Pricing and Rate Libraries

UK pricing sources — what each provides

BCIS (Building Cost Information Service). Operated by RICS, BCIS is the most widely used construction cost database in the UK. It provides:

• Elemental cost analyses of completed projects — real costs broken down by element
• Tender price indices and building cost indices — tracking market movement
• Rates for measured work — unit rates for NRM 2 items
• Regional adjustment factors — indices for adjusting costs between regions
• Preliminary percentage ranges by project type and value

BCIS is invaluable for benchmarking and early-stage cost planning under NRM 1. For detailed pricing, it provides a credible baseline that can be adjusted for your specific circumstances.

Spon's Architects' and Builders' Price Book. Published annually, this is the traditional reference for construction pricing. It provides all-in rates for a wide range of building work items, approximate estimating rates, and guidance on preliminaries and daywork.

Laxton's Building Price Book. Similar to Spon's but with a different editorial approach. Some practitioners prefer Laxton's for its treatment of certain work sections.

Internal tender data. Your own database of priced BOQs, subcontractor quotations, and final account records. This is the most accurate source for your specific market position but requires systematic collection and maintenance.

Subcontractor quotations. For specialist work (mechanical services, electrical installations, lift installations, curtain walling), the rate library should hold both benchmark rates (for early estimates) and recent quotation data. The gap between benchmark and actual quotation rates can be significant in a volatile market.

The anatomy of an all-in rate

Every rate in a BOQ is a composite. Understanding the components is essential for both building rate libraries and configuring AI to match rates intelligently.

Labour. The cost of the operative performing the work. This includes:

• Basic hourly or daily rate (governed by CIJC or BATJIC working rule agreements for directly employed labour)
• National Insurance, pension, CITB levy, holiday pay, sick pay
• Travel and subsistence allowances
• Supervision (typically expressed as a percentage — one foreman per 10-15 operatives)
• Labour productivity assumptions (output rates — m2 of plastering per operative per day)

Materials. The cost of materials delivered to site:

• Material purchase price (from merchants or manufacturers)
• Delivery charges
• Waste allowance (typically 5-10% for most materials, higher for cutting-intensive items like tiles or brickwork)
• Unloading and site distribution

Plant. The cost of equipment and tools:

• For most finishing trades, plant is minimal (hand tools, access equipment)
• For groundworks and structural work, plant costs can dominate (excavators, cranes, concrete pumps)
• Plant rates are typically daily or weekly hire rates allocated across the output achieved

Overheads. The cost of running the business that is not directly attributable to a specific project:

• Head office costs, insurance, IT, professional fees
• Typically expressed as a percentage of turnover (5-12% for most contractors)

Profit. The margin the contractor needs on the work:

• Typically 2-5% net for main contractors (yes, it really is that thin)
• Higher for specialist subcontractors in niche markets
• Variable depending on market conditions, risk appetite, and desire to win the work

For example, a rate for 100mm dense concrete blockwork in mortar, fair face one side, might build up as:

AI-assisted pricing — semantic rate matching

Once quantities are measured, each BOQ item needs a rate. This is where AI can assist — not by generating rates from nothing, but by matching BOQ item descriptions to the most appropriate rate in your library.

The matching problem. A BOQ item description reads: "Walls; 100mm thick; Tarmac Toplite Standard Block; stretcher bond; 1:1:6 cement:lime:sand mortar; fair face one side." Your rate library has entries for:

• "100mm blockwork, standard block, fair face" — GBP 52.40/m2
• "100mm blockwork, dense block, fair face" — GBP 55.60/m2
• "100mm blockwork, standard block, two coats plaster" — GBP 68.20/m2
• "140mm blockwork, standard block, fair face" — GBP 58.80/m2

A semantic matching AI can read the BOQ description, identify the key attributes (100mm thickness, standard block, stretcher bond, fair face one side), and match to the most appropriate library rate. In this case, it should match to the first entry — 100mm standard block, fair face.

The "rate confidence" score. The AI should output a confidence score with each match:

• 0.95+ — Strong match: the library rate matches all key attributes of the BOQ item
• 0.80-0.95 — Good match: most attributes match, but one or two differ (e.g., mortar specification not in the library rate)
• 0.60-0.80 — Approximate match: the library has a similar item but with meaningful differences. QS review required.
• Below 0.60 — No reliable match: the item is not represented in the library. New rate build-up needed.

This confidence score is critical because it directs the estimator's attention. High-confidence matches can be accepted with a quick glance. Low-confidence matches need manual rate building or subcontractor quotation.

Handling provisional sums, prime cost sums, and daywork

Not everything in a BOQ is measured and priced with all-in rates. Some items require different pricing mechanisms, and AI must handle these correctly.

Provisional sums. An allowance included in the BOQ for work that cannot be fully described or measured at tender stage. Under JCT contracts, there are two types:

• Defined provisional sums — the work is sufficiently described for the contractor to have made due allowance for planning, programming, and pricing preliminaries. The contractor is deemed to have allowed for these elements.
• Undefined provisional sums — the work is not sufficiently described. The contractor is not deemed to have allowed for related preliminaries, and an adjustment is made when the work is instructed.

AI cannot generate provisional sums — they require professional judgment about what work is insufficiently defined. AI should flag items where the specification is incomplete or ambiguous and suggest that a provisional sum may be appropriate.

Prime cost sums. An allowance for work to be carried out by a nominated subcontractor or for materials to be supplied by a nominated supplier. The contractor adds for attendance, profit, and any special attendance requirements. These are less common under modern procurement but still appear in some BOQs.

Daywork. A provision for work that cannot be valued by measurement — typically minor or unforeseen work instructed during construction. Daywork rates for labour, materials, and plant are included in the BOQ, usually priced as percentage additions to the RICS Schedule of Basic Plant Charges and relevant working rule agreement rates.

AI should recognise these pricing mechanisms in historical BOQs and handle them appropriately in new ones — including them in the BOQ structure without attempting to convert them into measured rates.

Updating rates — the maintenance challenge

A rate library is only valuable if it is current. Construction costs move with material prices, labour rates, and market conditions. A rate library that is 18 months out of date can be 10-20% wrong.

Material price volatility. Timber, steel, concrete, copper, and insulation materials have all experienced significant price movements in recent years. Your rate library must track these movements and update material components accordingly.

Labour rate changes. CIJC (Construction Industry Joint Council) and similar bodies publish annual wage awards. These feed directly into the labour component of every rate. A 5% wage increase translates to a 2-3% increase in all-in rates for labour-intensive items.

AI-assisted rate updating. AI can help maintain the library by:

• Monitoring published cost indices (BCIS, ONS construction output price indices) and flagging when library rates have drifted beyond a threshold
• Parsing subcontractor quotations from recent tenders and comparing against library rates
• Identifying items where the library rate is significantly above or below recent competitive rates
• Suggesting rate updates based on index movements, subject to QS approval

The discipline required. Rate library maintenance is unglamorous but critical work. The most effective practices update their libraries quarterly, incorporating data from completed tenders, final accounts, and published indices. AI can reduce the effort, but the QS must own the decisions — accepting or rejecting each rate update based on their market knowledge.

Key takeaways

1. A rate library is a core commercial asset — built from your own project data, benchmarked against BCIS and published sources, and adjusted for region and time.
2. Every all-in rate comprises labour, materials, plant, overheads, and profit — each component varies independently with market conditions.
3. AI assists pricing through semantic rate matching — comparing BOQ item descriptions against library entries and providing confidence scores.
4. Confidence scores direct review effort — high-confidence matches need a quick check, low-confidence matches need manual rate building or subcontractor quotations.
5. Provisional sums, PC sums, and daywork are pricing mechanisms that require professional judgment — AI should handle them structurally but not attempt to generate them.
6. Rate libraries must be maintained — quarterly updates incorporating tender data, final accounts, and published indices.

Next up: AI for Site Inspection.