Pillar guide

MEP Contractor Marketing and AI: How to Win More Bids Without Another Estimator

How a 10–49 person MEP contractor can use AI across the bid cycle: quote tables, clearer proposals and steady follow-up, with the estimator still in charge.

Written by the Miriad team with AI assistance. Every figure links to its source, and examples marked illustrative are not real cases. Updated 2026-10-21.

TL;DR: An MEP firm rarely loses a bid at the marketing stage. It loses it between the bid invitation and the award: quotes that arrive late, exclusions nobody wrote down, a proposal followed by silence. AI is useful in exactly that gap, for reading, sorting and drafting. The estimator keeps every price, quantity and labour figure. Start with one live bid, one comparison table and one follow-up rule.

It is Thursday afternoon. A general contractor's bid invitation for the mechanical and plumbing package on a school refurbishment lands in the shared inbox. The drawings sit on a portal, the bid is due next Wednesday, and your senior estimator is already pricing two other jobs. Three wholesalers will answer in three formats, one of them as a phone photo of a printed page. Somebody has to read the commissioning section of the spec and decide whether you want this job at all.

Most marketing advice for contractors starts with more leads. For a mechanical, plumbing or electrical contractor with somewhere between ten and fifty people, that is usually the wrong end. The invitations arrive. What runs out is estimating time. This guide follows the bid cycle step by step, shows where AI can take paperwork off the estimator's desk and where it must not touch the numbers, and points to the article in this series that goes deeper on each step.

Where in the bid cycle does an MEP contractor win or lose work?

Most bids are decided in the office, not in the market. Between the bid invitation and the award there are about eight steps, and a small MEP contractor usually loses work at three of them: the go/no-go decision nobody makes, the wait for supplier and subcontractor quotes, and the silence after the proposal goes out.

The table below is the cycle as we see it in building services offices. It follows the typical commercial path in our source notes: the GC posts the bid, the subcontractor's estimating team prices it, a proposal goes out, an award follows. The two measured facts around that path are that estimators are among the hardest roles to staff (AGC, 2025) and that proposal automation is the AI use early adopters rate highest (Roads & Bridges reporting Dodge and CMiC, 2025).

Step What happens Where it stalls What AI can draft What stays with a person
Bid invitation GC sends drawings, specs, addenda, due date Sits unread in a shared inbox A one-page summary of scope, systems and dates from the bid package Who owns the bid
Go/no-go Decide whether to price it Never formally decided A checklist against your own criteria The decision
Takeoff Quantities from drawings One estimator, several jobs A list of spec sections that affect the takeoff The takeoff itself
Supplier and subcontractor quotes Wholesalers and specialist subs price their part Late, incomplete, in different formats One comparison table with missing lines flagged Checking every price against the original
Pricing and labour Material, labour, plant, risk Depends on one person's memory Nothing; this is judgement All of it
Proposal Scope, exclusions, clarifications, price Copied from the last job A first draft from your own template Exclusions and sign-off
Follow-up Keep the GC informed until award Nobody owns it Short emails keyed to the tender calendar Whether and when to send
Award or loss Result and, if lucky, a reason No record kept A one-line entry in a bid log The lesson for the next bid

Where the days actually go between request and quote is mapped in Too slow to quote: why MEP bids are lost before price is compared. How to read your own record of wins and losses is in raising your bid win rate.

Why is hiring another estimator not the easy answer?

Because there are not enough of them to hire. In the Associated General Contractors' 2025 workforce survey of firms with $50.1 million to $500 million in revenue, 75% of firms with estimating openings said those roles were hard to fill (AGC, 2025). And a new hire still has to learn how your firm prices labour.

The same survey shows the trades behind the estimate are just as tight. Among firms trying to hire, difficulty was reported by 85% for plumbers, 89% for electricians and 78% for pipefitters and welders, although the sub-samples are small at 23, 35 and 57 firms (AGC, 2025). Nationally, 92% of construction firms report difficulty hiring, and 45% name worker shortages as a leading cause of project delays (NCCER with AGC, 2025).

Two caveats. The AGC segment is larger than most firms reading this, so treat it as direction, not as your market. And the data is American; a European firm will recognise the pattern but should not quote the percentages as its own.

The practical consequence is simple. If you cannot add estimating hours, every hour spent on a bid you were never going to win is an hour taken from one you could have won. The cost of a lost bid gives you a worksheet to count those hours. And if most of your pricing knowledge sits with one senior person, capturing estimating knowledge before they leave is the more urgent job.

What can AI realistically do in an estimating office?

AI is good at reading, sorting and drafting text, and poor at knowing what your labour actually costs. In an estimating office that means it can summarise a bid package, line up supplier quotes and draft proposals and emails. It should never produce a price, quantity or labour figure that nobody has checked against a source.

Adoption is already further along than many owners assume. In a 2025 Houzz survey of more than 700 construction and design professionals, 32% of construction firms used AI, and the share rose to 64% among construction firms with more than ten employees (Houzz, 2025). The most common uses were administrative work (47%), project and client management (42%) and sales and marketing (37%), and 92% of the contractors using AI had no formal training (Houzz, 2025).

The Dodge Construction Network and CMiC research points the same way and adds the catch. 87% of contractors expect AI to improve how projects are planned, managed and delivered, and 92% of early adopters say automated proposal tools work well, but only 26% say their data is in good shape (Roads & Bridges, 2025). That last figure is the one to remember. The model is rarely the bottleneck. Your old pricing spreadsheet is.

Good jobs for AI in a building services office Jobs that stay with the estimator
Summarising a bid package: systems, spec sections, addenda, due dates Deciding whether to bid
Turning supplier quotes in mixed formats into one table Accepting or rejecting a supplier price
Flagging lines that one quote has and another lacks The takeoff and the labour allowance
Drafting exclusions and clarifications from the estimator's notes Final wording of exclusions
Drafting follow-up emails Any start date or programme promise
Writing a case study from closeout documents Confirming the facts with the client

The rule we use: AI output is a draft, and every number in it is traced back to a document before it goes anywhere near a bid.

How do you turn three supplier quotes into one table you can price from?

Give the AI the quotes and a fixed list of columns, and tell it to copy, never estimate. The result is a draft comparison table with item, unit, quantity, unit price, delivery and exclusions, plus a list of lines one supplier priced and another did not. The estimator then checks each figure against the original quote.

Example (illustrative, not a real case): A 15-person building services firm is pricing the plumbing and heating package for a small office fit-out. Wholesaler A sends a PDF with unit prices per item. Wholesaler B sends a spreadsheet that bundles valves and fittings into one lump line. Wholesaler C sends a photo of a printed quote with handwritten delivery terms. The estimator pastes all three into a general AI assistant with this instruction:

Build one table from the three quotes below. Columns: item, size, unit,
quantity, unit price, line total, delivery, supplier note. Copy values exactly.
If a value is missing, write MISSING. Do not calculate or guess any price.
After the table, list every item that appears in one quote but not in another,
and every line where units differ (each, metre, box).

The table comes back in a minute. The useful part is the second list: Wholesaler B's lump line hides the pressure-reducing valves, and Wholesaler C quoted pipe per box, not per metre. Those are the two things that would have quietly broken the comparison. The estimator rings B for a breakdown and converts C's units by hand.

The full method, including how to handle subcontractor quotes for insulation or controls with different scope splits, is in turning subcontractor quotes into one comparable table.

What should an MEP bid proposal tell the general contractor?

A proposal should let the GC compare you line by line with the other bidders and see exactly what you left out. That means a scope broken down by system, explicit exclusions and clarifications, the drawings and addenda you priced, your submittal and commissioning assumptions, and a validity date. Marketing language comes last, if at all.

What decides a mechanical award is usually not the cover letter. It is whether the GC's estimator can drop your number into their bid without phoning you, and whether your exclusions will come back as change orders later. A proposal that lists "ductwork insulation by others" on page three is honest. A proposal that says nothing about insulation is a future argument.

Some lines belong in an HVAC proposal because a code or regulation stands behind them. In the US, the International Residential Code (Section M1401.3) requires equipment to be sized with ACCA Manual S based on Manual J load calculations, although enforcement is described as spotty (Green Building Advisor, 2017). Stating the load calculation basis protects you on residential and light commercial work. Technicians who handle refrigerant need EPA certification, and some states and localities add a licence (BLS, 2026). In the EU, refrigerant work must be done by certified personnel under the F-gas rules (IDM Energiesysteme, 2026).

AI helps here in a narrow way. It turns the estimator's scribbled notes into consistently worded clarifications, checks that every addendum on the bid invitation appears in the proposal, and keeps your template's section order. The section-by-section version, written from the GC's review checklist, is in the HVAC bid proposal template.

How do you follow up on a bid without chasing the GC?

Follow up against the tender calendar, not a sales funnel. A short note confirming receipt, one before the GC's own deadline to the owner, and one around the award date tell the GC you are still available and give them a reason to ask questions. Each message should carry a piece of information, not just a nudge.

The GC is assembling dozens of packages. Your follow-up is useful when it saves them work: confirming you have priced addendum 2, noting that a long-lead item has a delivery date that affects the programme, or asking whether a value-engineering alternative would help. It is a nuisance when it asks "any news?".

Be careful in peak season. Overtime and irregular hours are common for HVAC technicians during the heating and cooling peaks (BLS, 2026), and a follow-up email that promises a start date your crews cannot meet is worse than none. AI can draft the three messages from the estimator's notes in a few minutes. The estimator removes anything that sounds like a promise. The three emails, with wording, are in following up on a construction estimate.

Where do new leads and bid invitations come from for an MEP firm?

They come through two different pipes. Commercial work arrives through general contractors' bid lists, repeat relationships and references. Service and residential work arrives through search, Google Business Profile, Local Services Ads and reviews. The marketing you need depends on which pipe feeds your firm, and most firms of this size have some of both.

Commercial pipe. A GC adds you to a bid list because someone vouches for you or because your website and references show you have done this kind of system, at this size, on this kind of deadline. That is why what general contractors look for on an MEP contractor website and turning closed jobs into reference case studies are marketing articles in this series, even though neither mentions advertising.

Service and residential pipe. In the US, HVAC, plumbing and electrical firms are eligible categories for Google Local Services Ads, and the ads need a licence or business registration before the badge (Google Local Services Help, accessed 2026). Google automatically credits some charged leads it judges low quality, usually within 30 days, and lead credits are not available to advertisers in EMEA (Google Ads Help, accessed 2026). On Google Business Profile, a service-area business should hide its address, keep the service area within about two hours' drive of its base, and keep service keywords and locations out of the business name (Google Business Profile Help, accessed 2026).

Reviews matter more on this pipe. In BrightLocal's 2025 survey of 1,026 US adults, 74% of consumers used two or more review sites before deciding, and email was the most effective way to ask for a review at 40%, ahead of asking in person at 27% (BrightLocal, 2025). A trade-press survey found that over 90% of homeowners rate online reviews as important when choosing an HVAC contractor, yet only 61% of HVAC contractors show reviews on their own website (ACHR News, 2024).

The lead itself is usually lost after the request, not before: the quote comes too late or too vague. HVAC contractor leads connects where leads come from with how fast you answer them.

Which rules limit what you can promise in marketing and proposals?

Most of the rules that matter here are trade rules, and your marketing must not contradict them. Refrigerant work needs certified people, equipment sizing has a code basis, fossil-fuel boiler incentives are ending in the EU, and review practices are regulated. A web page or proposal that implies otherwise is a liability, not a sales line.

Rule Where it applies What it means for your copy
EPA certification for technicians handling refrigerant; some states add a licence (BLS, 2026) US Say who on your team is certified; do not imply every installer is
IRC M1401.3: size equipment with Manual S based on Manual J (Green Building Advisor, 2017) US, residential State the load calculation basis in the proposal
F-gas Regulation 2024/573: from 1 January 2027, self-contained heat pumps up to 50 kW with refrigerant GWP of 150 or more may not be placed on the market; split systems up to 12 kW using any F-gas are banned from 1 January 2035 (IDM Energiesysteme, 2026) EU Installed systems need not be removed; do not scare clients into replacement
EPBD recast, Art. 17(15): no public financial incentives for stand-alone fossil-fuel boilers from 2025 (Directive 2024/1275 via LawPlayer, 2024) EU Do not advertise boiler replacement subsidies that no longer exist
FTC rule banning fake reviews, including AI-generated ones, with civil penalties for knowing violators (FTC, 2024) US Never let AI write a review or testimonial
Google bans paid or incentivised reviews and selective asking of happy customers (Google Maps Help, accessed 2026) Global Ask every client the same way, with no discount attached

There is also a market shift underneath. EHPA data shows 2.88 million heat pumps sold in 21 European countries in 2025, up 13% on 2024 but below the 3.31 million record of 2022 (pv magazine, 2026). For a European building services firm, heat pump installation and commissioning references are worth documenting carefully, because that is where the next bid packages will ask for experience.

Where should a small MEP firm start with AI?

Start with one live bid, not a strategy document. Take the next invitation you decide to price and use AI for three jobs only: the scope summary, the quote comparison table and the follow-up drafts. Write down what it got wrong. After a handful of bids you will know which jobs to keep and what your data is missing.

A workable first sequence:

  1. Write your go/no-go criteria on one page. Job types you want, GCs you trust, minimum lead time, systems you will not price. AI can turn a ten-minute conversation into a checklist; the owner signs it off.
  2. Summarise the next bid package with AI. Systems, spec sections that touch your scope, addenda, submittal requirements, due date. The estimator reads the summary and the spec index side by side and notes what the AI missed.
  3. Build the quote comparison table using the instruction above. Every price is checked against the original document.
  4. Draft the proposal from your template, with AI rewording the estimator's clarifications and checking addenda.
  5. Draft three follow-ups keyed to the tender dates, and put them in the calendar.
  6. Log the result in one line: won or lost, by how much if you know, and the reason the GC gave.

Do not feed client drawings or prices into a tool your firm has not approved for that use. More than half of contractors already worry about data accuracy and security with AI (Roads & Bridges, 2025); decide which tools are allowed before an estimator decides for you.

The point of the first month is not speed. It is finding out how much of your estimating knowledge exists only in people's heads. That is the work that makes every later AI step reliable.

Which article in this series should you read next?

Pick the one that matches the step where your firm loses the most bids. If quotes go out late, start with the quote speed and comparison articles. If quotes go out on time but rarely win, start with the proposal, follow-up and win-rate articles. If the problem is too few invitations, start with the website and references.

All articles in the MEP and building services series:

If the first job turns out to be writing down how your firm works, what it prices, for whom and with which exclusions, that is what our Business Brain Starter Kit is being built for: a workbook and a chain of prompts that turn company knowledge into plain files with a source next to each line. Get notified when it launches.

FAQ

Can AI price an MEP bid?

No, and it should not try. AI can read a bid package, summarise the scope, line up supplier quotes and draft the proposal text, but it does not know your labour rates, your crews or the risks on a given site. Every price, quantity and labour figure has to come from the estimator and be traced back to a document before it goes into a bid.

Which AI tool should a small MEP contractor start with?

A general AI assistant that your firm has approved for business documents is enough to start. The tool matters less than the instructions and the checking routine: fixed columns for quote tables, a rule that the AI copies rather than estimates, and an estimator who checks every figure. Decide first which client documents may be uploaded, because drawings and prices are confidential.

Does AI help with commercial bids or only with residential leads?

Both, but in different places. On commercial work it helps with bid packages, quote comparison, proposals and follow-up, which is where most estimating time goes. On residential and service work it helps with fast first replies, review requests and website text. The rules differ too: review requests are regulated by Google and, in the US, by the FTC.

How do I know whether AI is saving my estimator time?

Measure it on real bids. For the next few bids, note the hours spent on the scope summary, quote comparison, proposal and follow-up, and what the AI got wrong in each. Compare with bids priced the old way. We do not publish a typical saving, because it depends on how organised your quotes, templates and pricing history already are.

What should never be written by AI in an MEP contractor's marketing?

Reviews and testimonials, start dates, certification claims and prices. The FTC's 2024 rule bans fake reviews, including AI-generated ones, and Google removes incentivised reviews. Certification claims must match the people you actually employ, such as EPA-certified or F-gas certified staff. Start dates and prices are commitments, so a person who can keep them has to write them.

Is this guide relevant for European building services firms?

Yes, with different rules. The bid cycle, the estimating bottleneck and the quote comparison problem are the same, but the regulation differs: F-gas Regulation 2024/573 sets heat pump and split-system bans, the EPBD recast ends incentives for stand-alone fossil-fuel boilers, and Google lead credits for Local Services Ads are not available in EMEA. Most survey figures cited here are American.

Sources

Related articles

Products from Miriad

Business brainBusiness Brain Starter Kit

Stop explaining your business from scratch. Write it down once, with a source next to every line.

49 USD

Get notified

See all prompt packs