How to Hire a Construction Estimator

What are you actually hiring for?

"Estimator" is used for three different jobs. Hiring the wrong one is an expensive mistake, and it shows up once the person you hired is either bored or drowning.

Takeoff technician. Measures quantities off the drawings and does not price them. Faster to hire, cheaper to carry, and the role worth ruling out before you hire for pricing. If you or your PMs already know your costs cold and the bottleneck is counting, hiring someone to price is overspending on the part you already do well.

Estimator. Quantities plus pricing. Builds the bid: labor hours at burdened rates, material with sales tax, equipment, subcontractor quotes and scope coverage, general conditions, bonds and insurance, permits and fees, escalation to the midpoint of construction, contingency, and markup. This is what most job postings mean, and it is the role where sector experience matters most, because the pricing half leans hard on past work — knowing what a scope really costs in your sector, and spotting when a number looks wrong — on top of current inputs the estimator has to go out and get: live sub and vendor quotes, the wage rates actually in force, material escalation, and how busy the local market is this quarter.

Preconstruction manager. Owns the pursuit rather than the number. Bid/no-bid calls, sub outreach, scope coverage, value engineering, client-facing conversations. Senior, and usually only justified when you are chasing negotiated or design-assist work. Hiring at this level to produce hard-bid takeoffs wastes the salary and usually loses the hire.

Should you hire in-house or outsource?

The question worth asking first is whether you need an estimator at all, or estimating capacity. Those are not the same thing. A salaried estimator is a fixed annual cost sitting against a bid pipeline that is not fixed — four invitations one month and twelve the next is normal, and there is no way to staff for it exactly.

There are three real arrangements. Hire in-house and carry the seat year round. Outsource per project and pay only on the jobs you bid. Or split it: an in-house estimator covering the trades you self-perform, with overflow and the divisions outside their depth going to an [outsourced estimating](/services/outsourced-estimating/) firm. The split is a common arrangement once bid volume and trade mix stop being predictable, because neither an all-in-house nor an all-outsourced model absorbs that variation well.

The honest version: if you bid constantly and always in the same scope, hire. At high, steady volume, outsourcing every bid is likely to cost more per year than a salaried estimator who knows your crews, your suppliers and your recent job costs — and an in-house estimator with that context will usually turn work around faster. Check it against your own fee quotes and bid count. Anyone who tells you otherwise is selling.

What volume justifies a full-time estimator?

There is no threshold that holds for everyone, because the break-even is driven by seat utilization, not company size. It depends on your own numbers: divide the fully loaded annual cost of a salaried seat by the per-project fee you would pay an outsourced firm, and compare that to last year's actual bid count. Run that arithmetic before using any external benchmark. These signals matter more than the count.

What do CPE, CCP and CEP actually mean?

Three credentials come up constantly on estimator resumes, plus an entry-level fourth, issued by two organizations. Requirements change, so confirm the current ones with the issuing body — but this is what each signals.

Is certification required to be a construction estimator?

No. Estimating is not licensed as a standalone occupation in the United States, and no certification is required to work as an estimator. Contractor licensing, public procurement rules and prequalification requirements vary by state and by owner, so confirm what applies in your jurisdiction. Many of the strongest estimators in the trade have no credential at all, having come off the tools or up through project management, and some of the weakest resumes you will read carry three sets of letters.

What a certification genuinely tells you: the person committed to a structured body of knowledge, sat a proctored exam, and had documented years of experience to qualify. That is a real signal about seriousness and about baseline vocabulary. It is not nothing.

What it does not tell you: whether they know your market's labor rates, whether they understand your sector's means and methods, or whether their numbers hold up against actual costs. A CPE whose entire career is data centers is unlikely to price your tenant improvement work as well as an uncertified estimator with ten years of TI behind them. Sector depth outranks credentials in almost every case where you have to choose.

Certification matters most in three places: cost-estimating and program-management services contracts, mainly federal and owner-side, where the scope of work can name a certified cost professional by title; owner-side and PM-side roles where the estimator is defending numbers to people outside construction; and firms whose clients ask. Worth saying plainly that contractor prequalification for public and DOT work is a different thing — it turns on bonding capacity, past performance, safety record and equipment, not on whether your estimator holds letters. If none of those describe you, treat it as a tiebreaker.

What should you ask in the interview?

Skip the questions with a correct answer they can rehearse. Ask about judgment under incomplete information, which is the job.

"Walk me through a bid you lost badly, and one you won by too much.". Tests whether they run post-mortems. A working estimator knows their spread history — how far off the next bidder they usually land, and in which trades they are consistently high or low. Someone who cannot recall a miss either does not track results or is not telling you the truth.

"The drawings and the specification disagree. What do you do?". The answer you want is that order of precedence is set by the contract documents, not by habit, that the conflict goes in an RFI, and that until it comes back the assumption gets priced and written onto the clarifications list. A flat "specs govern" with no qualification is a memorized answer, not a practiced one.

"Where do your production rates come from?". Published cost data or historical job costs off completed work — those are the two real sources of a production rate. A subcontractor quote is a price, not a rate: it tells you what someone will do the work for, not how many hours it takes, and you can only back a rate out of it if you know the crew and the duration behind it. Push one question further, because the adjustment is where the skill sits. Published rates have to be localized and carry a pricing date. Historical rates have to be corrected for crew size, overtime, trade stacking, weather and lift height. An estimator who reads a number straight out of a book and drops it into the bid has done half the step. Worse is not knowing which source a given number came from at all — it means the number cannot be defended when a PM challenges it six weeks later.

"How do you build general conditions?". This is the line item that quietly eats margin. You want to hear duration-driven pricing built off the schedule — supervision, temporary facilities, equipment on rent, site staff — not a flat percentage of hard cost. A percentage is a shortcut that is roughly right on a typical job and badly wrong on a long, low-value one.

"What happens when the package needs eight days and you have four?". The answer you want is that they tell you before they start, narrow the detail deliberately, and state in writing which parts are carried as allowances. The answer you do not want is that they get it done. Quiet guessing is the single most expensive habit an estimator can have.

"Send me the exclusions list from a recent bid.". Redacted is fine. This one artifact tells you more than the rest of the interview: whether they think in terms of scope boundaries, how precisely they write, and whether they have ever been burned. A thin exclusions list on a complex job means they have not been burned yet, which means you will pay for the education.

How do you evaluate a sample estimate?

Give the candidate or firm a plan set from a project you have already built, so you know the actual cost. Do not tell them the answer. Then read the output against this list — the structure tells you more than the bottom line does. One item matters more than anything else on a general contract and is easy to leave out: the leveled bid tab. You want to see who quoted each trade, what each one included and excluded, where the quotes disagreed on scope, and how the estimator plugged the gap — carried by another sub, priced in-house, or excluded and written onto the clarifications list. Sub numbers with no leveling behind them are arithmetic on other people's assumptions.

• Organized to a recognizable structure and applied consistently, not division 3 in detail and everything else lumped. Which structure depends on the stage: a documented hard bid sorts into CSI MasterFormat divisions, while conceptual and design-build estimates priced off sketches belong in UniFormat systems or an assemblies breakdown, because there are no trade packages yet to sort. Asking for MasterFormat detail on a sketch-stage estimate buys you false precision, not rigor.
• Every quantity traceable back to a sheet and detail, not a bare number in a cell
• Units correct for the trade and internally consistent. Concrete in CY, with formwork in square feet of contact area and reinforcing by weight, not slab area standing in for all three. Drywall and finishes in SF. Structural steel by weight, with connections, miscellaneous metals and piece count carried separately rather than priced per piece. Pipe and conduit in LF with fittings and devices in EA. Flooring SF against SY, aggregate and asphalt CY against tons. On earthwork, check which cubic yard is in the cell — bank, loose or compacted — and whether swell and shrink have been applied in the right direction. That one catches more bad takeoffs than everything else on this list.
• Waste, lap and shrink factors shown as visible, separate inputs rather than baked into the quantity
• Labor, material and equipment each on their own line, with labor hours shown and the labor rate identified as burdened or unburdened — payroll taxes, workers' compensation, liability, fringes, and any union agreement or prevailing wage determination stated — rather than a single blended unit price you cannot take apart.
• Crew composition and production rates visible somewhere in the workbook
• A written list of assumptions, clarifications and exclusions — its absence is disqualifying
• General conditions priced from project duration, not carried as a percentage of hard cost
• The pricing basis and its date stated on the estimate itself, so you know how stale the numbers are
• Delivered as an editable file you can work in, not a locked PDF
• Suspiciously round numbers chased down — a clean figure is usually an allowance nobody labeled
• Three quantities you spot-check against the drawings yourself, chosen by you, not by them

How do you run a paid test bid?

Pay for it. A free sample is a sales artifact, prepared without time pressure on a project chosen to flatter. A paid test is work, and it is the only part of the hiring process that resembles the job.

Use a completed project where you know the actual costs, give the same package and the same deadline you would really give, and say nothing about the outcome. Normalize before you judge the variance. Strip the change-order scope out of the actual cost so both numbers cover the same work, and escalate the historical cost forward to today's market before you set them side by side. Skip that and you are measuring what happened to material prices since you built it, not what the candidate can do. Then sit down and reconcile. The conversation matters as much as the number: a variance is not automatically an error, and the useful question is which assumption caused it. An estimator who can point at the line, explain the logic and tell you why they carried it that way is doing the job correctly even when they land high.

This works for outsourced firms too. Before any standing arrangement, run one real bid through them and reconcile it the same way. If a firm will not do a paid test project, or will not walk you through its own workings afterward, that answers the question. If what you actually need is the counting rather than the pricing, ask for a [quantity takeoff](/services/quantity-takeoff-services/) and price it yourself — it costs less, and your own numbers will beat anyone else's on your own trades.

What are the red flags?

These apply to individual candidates and to outsourced firms alike.

• An accuracy claim with nothing behind it. Accuracy ranges themselves are legitimate — AACE International's cost estimate classification system ties them to how complete the design is, so a Class 5 conceptual estimate carries a wide range and a Class 1 bid-ready estimate a narrow one. What should worry you is a blanket figure, "we are within two percent every time", quoted with no estimate class named, no statement of what design documents it was based on, and no answer on whether contingency sits inside the number or outside it.
• No exclusions list, or one that reads as boilerplate copied between jobs
• Cannot explain where a specific number came from when you point at it
• Will only deliver a locked PDF, so you cannot audit or adjust anything
• Unwilling to say "the drawings do not show this" — and instead fills every gap silently
• A takeoff fee that does not change with drawing quality, which suggests nobody is really reading the set
• A resume that jumps sectors every two years with no depth anywhere
• Reluctance to do a paid test, framed as a trust issue rather than a scheduling one

How do you compare in-house cost against outsourced cost honestly?

Do not compare a salary to a per-project fee. The in-house column is salary plus payroll taxes and benefits, takeoff software licenses, a cost data subscription, hardware, recruiting cost, and the ramp period before the hire is producing bids you would send out. Then add the months when bid volume does not fill the seat, which is the cost easiest to leave out entirely.

The outsourced column is the fee per project times your realistic bid count, plus your own review time, which is never zero. Someone in your office still has to read the estimate, apply your markup and own the number.

The comparison that means anything is cost per bid at your actual volume. Run both columns against last year's real bid count, not the count you hope for this year.

One thing sits outside both columns. An in-house estimator accumulates knowledge of your crews, your suppliers, your recurring clients and what your last thirty jobs actually cost to build. That compounds, and outsourcing does not replace it — an external firm can produce a rigorous, well-structured estimate and still not know that your framing crew consistently outperforms published production rates, or by how much. If you can genuinely fill the seat, hiring wins in the long run on more than cost. If you cannot, carrying an underused estimator to capture that benefit is the more expensive mistake.