Why the next 12 months will be the hardest to forecast in food and drink
By Helen Blair, Group Insight Director, F!S Group
Planning teams that go into the next 12 months with a clear view of what has actually shifted will make sharper calls than those still running last year's assumptions.
Forecasts have always carried error. Anyone who has worked in food and drink long enough has seen a launch case overshoot by 30% and another miss by half. That part is normal. What is not normal is what is happening now: three of the foundational inputs that planning models depend on are all moving in the same 12-month window, and most of the assumptions sitting underneath last year’s spreadsheets have not been re-tested.
I sit in client meetings most weeks where the model gives an answer it would have given two years ago, and the senior team in the room has stopped quite trusting it. That gut response is right. The model is not faulty. The inputs are.
The macro context is hardly helping. The UK Food and Drink Federation has revised its food inflation forecast from 3% to 9% by year-end following the latest disruption to the Strait of Hormuz, and the FAO Food Price Index rose 2.4% in March alone (FoodNavigator, April 2026). Stagflation is back in real-world conversations rather than economist textbooks. That is the backdrop. What follows are the three inputs that change how planning teams should be working inside it.
This piece is the longer version of a LinkedIn post I shared this week. It walks through the three inputs that have moved, why each one matters, and the practical steps insight teams can take in the next quarter to stop relying on assumptions that have already expired.
1. Volume per buyer is moving, and most businesses are slow to catch up
The first and most discussed shift is the impact of GLP-1 medications. Tirzepatide and semaglutide prescribing has scaled rapidly in the UK since the back end of 2024, and even with conservative estimates, the number of people now using these medications privately or through the NHS is large enough to start showing up in the categories most affected: confectionery, salty snacks, ready-to-drink alcohol, sugary soft drinks, and impulse bakery.
The visible edge of it is already in the trade press. NielsenIQ data picked up by FoodNavigator at the end of April found that mainstream chocolate sales fell in units in 2025, while super-premium chocolate value sales grew 16.7% and premium 9.8%. The story FoodNavigator told from the data was straightforward: healthier choices and GLP-1s are driving consumers away from chocolate consumption, but the super-premium category seems immune to the trend. New Nutrition Business has been tracking the same compound effect from the other angle, with the GLP-1 effect now boosting meat and high-protein sales (NNB, May 2026). What the data means in plain language: shoppers are not abandoning these categories. They are changing what they buy inside them.
What it does not do, yet, is show up cleanly in the headline numbers most planning teams rely on. The reason is the nature of the shift. The effect is compositional, not absolute: most GLP-1 users do not stop buying these categories, they buy less, less often, and in different formats. A change of that kind is quiet in any top-line read. It does not announce itself as a volume collapse. It surfaces slowly, as a gradual reshaping of what sits in the basket, and gradual reshaping is exactly the kind of signal that headline data takes time to confirm.
In our own work, we are seeing it cleanly in qualitative depth and in the custom panels we run for clients. The signal is there. The challenge for insight teams is to triangulate it in their own current data rather than wait for it to surface in the headline numbers, because by the time it does, planning cycles will already have set the range and the marketing budget for the year.
The practical action: if your category sits in the affected set, run a piece of work this quarter that captures GLP-1 status as a respondent variable. Even a small base will show you whether your shopper mix is starting to shift, and where.
2. Price elasticity is no longer a stable curve
The second input is more structural and, in some ways, harder to deal with. Three years of compounded price rises have changed the way shoppers respond to further price changes, and the change is not uniform.
The early signal is starting to show in the financials of the largest players. PepsiCo has reduced prices on certain snack lines by roughly 10 to 15% following weaker volumes in North America. Mondelez International has reported volume declines of around 2 to 3% in parts of its portfolio (FoodNavigator, April 2026). FoodNavigator’s stagflation analysis put it bluntly: food manufacturers are losing pricing power as consumers trade down, and the usual levers no longer align. Raise prices and lose volume. Hold back and compress margins. The teams who rely on three-year elasticity curves are reading a curve that no longer matches the consumer in front of them.
The same +5% on a branded SKU now triggers a different switching pattern depending on the income bracket of the shopper, the age of the household, and the day of the week. Younger shoppers under cost-of-living pressure have already moved to private label or own-brand on staples and are unlikely to pay back when prices ease. Older shoppers with more discretionary income have absorbed the increases without much switching, but their tolerance for further rises is uneven and category-specific.
The old elasticity curves were built on three relatively stable years of consumer behaviour. The new curves are being drawn in real time, and they have much shorter usable shelf lives. Elasticity work done in early 2025 has limited predictive value by mid-2026.
This matters for two reasons. The first is that the price-volume trade-offs in your innovation business case are likely wrong. The second is that the retailer negotiation arguments built on 2023 elasticity are going to be challenged, and the retailers will be doing more recent work than the brand teams will.
The practical action: do not rely on elasticity curves more than 12 months old for any decision worth more than the cost of refreshing them. If your category has not been re-tested since the last big price round, it is overdue.
3. Channel mix has settled at a different equilibrium
The third input is the one most planning models still treat as if it is mid-cycle. It is not. The channel mix that food and drink sits inside has reached a new equilibrium that looks different from anything we modelled five years ago, and most of the shifts are not reverting.
Discounter share has plateaued at higher levels than nearly every planning model written in 2022 expected, and the shoppers who moved have stayed. Convenience is up structurally, partly because of work-from-home patterns settling and partly because in-flight consumption is now the bulk of impulse. Online grocery has stopped growing in the way that the 2020-2021 acceleration suggested, and the platforms are restructuring around that reality.
The supply side is moving in step. UK farmgate milk prices have fallen by up to 40% between October 2025 and early 2026 due to oversupply, while processors continue to face higher operating costs (FoodNavigator, April 2026). That gap between farm and processor is hard to close, and is one of several places where channel and supply assumptions written in 2022 no longer reflect the market in front of them.
Each of these on their own would be manageable. The compound effect, though, is that the weighted distribution targets and channel-by-channel volume forecasts in most plans are sitting on assumptions that no longer reflect where shoppers are actually buying. That has direct consequences for trade investment, supply forecasts, and the case made internally for branded versus own-label development.
The practical action: rebuild your channel weights using the most current 12-month moving average rather than three-year norms, and stress-test your distribution plan against a flat-discounter, growing-convenience, slow-online scenario. If the strategy still works, you have something to defend. If it does not, you need to know now, not in October.
What changes for the way insight teams work
The honest answer is that all three of these problems are solvable with current research methods. None of them require new technology. What they require is the willingness to put time and budget into refreshing assumptions that most teams have been carrying forward unchallenged because the team that owns them is busy with launches and reviews.
A short list of what we are doing with our clients this quarter:
· Auditing the planning assumptions that are still using pre-2024 inputs and prioritising the three that, if wrong, would change the most decisions
· Running quick, current-data elasticity work on the categories where pricing rounds are imminent
· Re-segmenting shopper bases by GLP-1 status, by income tier, and by recent switching behaviour, rather than by the demographic cuts that worked when the consumer was more uniform
· Building scenarios rather than point forecasts, so leadership teams have a clear range and a sense of which inputs they are most exposed to
· Bringing qual and quant into the same room earlier, because qualitative depth often surfaces an emerging signal before it is visible at scale
None of this is glamorous work. It is the work that protects the strategy that follows it. The cost of getting any one of these inputs wrong has gone up sharply, because each of them now compounds with the others. That is what makes the next 12 months different.
Where to start
The single most useful thing an insight director can do this month is convene a 60-minute session with the planning team, the innovation team and the commercial lead, and put one question on the table: which three things about consumer behaviour, price response or channel mix are we still treating as fixed in our 12-month plan? Then ask what would change if they were not.
That conversation will tell you whether you have a forecasting problem or a strategy problem. Most teams will find they have both.
Get in Touch
Talk to us about your forecasting assumptions.
If your team is rebuilding planning assumptions for the next 12 months and you would value a sounding board, drop us a note. We have run this exercise with several food and drink leadership teams over the last quarter and would be happy to share what we have learned.
Helen Blair is Group Insight Director at Good Sense Research, leading consumer insight programmes for senior teams in branded and own-label food and drink, retail and foodservice.