For private equity firms with consumer portfolios, the last few years have bit by bit closed off most of the playbook they’ve been used to.
Financial engineering made sense when leverage was cheap and multiples were expanding. Neither of those things is true anymore. Buy-and-build as a growth strategy runs into a consumer who’s increasingly cautious about discretionary spend… in the UK, discretionary spending hit a three-year low in Q1 2026, with nearly half of adults planning to cut non-essential spending. Cost optimisation has been wrung hard enough that the hospitality sector is already running on margins of 3-5%, with 82% of operators having already raised prices and cut staffing hours in response to legislative wage increases and persistent food inflation.
The hold clock is also ticking louder than it has in years. More than 63% of active North American portfolio companies have been held for over four years. LP distributions have roughly halved from their historical average, from around 25% of NAV during 2013-2021 down to just 12% across 2022-2024. The exit that was supposed to happen hasn’t happened, and the pressure to make something of the hold period has nowhere to go except operations.
Which is why 72% of GPs are now citing operational improvements as their primary value creation lever. Not financial engineering, not top-line growth, not acquisition. Operations. And more specifically, the data that tells you what’s working and what’s about to go wrong.
The consumer side isn’t getting easier either
The portfolio companies themselves are being squeezed from the other direction.
Brand loyalty is eroding faster than it has in a decade. 38% of shoppers in 2025 said they were loyal to five or fewer brands, up from 22% in 2023… a dramatic shift in two years. And 60% said they’d switch brands if prices went up. In a market where operators are being forced to raise prices by legislative wage increases, food inflation, and tariff pressure, that’s a direct conflict.
The consumer businesses in a PE portfolio right now are navigating genuinely thin margins, a customer base that’s more promiscuous than it’s ever been, and a macroeconomic environment that’s suppressing discretionary spend across every income group. Organic volume growth isn’t picking up the slack. And acquiring a new customer costs five to 25 times more than retaining one you already have.
Which brings me to the one thing most of these businesses are actually sitting on.
The asset that’s already there
Every consumer business generates it continuously: transaction records, visit frequency, basket composition, recency patterns, loyalty interactions. It accumulates. And in most cases it’s being analysed with tools built for a different era… basic RFM tiers, demographic segments, boolean rules designed around assumptions about what customers look like rather than what they actually do.
The signal is in there. Most of it just hasn’t been mined. And in an environment where every other lever has been squeezed close to its limit, extracting that signal is arguably the highest-ROI thing left on the table.
What deep learning does differently at the portfolio company level
Standard segmentation treats customers as static categories… a “high-value” tier defined by spend above a threshold, a “lapsed” cohort defined by recency. Clean enough to explain in a board update, but by definition built around assumptions that came from a human analyst deciding what to look for in the first place.
Neural networks don’t work that way. They learn latent behavioural patterns directly from the data, surfacing clusters of customers that a human analyst would never have thought to look for, because the patterns don’t map to anything with a simple name. Not “high-spend and high-frequency” but something more like “this group of 4,000 people has a distinctive and consistent way of moving through your product that correlates strongly with lifetime value, and they’re more loyal than their spend tier suggests.” You couldn’t have written that as a WHERE clause.
The precision advantage shows up most clearly at the edges. Every customer in a deep learning-derived cluster comes with a distance-to-centroid… a measure of how archetypal they are within their group. The customer at the centre of the cluster is the cleanest expression of that behavioural pattern. The customer at the edge is still technically in the segment, but they’re already drifting. That drift signal is where churn prediction gets genuinely useful, not as a retrospective label applied after someone’s already left, but as an early indicator that intervention has value.
When you’re operating on 3-5% margins and every retained customer is expensive to replace, the difference between broad retention spend and targeted intervention matters enormously. A 5% improvement in retention can lift profits by 25-95%. The maths hits differently when the baseline margin is that thin.
The cross-portfolio view
This is the angle that’s harder to see from inside any single portfolio company, but it’s the one that should interest operating partners most.
A PE firm with eight or ten consumer brands across restaurants, retail, and hospitality is sitting on a cross-portfolio data asset that no individual company management team can access on their own. When you run consistent deep learning segmentation across the portfolio, patterns start to surface that are invisible at the individual company level.
Which customer behavioural segments are shared across multiple brands? If a cluster at one portfolio company looks structurally similar to a cluster at another, that’s either a cross-sell opportunity or a signal that the same retention playbook that worked in one place is worth testing in another. Which portfolio companies have the cleanest, most loyal underlying customer bases ahead of exit, and which need work before they’ll command a premium multiple?
That last question is important. Acquirers right now are specifically awarding premium valuations to businesses that can demonstrate strong customer retention and a coherent customer base. Walking into an exit process with cross-portfolio benchmarks on retention quality, churn trajectory, and lifetime value distribution isn’t just a stronger narrative… it’s a demonstrably differentiated position. Acquirers are paying up for exactly this right now. The businesses that can show it clearly, derived from deep learning rather than analyst intuition, will outperform comparable assets at exit.
This isn’t new technology
None of this is novel, by the way. Google, Meta, and Amazon all made deep learning the underlying intelligence behind their recommendation and ad systems roughly a decade ago. The results are visible in where media budgets have gone ever since… these platforms have captured an ever-increasing share of global advertising spend precisely because their models understand customer behaviour at a level of granularity that no human analyst, and no rules-based system, can match. That shift wasn’t accidental. It was the direct consequence of applying neural networks to first-party behavioural data at scale.
The imperative for consumer businesses is to do the same thing, but on their own data. Transaction histories, loyalty interactions, visit patterns… the raw material is there. What’s been missing is the tooling and the institutional weight to do something genuinely sophisticated with it, which is a large part of why this has remained a big-tech privilege for so long.
The catch is that most portfolio companies can’t bridge that gap on their own. They’re not Meta. They don’t have the engineering headcount, the infrastructure, or the data science teams to build this from scratch, and they can’t justify the investment individually. Which is precisely where the private equity owner has a structural role to play… not just as capital provider, but as the enabling layer that gives individual portfolio companies access to capabilities none of them could build alone.
What this looks like in practice
At Neuralift this is what we do for consumer businesses. Transaction data, event streams, loyalty interactions, and purchase sequences go in through a properly structured input layer. Roughly six to seven deep learning models run iteratively until the latent space converges, dropping every customer into the segment they actually belong in based on their real behaviour. The output is a set of stable, actionable segment IDs, each with a distance-to-centroid that lets you tune precision depending on what you’re trying to do with the segment.
For a single portfolio company that might mean identifying the 6,000 customers whose behaviour puts them at genuine churn risk in the next 90 days, and separating them from the 3,000 whose apparent inactivity is seasonal rather than structural. For a portfolio operations team it might mean running consistent segmentation across multiple brands and building a shared intelligence layer that nobody in the portfolio has had access to before.
This isn’t theoretical positioning either. We’ve already started seeing PE firms introduce us directly into their portfolio companies… warm intros from a partner who’s seen the model land in one place and wants the same lens applied across others. I expect a lot more of that pattern over the next 12 to 18 months as the operational-improvement imperative bites harder. The funds that move first will be working with a meaningfully better data picture than the ones that wait.
I’ve written about this before in the context of lookalike seed quality, and the underlying argument is the same in the PE context. The data is already there. The urgent need is to get more granular with it… and that’s not something you can do with human-only approaches in businesses that are already resource-constrained. The analyst time isn’t there, the headcount isn’t there, and the pattern complexity is beyond what rules-based segmentation can realistically surface anyway.
For a lot of PE consumer portfolios right now, that’s the only lever that hasn’t been pulled.