A long-form diagnostic for PE partners on the variable conventional diligence cannot see. Why the smartest money is leaving consumer brands, what is being missed, and what changes once the variable becomes measurable.
We are five years into the most data-rich era in the history of capitalism. AI reads every review. Sentiment dashboards monitor every social platform in something approaching real time. Diligence processes run for weeks across financial, legal, commercial, operational and ESG dimensions, supported by tools and datasets that did not exist a decade ago.
And yet.
In 2024 and 2025, PE-backed consumer brand failures kept arriving. Outdoor Voices. Joann. Red Lobster. Party City. Forever 21. The Container Store. 99 Cents Only. Saks. Eddie Bauer. All bought at premium multiples by sophisticated firms. All collapsed during the hold.
The puzzle is not why some deals fail. Some always will. The puzzle is why so many keep failing despite more data than the industry has ever had access to.
This paper argues there is a single variable missing from the diligence stack. The rest of this section shows what the data looks like once you accept that the variable exists.
Three numbers, fast. The reader has heard most of them.
Source. MSCI Private Capital, via CEPR 2025. Annualised returns over the 2022 to Q3 2025 window. Consumer PE has materially underperformed the public market over the same holding period.
Multiples paid are too high. Returns are too low. Hold periods are too long. These are symptoms. The interesting part is what they are doing to the people inside the system.
The category-level retreat is already documented and named.
In October 2023, Carlyle disbanded its US Consumer, Media and Retail investing team. Four senior dealmakers, decades of relationships, networks and sourcing edge, asked to leave. The remaining lead was repositioned to manage out the existing portfolio rather than build new deals. Reuters reporting in March 2024 established a pattern that ran deeper than a single firm.
Four anchor sponsors of the consumer category. All effectively gone from new consumer investing inside a decade. Careers built on a category that no longer exists for them.
Source. McKinsey Global Private Markets Report 2026. The four-year exit window is the industry benchmark for when an LBO should be sold. 52% of all buyout-backed companies are now past it.
The personal stakes are not abstract. A senior PE partner's compensation is built on carry, the 20% performance share that pays out only when a fund clears its hurdle rate. For partners whose long-term wealth was modelled on carry from the 2018 to 2021 vintage consumer funds, the practical implication is years of high-intensity work, with the holiday home, the school fees, the retirement number, and the standing among peers all built on a line item that is now unlikely to materialise.
Paul Weiss's 2024 fund terms survey found that 64% of new PE funds now include interim GP clawback provisions. LPs have written the legal mechanism to take carry back off partners after the fact, expecting that they will need to use it. Hooke's analysis at Johns Hopkins, replicated by Institutional Investor, places several flagship mega-funds below the public market index over their fund life. Those partners will earn no carry at all.
The cleanest named cautionary tale is Casper. Founded 2014. Backed at peak by some of the best-known venture and growth investors in the world. Private valuation of $1.1bn in 2019. IPO in February 2020 at $12 a share, already a markdown from the $17 to $19 target range. Taken private in 2021 at $6.90 a share. Founder Philip Krim watched a category-defining brand collapse from celebrated to cautionary inside eighteen months. The diligence had been done. The product was good. The advertising was iconic. The data, on paper, was strong. None of it caught what was actually happening in the minds of repeat buyers.
CNBC's February 2026 piece on the industry's Darwinian era put the firm-level consequence more directly.
The reputational climate is sharper still. The Stop Wall Street Looting Act was reintroduced in 2024. "PE-backed" has become a near-pejorative phrase in mainstream business and political coverage. Coller Capital's 41st Barometer found that over half of public pension plans, endowments and foundations now identify reputational risk from PE activist criticism as a growing concern. Consumer brands, given their visibility, are the most exposed sub-sector to that drag. The partner who lost on Casper does not just lose carry. They lose the next allocation, the next fund, the next board seat, and, at some industry dinners, the next conversation.
PE's share of total private capital raised fell from 41% in 2024 to 33% in 2025. Buyout fundraising specifically hit a seven-year low of $414 billion. Source. PitchBook Q4 2025 Global Fundraising Report.
Where the money went is more revealing than the fact that it left.
This is not a verdict on whether consumer brands are good investments. It is a verdict on whether the industry's diligence processes can reliably tell the good ones from the bad ones. Capital flees what it cannot underwrite.
Multiples too high. Returns too low. Hold periods too long. Anchor sponsors walking away. Capital fleeing to assets that need no consumer to choose them. All of it happening while diligence has never been more sophisticated, AI has never been more powerful, the data has never been deeper, faster or cheaper.
So the obvious question, the one no one is asking out loud. Why is this still happening in 2026?
The failure is not vendor failure. The research houses do good work. The consultancies do good work. The operational diligence firms do good work. The failure is upstream of all of them. The failure is in what PE has decided to ask for. PE houses spend ten to twenty times more on operational and financial diligence than on anything that touches the consumer's mind. The unstated assumption is that finance and operations are real, and what the consumer thinks is decoration.
This is not new. Every category that has been disrupted held, at the point of disruption, a limiting belief about what counted as real and what counted as decoration. PE is in the same position now.
The clue that the belief is wrong is in McKinsey's own 2026 data. PE firms have been selling their brand-healthy assets first, leaving the brand-broken ones stuck in the 16,000-strong backlog. The portfolio is sorting itself by what consumers believe to be true about the brands. That sorting is the statistical fingerprint of the variable PE has decided does not count.
The thing the diligence stack does not count is one layer beneath the spreadsheet. It is in the human being who buys the product. The story they tell themselves at the moment they reach for the shelf, or scroll past it. The things they believe to be true about the brand, held in their mind as fact, generating the feelings that drive the behaviour that becomes the sales line. That is consumer belief. And it is measurable.
Each of the three brands below has been written about exhaustively in financial and operational terms. Each was diagnosable in belief terms long before the financial collapse confirmed it. What follows is the belief read on each, drawn from publicly available consumer signal that any diligence team could have pulled. Three brands. Three curves. The same shape, every time. Belief breaks first. Price follows.
Illustrative reconstruction. Public consumer signal shifted six to nine months ahead of the financial line. The pattern is consistent with the Brand Belief Score lead-time observed across categories.
Illustrative reconstruction. The retail-experience belief shifted in forums and review platforms from 2013 onward. Chapter 11 followed in September 2017. A two-to-three year lead time between belief and financial confirmation.
Sales is the score. It is not the early warning. The reason is a chain that every consumer transaction runs through, and the chain takes time. Once a reader sees the chain, the question of why sales lags answers itself.
Every purchase, in every category, runs through the same five-link sequence.
The sale is not the cause. The sale is the receipt. Four invisible stages happened first. By the time the till rings, the belief has already done its work. What this means in practice. Belief is upstream of sales. Always.
For one consumer the chain runs in seconds. For a market of millions, the chain runs in months. A belief shift in one shopper changes their next purchase. A belief shift across a whole consumer base only shows up in the sales line once millions of individual chains have re-run with the new belief in stage 2. That takes time. Roughly nine to twelve months, in most consumer categories.
A diligence team reading the EPOS data at month nine sees a healthy brand with stable share. A diligence team reading the consumer language at month one sees the belief already breaking. The first team underwrites the deal. The second team prices the risk correctly.
This is the variable the diligence stack does not count. It is measurable. It is observable. It is already written down, in public, in the consumer's own words. The next section shows what it looks like when you plot it.
A belief can change in an instant. Behaviour, across a whole market, does not. That gap is why belief is the leading indicator, and why the lag runs 6 to 18 months rather than a fixed date.
A person can change what they think, feel and do the moment they meet a new brand experience. Many do not. Beliefs that have hardened over years run on automatic. The neural pathways are deep, and without a fresh experience to judge the brand against, the old belief, and the behaviour it drives, simply persists. Some people update fast. Others are slow. Most sit somewhere between.
Now multiply that across a market. A brand turning over £20m has tens of thousands of buyers who once believed the same thing about it. Give them a new experience and behaviour shifts quickly for some, slowly for others, at every pace between. Summed across that population, the belief-to-behaviour shift feeds into the sales line gradually. That is the lag. The range is the mechanism showing itself. Not everyone moves at once.
There is an exception. Occasionally one definitive signal flips an entire consumer base almost overnight. We call it the shoaling effect, a whole shoal turning as one. In 1985 Coca-Cola replaced its formula with New Coke. The backlash was immediate and the original returned within months, because the real thing cannot be new. In 1991 Gerald Ratner stood up and called his own jewellery "total crap." The line went viral before virality had a name. Belief in the brand collapsed, an estimated £500m of value with it, and the Ratners name was retired. One signal, the whole shoal gone.
Most belief change is the slow aggregate. Shoaling is the rare, violent shortcut. Either way, belief moves first and the P&L follows. The investor who reads belief sees the turn while there is still time to act on it.
Oatly listed on the Nasdaq in May 2021 at approximately ten billion US dollars. Within thirty-three months, market cap had fallen below one billion. A ninety percent destruction of value. The financial press blamed supply chain. The investor decks talked margins. The consumer had been telling a different story for the better part of a year before any of it.
BeliefTrak runs the diagnostic continuously. It produces a single headline number per brand per month, the Brand Belief Score, plotted over time. When the score turns down, it does so months before the share price registers the shift. By the time the P&L confirms the trajectory, the belief curve has already been falling.
Here is what the Oatly belief curve looked like, plotted alongside the share price.
Illustrative belief curve constructed from publicly available consumer sentiment. The Brand Belief Score is the BeliefTrak headline output, plotted monthly. The six-month lead time shown here is consistent with the 6 to 18 month range observed across categories.
The belief score broke first. Consumers were already writing about taste regression, price hikes and identity backlash months before the share price registered the shift. An owner running BeliefTrak would have seen the gold line break. They would have had a 6 to 18 month window to intervene. Reformulation. Repricing. New communications addressing the identity wound. Instead, the conversation only started after the share price had already halved.
If belief is the missing diligence layer, then someone needs the discipline to read it. BeliefCo. is the firm that does. Belief Engineering is the methodology.
BeliefCo. was founded by Dan Craddock, who spent more than thirty years inside major consumer brand businesses before naming the methodology. Heinz. Coca-Cola. Guinness. Philips. GSK. Across that career, thirty-four brand transformations completed, every one of them with the same underlying mechanic, though it had no name at the time.
The mechanic is this. Every consumer behaviour is driven by a feeling. Every feeling is produced by a belief. The belief is the meaning the consumer has given to the experience, held as true despite contrary evidence. Shift the belief and the feeling shifts. The feeling shifts and the behaviour shifts. The behaviour shifts and the P&L shifts. That sequence had been working in the background of every one of those thirty-four transformations. The Philips work, in particular, made the mechanic explicit. The product had not changed. The price had not changed. What had changed was what people inside the organisation believed about the brand, and the belief shift produced the commercial shift.
Belief Engineering is the systematic application of that mechanic. Read the consumer belief currently in market. Identify the limiting belief that is suppressing growth. Identify the empowering state the brand needs the consumer to hold. Engineer the five touchpoints, product, price, pack, environment, communications, to shift one belief into the other. Track the belief on a monthly cadence to confirm the shift is taking hold.
The discipline is not new. The thirty-four transformations are the evidence that it has been working for decades. What is new is naming it, codifying it, building the diagnostic instruments that read belief at scale from public consumer signal, and offering it as a service the PE and M&A community can buy in the windows where it matters most.
A working definition, kept tight enough to operationalise. This is the foundation Belief Engineering is built on.
The definition has four parts and each part matters.
Meaning given to an experience. Belief is not the experience itself. Two consumers buying the same product can hold opposite beliefs about it. The product is the input. The belief is the meaning the consumer constructed from it.
Held as true despite contrary evidence. This is the operational test. Beliefs do not yield easily to data. A consumer who believes a brand is cheap and nasty will discount evidence that the new range is premium, until enough cumulative experience overrides the prior belief. This is why advertising alone cannot do the deep work, and also why bad experiences compound faster than good ones.
Generating feelings. Beliefs produce feelings. Pride. Trust. Suspicion. Embarrassment. Excitement. The feelings are the bridge from the cognitive belief to the behavioural action.
That drive the behaviour. Behaviour is the output. Reach for the shelf. Click the link. Recommend to a friend. Quietly stop buying. Behaviour is what shows up in the P&L 6 to 18 months later.
The distinction matters operationally. Belief in is loyalty, affection, identification. Soft. Hard to measure. Easy to fake in survey responses. Belief that is a factual claim held in the consumer's mind. Allbirds wears out in months. Oatly is in every coffee shop now. Toys R Us is dingy. Belief-that statements are observable in consumer language, written down in their own words, on platforms where they have no incentive to perform for a researcher.
BeliefCo. diagnostics measure belief-that statements. They are operational, observable, and they move ahead of behaviour.
Every belief a consumer holds about a brand was built from things they experienced. The product itself. The price they paid. The packaging. The environment they bought it in. The communications they encountered. Every brand action reaches the consumer through one of a limited number of channels. Belief Engineering reads which channel built the belief, and which channel can shift it.
Traditional brand diligence uses three instruments. Focus groups, surveys, and management interviews. Each has a structural reason it cannot see belief.
Focus groups ask eight people to perform a thought process for the moderator. They are useful for hypothesis generation. They are unreliable for diagnosing belief because the act of being asked changes the answer.
Surveys are constrained by the question architecture. The researcher decides in advance which beliefs to test. Beliefs the researcher did not anticipate cannot be detected.
Management interviews are useful for understanding the brand's intent. They are structurally unable to test whether that intent is landing with consumers.
Scraped consumer language is different in three ways that matter.
Volume. A BeliefCo. diagnostic typically classifies 1,000 to 1,800 data points per brand per category. Roughly two orders of magnitude more than a typical qualitative study.
Spontaneity. The consumer was not asked. They wrote what they wrote because they wanted to. The signal is unprompted, which is the strongest form of consumer evidence available.
Specificity. Reviews, forum posts, social comments, video transcripts. The language is concrete, granular, and grounded in actual experience. Belief-that statements appear in the consumer's own words.
Scraped sentiment cannot see what consumers do not write about. It is over-weighted toward consumers willing to publish. It misses non-buyers who never engaged with the brand. It is supplemented in BeliefCo. diagnostics with structured listening on closed platforms and, on Tier 2 and Tier 3 reports, primary research. The diagnostic is stronger when scraped and primary data triangulate. It is not a magic instrument. It is a missing one.
Two dimensions. Four positions. Four investor actions. The diagnostic places every target brand on this map, and the position tells the investor what to do.
Brand Strength captures the structural resilience of the brand. Limiting Belief Intensity captures the size and entrenchment of the consumer belief capping growth. Example brands shown are illustrative positions only.
Buy Now. Strong brand, weak limiting belief. Capital is already compounding. Belief work expands the audience and opens adjacent categories. Buy if the price reflects the runway.
Buy & Transform. Strong brand, strong limiting belief. The highest-return position in the matrix. A single ceiling belief is suppressing an otherwise sound asset. Shift the belief, the whole brand moves.
Build & Hold. Weak brand, weak limiting belief. Founder-led brands, challengers, emerging assets. Architecture first, then belief.
Sell Now / Avoid. Weak brand, strong limiting belief. A live limiting belief is eroding value in real time. Twelve to twenty-four month window before lock-in. If the brand is already in the portfolio, exit. If it is on the buy list, walk away.
Once the dominant limiting beliefs in a category are identified and sized, they plot on a single map. Resistance on one axis. Audience on the other. Revenue opportunity in the bubble size. The map tells the investor which beliefs are worth attacking, in what order, and where the capital should land.
Worked example. Oat Milk Category. The map is read in seconds. Which beliefs are big, which are shiftable, which deserve capital. Bubble names are illustrative.
The Category Belief Map shows the picture. The M&A League turns it into a living competition you subscribe to. It is the only quarterly ranking of consumer brands scored on consumer belief, and it is the foundation the rest of the stack is built on.
Every brand carries a belief score, expressed in Belief Points. Net empowering belief minus net limiting belief. We never publish the recipe, only the number, the way a credit score is read without the formula behind it. Plotted across a market, those numbers form a table. Plotted across four editions a year, they form motion.
Motion is the asset. The League reads the change, not just the state. Who is rising and who is sliding. Who has crossed whom. Promotion and relegation between divisions. A form line on every brand. And a quarterly editorial on what moved and why. A sales-based tracker cannot do this, because it reports a decision already made. The League reports the decision forming.
This is the entry point for a buy-side team. Subscribe to a sector, and you are watching belief value move across a whole market before it reaches anyone's P&L. Everything that follows, the Reports, the diagnostics, the post-deal tracking, hangs off this one living view.
The four stage engine. How Belief Engineering runs end to end. Shown here on the UK oat milk category. A live BeliefLab. Deep-Dive run against publicly available consumer signal. The numbers behind each stage are illustrative of the output structure. Real reports carry full source attribution and signed-off confidence levels.
BeliefLab. Deep-Dive scraped and classified over a thousand consumer quotes from Reddit, Mumsnet, TikTok, Amazon and the UK food and nutrition press. Output. Fourteen distinct limiting beliefs holding the category back. Ranged from price barriers at the top of the prevalence list, through taste regression, UPF panic, blood sugar concerns, seed oil fear, identity resistance ('not real milk'), tea-occasion performance, protein anxiety and several lower-volume beliefs further down.
The three with the largest combined impact on category value. Taste arbitrage. The gap between blind preference for oat milk and actual purchase. UPF defence. The fast-growing belief that oat milk is ultra-processed. Tea occasion. The largely untapped belief barrier among UK tea drinkers.
Each belief sized by prevalence in the data set and crossed against UK household population. The top belief, the price barrier, held by roughly 37% of relevant consumers. Identity resistance, the bottom of the priority pile, held by around 10%. Most striking, the tea occasion. Plant milk uptake in coffee runs above 25% of users. In tea it runs at roughly 4%. Ninety-six percent of UK tea drinkers untapped.
Each belief translated into revenue. Current UK oat milk market roughly £275M annually. Beliefs in the data set are suppressing an estimated £140 to £205M in unrealised category revenue. Investment required to reverse them, in the order of £70 to £105M. Three-year ROI of two to two-and-a-half times.
The single largest, fastest-payback intervention. Taste sampling at national scale. Modelled at £15 to £25M investment to unlock £35 to £50M in annual revenue, with a twelve to eighteen month payback. The most urgent intervention. UPF defence. Clean-label reformulation and independent clinical evidence. Lower ROI but defensive necessity given the growth rate of the underlying belief. The structural opportunity. Cracking tea performance. Daily-frequency occasions across tens of millions of UK households.
BeliefTrak then watches each of the fourteen beliefs on a monthly cadence. The intervention plan has belief milestones. Taste belief moves first if sampling lands. UPF belief is monitored as the early warning on share-stealing reformulators. Tea-occasion belief is watched for the brand that cracks tea first.
The monthly track produces one number per brand per month, the Brand Belief Score, plotted on the curve the previous section showed. The same lead-time logic. Belief moves first. Behaviour follows. The P&L confirms last. The owner running BeliefTrak has 6 to 18 months of warning before the sales line confirms either the recovery or the further decline.
Three points in the deal cycle where the belief read materially changes the decision.
The belief diagnostic runs alongside QofE and CDD during the exclusivity window. Output. A ranked belief map of the target. The dominant limiting belief, sized, costed, and translated into the value creation thesis. The diligence answers the question commercial DD cannot. Not does the demand exist, but does the consumer still believe what we are paying for them to believe.
The diagnostic sets the operating priority for the first 100 days. The limiting belief defines what the business must change at the five touchpoints to release growth. The empowering state defines the destination. The brand value creation plan stops being a marketing plan and becomes a belief plan, with belief metrics, belief milestones and belief accountability inside the operating cadence.
Consumer belief leads sales by 6 to 18 months. A rising BBS confirms a brand is gathering value the next buyer will pay for. A falling BBS warns that the headline financial story may not survive the next twelve months of P&L. The owner who watches BBS can time exit on the belief curve rather than the sales curve, and price the asset on what the next twelve months will look like, not what the last twelve months reported.
One algorithm, The Cipher, scores every brand on a single unit: Belief Points. Net empowering belief minus net limiting belief, a number in the hundreds, plus or minus. Like a credit score, the number is the headline and the method stays proprietary. Everything below reads off that one unit.
Four layers. The whole consumer universe, down to a single buyable brand. One algorithm runs underneath all four.
A table is a five-minute read. Belief in motion is the asset. Each edition reads the change, not just the state: the movers and crossings, promotion and relegation, a form line on every brand, and an editorial on what moved and why. Pure motion. The early warning no sales-based tracker can give.
Read in sequence, the stack does one job for an M&A team: it removes the blind spot that loses money in consumer deals, and it finds the upside a sales line cannot see.
The League shows where belief value is moving across a market, quarter on quarter. The M&A Snapshot opens a single name. The Category Report frames the ground it sits on. The Brand Report diagnoses the target in full. The Deal Report costs the fix and models the value it unlocks across the hold. The IC Report proves it to the committee. After completion, BeliefTrak runs The Cipher continuously, so belief is watched live for the length of the hold.
Less downside. Belief turns 6 to 18 months before the sales line confirms it, so a buyer sees a brand sliding while there is still time to walk, or to pay a lower multiple. The Oatly reconstruction earlier in this paper is exactly that turn, the belief line bending months before the financials moved.
More upside, sooner. The same signal works in reverse. A brand whose belief is climbing ahead of its sales is mispriced in the buyer's favour. Buy it before the P&L catches up, act on the beliefs the Deal Report ranks, and the multiple grows on a clock the seller could not see. Belief tells you what is about to happen. Sales data only tells you what already did.
Send the target. We come back with a free read on the dominant limiting belief and the sized opportunity behind it. No NDA on the first call. No deck. No pitch. Just the read.
Begin the read