2026 State of the UK Construction Industry

Total UK construction output recovered after the 2020 shock, but the headline number is a poor planning tool. ONS output data (chained volume, seasonally adjusted) shows a sharp rebound in 2021, then a long plateau through 2022 to 2024. The Construction Products Association (CPA) Winter 2025/26 forecast puts 2026 growth at about 2.8%, which takes the total output index to roughly 112 (2020=100). That can still mean flat turnover for your firm, because the work mix is moving under your feet.

The split that matters is simple. Infrastructure is rising while private housing is falling. On the same 2020=100 index, infrastructure is forecast around 128 by 2026, and private housing around 94. That is why two firms can both say “the market is steady” and still face opposite tender conditions. If you are weighted to housing and housing-led supply chains, you are fighting for volume in a shrinking pool. If you are weighted to utilities and civils, you are fighting for labour, access windows, and approvals on complex programmes.

Housing starts are the early warning signal that hits your pipeline first. DLUHC housing supply tables show private housing starts in 2024 ran roughly 18% below their 2022 level. That drop feeds through into groundworks and RC frames first, then MEP and fit-out later. Output can still look “fine” at national level because other sectors fill the gap. Your business still takes a hit if the replacement work has longer payment cycles, higher prelims, or heavier testing and commissioning.

Repair and maintenance (R&M) has grown as a share of output, to about 38% in ONS data. Many teams treat R&M as safer because jobs are smaller and repeatable. Commercially, it bites in a different way. R&M generates more instructions and small changes, so you get more chances to miss notice periods under JCT or NEC contracts. That is margin leakage, not a site performance issue. Weekly cost-to-complete and a weekly variation review stops the slow bleed, because you catch it while the labour and plant decisions are still changeable.

Practical takeaway for 2026: plan around mix, not the headline. Split your live work and pipeline into private housing, infrastructure, and R&M. Set different cash rules, reporting cadence, and contract admin checks for each. Archdesk customers do this with consistent project tagging and cost codes, so the mix shift shows up in valuations, prelim burn, and variation run rate before it hits margin.

Material prices stopped driving the story. Labour still does. CPA forecasts through 2025 into early 2026 flagged cooling product inflation, but “cooling” does not mean “back to 2020”. Your cost base has reset at a higher level. Then wages keep climbing on top. Flat supplier quotes do not fix margin if the job runs long and you are paying higher weekly burn the whole time.

The useful number to watch in 2026 is the spread between materials and wages, not either line on its own. If materials settle and earnings keep rising, labour-heavy trades become the inflation engine for the whole job. That is the moment you need more estimating discipline on hours and output, not on the materials basket. Fit-out, drylining, ceilings, MEP first fix and second fix, and testing and commissioning are the packages that keep biting even on “calm” material markets.

Prelims is where wage pressure hides because it shows up as time, not as a line item on measured works. BCIS tender price reporting in late 2025 showed prelims allowances in winning bids rising by 6% to 8% year on year, even where measured works rates were not moving much. That single stat should change how you think about delay. The same two-week slip now costs more cash because the weekly burn for site management, welfare, logistics, and temporary works is higher.

Fixed-price change is where the costs vs wages gap turns into a real loss. Under JCT and NEC, variations are often priced using contract rates set at tender. Those rates can be 12 to 18 months old by the time the instruction lands on site. Labour-heavy packages feel it first. RICS Contracts in Use surveys show fewer than 30% of sub-£20m projects include fluctuation clauses. Most firms are carrying wage drift inside “fixed” rates, then arguing about it after the work is done.

Practical takeaway: split your allowances and manage labour like measured work. Track planned versus actual hours and output weekly by trade, then link that to progress and variations. Archdesk is built for this. It ties labour hours, programme status, and commercial records together, so you can price change from live site evidence and protect margin before the job drifts.

Labour shortage hits your margin through programme damage first, wage rates second. One missing gang forces resequencing and stacked trades. That leads to rework, missed inspections, and late commissioning. The real cost sits in prelims extension and disruption, not the day rate premium.

Capacity pressure is structural. CITB workforce modelling (2025) put the extra headcount needed at about 251,500 workers by 2028. CITB also flagged retirements at roughly 36,000 a year against about 27,000 new entrants in 2024/25. That gap does not land evenly across trades. It lands at hold points, where one missed week turns into three.

Critical trades in 2026 are the ones that gate close-up and sign-off. MEP first fix is the common bottleneck on fit-out and MEP-heavy builds. If rough-in slips, drylining cannot close, firestopping cannot sign off compartments, and ceilings wait. Test and commission then gets crushed into the last few weeks, when you have no float left. This is where “one missing week becomes three” is real, because follow-on gangs get redeployed to other sites.

Prelims is where labour pressure hides in plain sight. BCIS tender price reporting in late 2025 showed prelims allowances in winning bids rising by 6% to 8% year on year, even where measured works rates were not moving much. That number matters because prelims is where you pay for disruption. If you accept a programme that cannot be resourced, prelims becomes a silent contingency. It gets burned before you reach the critical finishes.

Contract form changes how early you see the problem. Fixed-price JCT Design and Build pushes risk down the chain, so you get shorter tender validity, more exclusions on access and sequencing, and start dates that are “subject to availability”. NEC target cost (open-book pain/gain) flushes the issue out earlier. Labour assumptions and output rates get tested before the baseline programme is treated as fixed. Archdesk teams see the same pattern across projects. Weekly resource-loading against named gangs catches slippage early enough to change sequence, not just argue about it later.

Overtime is not a recovery plan. It is a defect risk you pay for twice. CIOB’s 2024 Quality in Construction report found that projects where more than 20% of site hours were overtime saw defect rates 35% higher than projects at or below normal hours. That cost lands at handover and through the Defects Liability Period, the period after handover when the contractor fixes snags. The commercial hit with clients and main contractors often lasts longer than the snagging.

Two markets are running side by side, and they behave differently under pressure. England private housing starts fell to roughly 132,000 in 2023/24, down from about 177,000 in 2021/22, according to DLUHC live tables. The IPA’s National Infrastructure and Construction Pipeline (late 2025) puts active projects at over £700bn. Planning labour, cash, and overhead off a single “construction market” number hides where your real risk sits.

The difference is funding and control, not just volume. Regulated programmes in water and energy move on multi-year cycles with defined client teams and planned procurement. That reduces stop-start risk, but it raises the bar on audit trail, evidence, and reporting. Private housing moves with sales rates and debt, so work can vanish fast, even after you have priced it, resourced it, and booked materials.

Contracting risk changes with the workmix. Infrastructure is more likely to come through frameworks and NEC-style (New Engineering Contract) management, where early warnings and compensation events live or die on records. Housing-led work is more often fixed price, where margin depends on continuity and clean access. The same trade gang can look “expensive” on housing after two stop-start weeks, and look competitive on infrastructure with a steady front and fewer remobilisations.

Cash exposure also splits. Regulated clients tend to run predictable valuation cycles, but certification can be slow and admin heavy. Housing developers can be quick when sites are flying, then stretch terms or pause works when sales dip. That creates a hidden working-capital swing for subcontractors that straddle both. If you do not price for that swing, you fund it from overdraft and margin.

Practical takeaway: split your forward order book into two work banks, regulated infrastructure and private housing-led. Set different tender assumptions for payment terms, stoppage risk, and reporting workload. Tag live projects the same way in Archdesk, then review margin and cash by work bank each month so a strong run on one side does not hide a leak on the other.

Regulated and compliance-led programmes are the safest workload in 2026 because the client has to spend. They do not spend because they feel confident. Water and energy sit at the top of that list. Ofwat’s PR24 settlement has set funding for AMP8, so the fight is now about access windows, outage planning, and delivery capacity. According to the IPA National Infrastructure and Construction Pipeline published in late 2025, active projects sit at over £700bn. That scale does not remove risk. It shifts it from “will it start?” to “can we staff it and prove it?”

Private new-build housing stays the weak link because starts are discretionary. DLUHC data shows England private starts fell to roughly 132,000 in 2023/24, down from about 177,000 in 2021/22. That is around a 25% drop in two years. The commercial impact is predictable. Rate pressure goes up, payment terms get tighter, and more risk gets pushed into sub-contract amendments. If your order book is still led by private sale work, you will spend more time negotiating exclusions and chasing cash than building margin.

Regulated infrastructure has a different profit model to “measured works at speed”. Energy, water and large transport programmes are heavy on possessions, temporary works, permits, assurance, and handover evidence. That cost sits in prelims, supervision, planning and QA. It does not sit in unit rates. The firms that hold margin price access and assurance as deliverables. They show their resourcing plan up front and they protect it contractually. The firms that treat it like a rates race win work and then bleed on overtime, night shifts, re-sequencing and documentation.

Commercial refurbishment is a winner for a different reason. The driver is compliance and asset value. From April 2027, leased commercial buildings need a minimum EPC rating of C. That pulls work forward into 2026 because landlords cannot leave it late and still keep tenants. Fit-out and MEP firms make money here when they run the job like a controlled shutdown. They plan access by floor and by tenant. They agree evidence requirements before they start. They also price the cost of working around an occupied building, not the cost of working in an empty one.

Industrial and logistics is not dead, but the work has changed shape. Cold storage and battery storage push complexity into power distribution, controls, fire strategy, and commissioning. That is strong for MEP, structural, and envelope specialists who can take design risk and deliver against a test plan. The trap is scope drift through “value engineering” after you mobilise. Lock design freeze dates and interface responsibility early, or you will pay for rework and disruption that you cannot claim back.

Action for 2026: sort your pipeline by funding type, then price and resource it differently. Put your best planners, QSs and commissioning leads on regulated and compliance bids because certainty is what the client buys. Treat private new-build volume as optional. Control the start date, tighten quote validity, and price programme risk as a real cost. Archdesk teams see margin hold when projects track cost to complete weekly and keep interface scope tight, before small delays turn into unpriced disruption.

London is not a higher-cost version of the regions. It is a different delivery environment. The same trade rate can look fine on paper and still lose money because the real burn sits in access, deliveries, supervision, and resequencing. One UK average creates a double error. You underprice London prelims and disruption. You overprice regional work and lose on price.

ONS ASHE pay data shows London construction pay sits above the UK average for comparable roles. The bigger issue is not the rate. It is the non-productive time that comes with central sites. Travel, inductions, delivery booking, marshalling, security, permits, and more supervisor layers all sit outside measured work. If you do not price those items as prelims, they still happen. They just arrive as margin loss.

Output rates split just as hard. London has more refurbishment, Cat A and Cat B fit-out, and MEP-heavy work in live buildings. Access windows drive the sequence. Outside the M25 you get more repeatable workfaces in industrial, civils, and utilities programmes. Same gang and same nominal rate, but different metres per day. If your estimating team carries one output norm across both markets, you price the wrong programme. The programme then drives the wrong prelims and the wrong cash curve.

Client behaviour also changes the commercial risk. London residential has a bigger build-to-rent and student mix than most regions. Those clients focus on building performance and handover evidence. That means heavier testing and commissioning, higher O&M expectations, and more return visits in the Defects Liability Period, the period after handover when the contractor fixes snags. Regional private-sale work reacts faster to mortgage demand. Pressure shows up in price, payment terms, and phased handovers. Change control and cash assumptions that work regionally will not protect you in London, and the reverse is also true.

Practical move for 2026: run two playbooks. Split cost codes, output norms, and prelim templates into “London” and “rest of UK”. Report margin that way too, not only by sector. Archdesk teams who do this see the problem earlier, because the dashboard shows prelim burn versus progress before the job is in trouble. Keep your London method statement priced like a cost plan. Keep your regional bids sharp on measured-work outputs and supply chain rates.

Tender risk in 2026 sits in programme reality, not in your measured works rates. Lead times for key building components still drive start dates and sequencing. A 2025 industry benchmark from BCIS put typical lead times at 14 to 18 weeks for structural steel and curtain walling. If your tender programme assumes 10 weeks and you do not qualify it, you carry the acceleration and out-of-sequence cost. State assumed lead times for your top ten long-lead items in the tender return. Tie them to the order dates you need from the client.

Prelims is the margin trap most bids still miss. RICS Contracts in Use 2024 found prelims often land at 12% to 16% on £5m to £20m projects. Many complex fit-out and structural packages now push closer to 18% once you price supervision, logistics, welfare, security, and temporary works properly. Prelims tracks wages and access constraints. It does not track your BoQ. Price it as a weekly cost with a clear duration. Stress-test it against access windows, shift patterns, and the real supervision load across interfaces.

Contract strategy needs to be picked at tender, because JCT and NEC punish different mistakes. Under JCT Design and Build 2024, if you do not include a fluctuation route, you carry the full cost movement. Under NEC4 Option C, you will get challenged on output rates and the duration you assumed. Either way, the commercial win comes from being specific. Name the trades that drive your critical path. State the output assumptions and working hours you priced. Clients argue less with an hours story than a rate story, because they can see the programme link.

Evidence is a tender decision, not a month ten scramble. NEC compensation events live or die on timescales and records. The common failure mode is simple: teams spot changes, but notify late and cannot tie cost to cause. Build the evidence plan into your tender clarifications. Commit to daily diaries, labour allocation sheets, and time-stamped photos for access constraints and stoppages. Archdesk helps here because it links programme, costs, and change logs, so your QS is not rebuilding the story from emails.

Practical move for your next tender review. Make the estimator show four separate lines before you sign it off: time-related prelims per week, labour hours by trade on the critical path, top ten lead-time assumptions with required order dates, and the record set you will keep to prove change. If any line cannot be explained in two sentences, it is not priced. It is hidden risk, and hidden risk is the part you never get paid for.

Good operators don’t win by calling the market perfectly. They win by moving first when the data changes. Most contractors still take four to eight weeks to turn an external signal into a bid decision, a buy order, or a resourcing move. That delay is pure margin leakage. Set up a small set of signals with a clear trigger for each, then act inside days, not weeks.

Use ONS new orders as your forward workload warning, not ONS output. New orders arrives earlier in the cycle. It tells you what will hit sites later. Track it by the sectors you actually sell into, not as a single national number. If it drops for two quarters in a row in your target sector, stop “keeping busy” with low-quality tenders. Cut the bid list, push negotiated routes, and protect estimating hours for work you can win at the right terms.

Treat planning approvals as a separate signal from site starts. The gap between approval and start is where work dies quietly. Draft boards miss this and assume a strong pipeline means strong work in hand. Watch approvals versus starts in your patch. If approvals rise but starts fall, clients are hesitating. That’s when you tighten bid conditions, shorten price validity, and ask harder questions about funding and decision dates before you commit pre-construction effort.

Split cost signals by what you can control. BCIS resource indices are useful because they break out labour, plant, and material groups. Don’t just watch “materials inflation”. Watch divergence. Draft bid reviews in early 2026 showed specialist M&E components rising while general building materials were flat. That’s how you end up buying your M&E kit at the worst moment. If the index for a critical category rises for three months, move that package forward. Place orders earlier, get quotes with clear expiry dates, and build a procurement plan that matches lead times, not the tender programme.

Treat HSE enforcement as a real operational signal, not a compliance footnote. HSE published over 8,200 enforcement notices in construction during 2023/24, with a rising share linked to work at height and structural stability. Higher enforcement in a region often sits alongside rushed programmes, thin supervision, and congested sites. That hits productivity before it hits your cost report. If enforcement activity rises where you work, assume more lost time and more rework. Price supervision and access properly in prelims. Don’t accept impossible windows and then hope to recover on measured works.

Practical step for 2027: assign one owner for the signal dashboard, usually the commercial director or head of pre-construction. Review it on a fixed monthly cycle. Every review must produce three outputs. Decide which tenders you will not bid. List the packages you will buy early if indices keep rising. Confirm the next quarter’s resource plan by trade and supervision ratio. Archdesk clients do this by putting external signals beside live job cost and programme data, so actions land fast and stay consistent across the business.