held up under repeated exposure.
The central question was straightforward. Does this ad work, for whom, and why? More specifically, where does the creative earn attention, where does it lose trust, and which objections are problems with the ad versus problems with the product or price point.
The Ad Creatives
The creative is a single static image, a studio flat-lay of a prepared Dragon Smoothie Bowl, shot from above on a clean background. Ingredient attributes are called out using hand-drawn labels: "no nasties," "supereasy," and "superyummy," alongside a visual breakdown of the bowl's components. The copy runs in a German-English mixed register, and the caption, "The anatomy of a Wholey bowl 🫶", leans into an educational, dissection-style framing rather than a conventional benefit claim.
There is no call-to-action button, no price point, and no brand logo visible on the image itself. That creative choice is the single thread running through most of the friction the simulation surfaces: across multiple groups and dozens of rounds, agents who were initially curious found themselves unable to act because the ad withheld the information they needed to move forward.
The question Imago set out to answer: could a simulation tell Wholey which audiences this creative is actually working for, which segments it is actively alienating, and what specific changes would close the gap between scroll-stop and conversion, before a single euro of real spend is committed?
How Imago Approached This
Imago's method is built around the idea that before you test an ad in market, you can test it against simulated audiences that behave like real people. Not simple A/B testing or focus groups. A simulation that generates the kind of reactions, hesitations, and language your actual customers would have when they scroll past your ad. Here is what that looked like in practice for this campaign.
1. Inputs
We fed the system the two ad creatives, the full copy, the brand context, and the campaign brief. This includes everything a real person would see: the image, the headline, the supporting copy, the product price range, and the platform context. The simulation needs enough signal to construct a credible impression of the ad as it would appear in a real feed.
2. Defining the Personas
Based on the brand context and campaign targeting, we defined three audience segments. Each one represents a meaningfully different type of consumer the ad was likely to reach in the German market.
- Convenience-Oriented Young Professional: Urban professionals in their late twenties, city-based and time-pressured, for whom breakfast is an afterthought solved in under two minutes. They are in-market for convenience products in the abstract but evaluate everything against a cost-and-effort benchmark that most premium formats fail.
- Active Lifestyle / Functional-Nutrition Adult: Gym-committed adults in their early thirties who track macros, compare protein-per-euro, and apply the same scrutiny to a smoothie bowl as they would to a whey supplement. A compelling visual is not enough; the label has to clear their threshold.
- Mainstream Supermarket Family Shopper: Working parents in their forties doing a weekly shop for the household, price-aware, coupon-using, and genuinely interested in lower-sugar options, but only when the price-per-100g comparison holds up.
3. Running the Simulation
300 AI agents, 100 per group, were exposed to the ad across 41 rounds and three contextual states, grocery-shopping Instagram, late-night TikTok, lunch-break TikTok, and workout-recovery Instagram, simulating the situational breadth of a real in-feed placement. Each agent could react the way a real person would: posting, commenting, liking, ignoring, clicking, or purchasing, with every reaction driven by how well the ad landed against their individual profile. Agents weren't exposed to the ad in every round, mirroring real-world impression frequency, and the simulation tracked each agent's accumulating purchase intent over time, gating actual conversion on a prior click having occurred, the same way paid-ad attribution works in practice.
4. What the Simulation Captures
The output is two things working together. First, a set of quantitative metrics, CTR, engagement rate, purchase rate, sentiment scores, ROAS, and CAC, measured per segment and in aggregate. Second, and more importantly, the verbatim language agents used when they reacted: what stopped the scroll, what broke trust, and what question the ad consistently failed to answer. That combination shows not just which segment performed best, but exactly why the same creative that efficiently converts family shoppers actively alienates time-pressed young professionals, and what would need to change for that gap to close. That is what the following sections break down.
2. What the Simulation Found
Across all four groups and 41 simulation rounds, the "anatomy of a Wholey bowl" creative generated a ROAS of 4.26, indicating the creative does stop the scroll at an above-average rate. Among agents who reacted, 77.9% of their actions were active rather than passive, posts, comments, and quote-shares rather than idle likes, confirming that when the ad lands, it generates genuine engagement rather than polite indifference.
The overall purchase rate was 8%, but the report made clear that the campaign is generating considerably more curiosity than it is converting into commitment. Affinity without action was the dominant pattern across the run.
The Three Groups at a Glance
| Segment | Purchase Rate | Positive Sentiment | CTR | ROAS | CAC |
|---|---|---|---|---|---|
| Mainstream Supermarket Family Shopper | 13.3% | 34.9% | 4.0% | 7.09x | $3.52 |
| Active Lifestyle / Functional-Nutrition Adult | 10.7% | 31.2% | 3.9% | 5.67x | $4.41 |
| Convenience-Focused Young Professionals | 5.3% | 41.5% | 3.8% | 2.84x | $8.81 |
The performance split across groups is where the simulation's most actionable findings live. The same creative that worked efficiently for one audience actively alienated another, and the gap is wide enough to carry real targeting implications.
What the Ad Got Right
The creative's scroll-stopping power is real. An 18.3% overall response rate across three distinct audience types and three exposure contexts confirms the flat-lay aesthetic, hand-drawn labels, and ingredient-forward framing are capable of generating genuine attention in a crowded social feed. The "Natural Sweetness from dates" (part of the Instagram Page description) claim in particular operated as a consistent hook: it was one of the most frequently referenced elements in positive agent reactions across every group, surfacing in comments from family shoppers looking to reduce sugar for their children, from young professionals curious about plant-based alternatives, and from active adults scanning for clean-label options.
Engagement quality was also strong. A 77.9% active engagement rate means that the agents who did stop were motivated enough to post, share, or comment and not just to double-tap and move on. That translates into organic social spread within the simulated environments and suggests the ad has the raw material to generate word-of-mouth when it reaches the right audience in the right context.
What the Ad Struggled With
The single most consistent friction point across all three groups and all three exposure contexts was the same: the creative withholds the information people need to act. No brand name on the image. No price point. No retailer or availability signal. For every group, the agent evidence shows a version of the same loop: the ad creates curiosity, agents try to resolve that curiosity, and when they cannot, the initial positive reaction cools. One agent in the Convenience-Oriented Young Professional group moved from "if this is Wholey from Rewe I might actually impulse buy it on Flink right now" to a flat dismissal after the purchase path proved too opaque to navigate quickly. The ad opens a door it does not help people walk through.
A secondary issue is the ad's framing mismatch with audiences who bring a price-comparison orientation to food decisions. Agents across the DACH Convenience-Focused Young Professionals and Convenience-Oriented Young Professional groups benchmarked the bowl against their existing breakfast setups and consistently concluded the ad's proposition did not pass their value test. This is not a creative execution problem in the traditional sense. It reflects a genuine tension between the product's positioning and the purchase logic of one of the three segments tested.
The Mainstream Supermarket Family Shopper group, by contrast, applies a similar comparison mindset but starts from a different frame of reference (lower-sugar, ingredient-quality, family value) which is where the ad's claims land most cleanly.
"5 minutes?? I would've been in and out in 2 with a random bar that looked good. Ain't nobody got time for price per 100g math at 8am before work."
Convenience-Oriented Young Professional"Natural Sweetness from dates sounds really exciting! I'm always on the lookout for breakfast options with less sugar for my boys."
Mainstream Supermarket Family ShopperThe two quotes above illustrate the audience split more clearly than any metric. The same ad reads as irrelevant friction to one group and as a genuine product discovery to another. That divergence is what the group deep-dives in the following sections break down in detail.
"Mainstream Supermarket Family Shopper"
This was the strongest-performing segment in the simulation. The highest purchase rate, the lowest CAC, and the highest engagement rate of any group and uniquely, the group where the ad's ingredient-quality framing landed as a genuine purchase signal rather than a decorative claim. For a shopper who is already scanning breakfast shelves for lower-sugar options and reading on-pack claims before anything else, an ad that leads with a recognisable, clean-label sweetener alternative speaks directly to an active search that is already underway.
The engagement rate of 85.5% was the highest in the entire simulation, meaning that among agents who noticed the ad, the overwhelming majority engaged actively rather than passively (they shared, compared, and discussed). That social behaviour is the mechanism behind the purchase rate. When the ad landed, it generated enough genuine interest to circulate within the group's simulated community layer, giving the creative reach beyond the initial impression. The CTR of 4.0% sits above cold-audience benchmarks, and the ROAS of 7.09 with a CAC of $3.52 reflects a group that, when it converts, does so efficiently.
The agent language in the positive reactions is consistent in what it responds to. One agent, early in the run, was direct: "'Süße aus Datteln' klingt richtig spannend! Ich bin auch immer auf der Suche nach Frühstücksoptionen mit weniger Zucker für meine Jungs."
"Natural Sweetness from dates sounds really exciting! I'm always on the lookout for breakfast options with less sugar for my boys."
The quote points to the underlying dynamic: agents in this group bring a comparison-shopping lens to every food purchase, and the ad's ingredient framing gave them something tangible to evaluate. That is a much stronger entry point than aesthetic appeal alone.
The negative reactions are clustered around a single concern: price and availability transparency. Once agents moved from the image to asking where the product could be bought, what it cost, and how it compared to their existing weekly shop, resistance built. Agents who did not convert consistently cited the absence of a price point or retailer signal as the reason they did not follow through. One agent's arc was representative: early rounds showed genuine curiosity and an intention to look for the product in-store at the next Rewe visit; by later rounds, when no easy purchase route had materialised, that intention had quietly dissolved. The interest was real. The conversion infrastructure was not there to catch it.
The broader implication for media planning is straightforward. This group represents the clearest case in the simulation for concentrated spend. The metrics are efficient, the agent behaviour is commercially legible, and the creative's core claims map onto a genuine purchase motivation this audience already holds. The ceiling is also visible: 86.7% of agents did not convert, and the qualitative record shows that most of them were not hostile to the product but they simply ran out of information before they could act. Adding a retailer availability signal and a price anchor to the creative, even in the caption rather than the image itself, would directly address the most commonly expressed barrier this group raised.
"Active Lifestyle / Functional-Nutrition Adult"
This group sits in the middle of the performance table with a 10.7% purchase rate and a ROAS of 5.67 look reasonable in isolation but the agent behaviour tells a more complicated story than the numbers suggest. The ad did generate genuine initial interest. Agents in this group are familiar with food brands like 3Bears and Verival, are comfortable evaluating new breakfast formats, and responded to the "no nasties" and ingredient-quality framing with real curiosity in early rounds. The CTR of 3.9% and engagement rate of 75.3% confirm the creative cut through for this audience.
The friction that emerged was specific and consistent. This group approaches food products with an evaluative mindset. They want to know what is in something and whether it earns its place in their routine. The ad's aesthetic-first, information-light creative did not give them enough to work with. Once agents started asking questions the ad could not answer from the image alone, frustration replaced curiosity. One agent was candid:
"Hiding the logo just wastes my time, I want to know the protein per serving and price per 100g, not play guessing games."
The 13 agents who did convert likely represent those whose prior orientation toward the bowl format was already positive enough to survive the information gap. The remaining 87 agents, most of whom engaged actively and had measurable awareness of the brand, did not cross the threshold, because the ad never gave them the specific signal they were looking for. There is a meaningful audience here but reaching them would require a different creative register: one that leads with what the product contains rather than how it looks.
"Convenience-Focused Young Professionals"
This group produced the most interesting internal split in the simulation. Their positive sentiment of 41.5% was the highest of any group yet their purchase rate of 5.3% and ROAS of 2.84 sit well below the campaign average. The gap between those two numbers is the finding. This group liked the ad more visibly than almost anyone else, and converted far less.
The agent language in the early rounds explains why. The ad's aesthetic landed strongly. The visual, the bowl format, the ingredient labelling generated genuine enthusiasm from agents who are already immersed in food content on Instagram and TikTok and respond to creative that looks the part. One agent described seeing the ad and thinking
"Omg aesthetic, Natural Sweetness from dates?? take my money"
which is the kind of impulsive, scroll-stopping reaction a brand wants. But that enthusiasm was consistently fragile. This group is also price-aware, comparison-minded, and well-informed about what alternatives cost. The core dynamic here is not that the ad failed to generate attention. The issue is that the ad created a moment of purchase intent without providing the infrastructure to act on it. No price, no retailer, no immediate path to buy. For an audience whose food decisions are impulsive and time-sensitive, interest that cannot be acted on immediately tends not to survive the next scroll.
This is Imago
What a real-world campaign cannot tell you is any of the above. It can report that certain audiences didn't convert, but not that the Convenience-Oriented Young Professional rejected the ad's register before they ever evaluated the product, that the DACH Convenience-Focused group was ready to buy until the purchase path ran out, or that the exact fix for the Family Shopper is a price anchor and a retailer name next to the ingredient claim they already responded to. The simulation identified the segment worth leading with, named the single creative change that would lift performance across all three groups, and gave Wholey clear grounds to deprioritise a targeting segment that the metrics alone would have taken months of real spend to surface, all before a single euro was committed.