In-House vs Outsourced Estimating: Worksheet, Scorecard, Back-Test

In-house vs outsourced estimating comes down to one comparison: your own takeoff hours against the outside quote, counted in the same hours, plus the hours you still spend checking the file. If your hours come in below that total, keep the work in house. If they come in above it, or you will run out of hours before a bid is due, send it out.

This guide gives you a worksheet for that sum, a decision table, a scorecard for any provider and a back-test on a bid you already priced. Takeoff Services LLC sells outsourced takeoffs, so weigh it with that in mind; how an outsourced takeoff works shows what you would be buying. The scorecard is written so any provider can be tested, and the worksheet can end in keep it in house.

In-house vs outsourced estimating: compare hours, not fees

Outsourced estimating means an outside firm does part of the estimate, most often the quantity takeoff. This guide tests the takeoff, because you can check it against a bid you already priced.

Estimating is three jobs: counting from the drawings, pricing the count and deciding what to bid. The count needs the least from your office: the drawings, the specifications, the addenda and your trade list. Pricing leans on your cost history and your subs, and the bid decision is yours. So the choice is rarely all or nothing.

The outsourced estimating pros and cons come down to four trade-offs: cost per bid, control of method, spare capacity in a crunch, and risk to the date and the files. A worksheet prices the first. A scorecard and a back-test cover the rest.

Write down three numbers first: bids, hours and hit rate

Three numbers drive the whole sum. Take them from your records, not from memory.

• Bids per month: bids you submitted, not invitations you received.

• Hours per takeoff: your timesheets for the last five similar jobs, with the rework after addenda included.

• Hit rate: wins divided by bids submitted, the figure the bid desk guide calls a win rate. A few dozen bids give a wide range, and the win-rate interval table shows how wide.

Suppose you submit 6 bids a month at 14 hours each and win 20 percent. That is 84 takeoff hours a month. The hours spent on bids that lose are 6 x 14 x (1 - 0.20) = 67.2, which is 80 percent of the total. The hours spent per win are 14 / 0.20 = 70. Seventy hours per win is the figure an outside desk has to beat, with your checking time included.

The break-even worksheet, with your own inputs

Fill the table from the top. The example column uses the inputs above and invented figures for the rest.

Table 1. The worksheet. Only B, H, W and the wage are facts you already hold.

Line m needs a source. In the Bureau of Labor Statistics employer cost data for June 2026, wages and salaries were 69.5 percent of compensation costs for private construction industry workers. Benefits, which include paid leave, were 30.5 percent. Total compensation is then 1 / 0.695 = 1.44 times the wage. That is an industry average per hour worked, so swap in your own benefit costs.

Seat extras are software, training, a desk and management time, entered as a fraction of the hourly wage: 0.15 means 15 percent of the wage. The productive share is the part of hours at work spent estimating, after meetings and admin. Leave stays out of that share because the Bureau figure already spreads paid leave over hours worked.

In the example the desk costs 8 + 4 = 12 hours against your 14. That saves 2 hours a bid and 12 a month. The saving ends if the quote passes 10 hours or your checking passes 6. Per win, that is 70 hours in house against (4 + 8) / 0.20 = 60 outside. A margin that thin disappears with one slow review.

The second result is time. You get back 6 x (14 - 4) = 60 hours a month. Those hours earn money only if you decline bids today or have better work for them. Count last quarter's declined bids before you count any saving.

Keep it, outsource one bid, outsource a trade, or split

When to outsource estimating depends on the test in the last column, not on the size of the firm.

Table 2. Four choices and the test for each.

The first row is a real answer. If S is zero or negative and you decline nothing, an outside desk adds a handoff and nothing else. Keeping the count in house also has a gain no worksheet counts: a team that does its own counts learns its drawings, assemblies and waste factors. If you are growing an estimator, keep the takeoffs that teach and send out the ones that only take hours.

If coordination, not counting, eats your week, look at the bid process instead. The jobs that can move are listed under the bid desk. Bidding software or a bid desk compares a tool with a person for them. Takeoff Services LLC offers three engagement models, from a single bid to a named estimator, and which model fits the way you bid can also answer that none does.

A restoration contractor shows why the split matters. The estimator who walked the loss holds the photos and readings that support a supplement, so that documentation has to travel with the file or the work stays close. Read about a supplement backlog before moving anything, because a better file may fix the problem without a handoff.

Hiring an estimator versus sending the takeoff out

A new seat, meaning one more estimator on payroll, is a fixed cost, so utilization is the number to watch. One seat gives P = 130 productive hours a month (2,080 hours a year / 12 x 75 percent; the example takes no leave out, so a real seat gives somewhat fewer). At 14 hours a takeoff, that carries 9.3 takeoffs.

Table 3. One seat, takeoffs only, at 14 hours each. The outside desk costs 12 hours per takeoff in the example.

In the example the desk beats the seat at every volume the seat can carry. Break-even volume is P / (Q + R) = 130 / 12 = 10.8 bids a month, above the seat's capacity of 9.3. Raise the quote to 12 hours and Q + R becomes 16, so break-even falls to 130 / 16 = 8.1 bids a month and the seat wins above it.

A real seat also prices, buys out and tracks addenda, and the work around the count is more than counting. Table 3 gives the seat takeoffs only, so it overstates the seat's cost per takeoff when the seat would do that other work too. Credit it for any of that work you would otherwise do yourself. A new hire also ramps up, so expect slower output in the first months.

If the choice is wider than in house or out, use software, a part-time hire or an outside desk. That guide holds the side-by-side table.

Outsourced estimating pros and cons, with a protection for each

The gains are plain: hours back, as the worksheet shows; capacity when due dates bunch up; depth in a trade your team does not carry; and no seat sitting idle at low volume. The costs are below, each with a protection you can ask for before the first file moves.

Table 4. Six downsides and the protection for each.

Paying for a workbook does not by itself settle who holds the rights. The Copyright Office says a commissioned work counts as made for hire only if it falls in one of nine listed categories and both parties sign a writing that says so.

A bid document can also limit who may see its drawings, so read the solicitation before you share a set. One label for such limits is controlled unclassified information, which the National Archives describes as information that needs safeguarding or controls on who receives it. If the terms limit who may see the drawings, keep that job in house unless they say otherwise. This explains the rule and is not legal advice.

A scorecard for any outside desk, this firm included

Score each check 0 for missing, 1 for verbal or vague and 2 for specific and in writing. Checks 1 to 4 show in a delivered file. Checks 5 to 8 live in the quote and terms. Set the pass line before you see anything: 12 of 16 is workable, with no zero on checks 5 to 8, because none of those can be repaired after the files arrive.

Table 5. The scorecard. Checks 5 to 8 live in the quote, and a delivered file cannot show them.

A quote you can score names the trades and the drawing issue it prices. It lists the addenda, the deliverables, the delivery date, the price and what would change it. If any item is missing, ask before you send files. What a complete takeoff file holds is laid out in a worked example, so you know what a 2 looks like on checks 1 to 4.

Takeoff Services LLC quotes one fixed figure before work starts and bills nothing until you approve it, with delivery 24 to 96 hours from an approved quote. Score it the same way. To grade checks 1 to 4, look at a sample package; the library asks for your name, company, work email and phone number, and the set arrives by email. Grade checks 5 to 8 on the quote.

Run a back-test on a bid you already priced

A back-test sends an outside desk a job you have already priced, so no live bid rides on the result. It is still a purchase: ask whether the provider takes closed jobs, and approve the quote as you would for any other. Run it before a live bid, and again when the worksheet comes out close.

1. Pick a finished bid whose set, addenda and quantities you still hold and whose documents you may share. A job you built is best, because you know what was true.

2. Set the tolerance before the file arrives. The example uses 2 percent on areas and lengths and an exact match on counts.

3. Send the same drawings, addenda and trade list, without your numbers. Log the hours from approval to delivery. Then review the file as you would on a live bid, before you open your own quantities, and log those hours as R.

4. Price both quantity sets with your own unit costs and compare line by line. Name the cause of each miss: the desk's error, your error, a drawing conflict or a different reading of scope.

5. Convert the quote into Q, enter R from step 3 and rerun the worksheet.

Here is an invented tenant build-out. Weight is each line's share of your priced total, and effect is the weight times the quantity difference, in percentage points of that total.

Table 6. Eight lines, weights summing to 100 percent.

The net effect is -0.13 percentage points, which looks like agreement. The gross movement is 2.31 points, about 18 times larger. Four lines differ and three fall outside the 2 percent tolerance. Recount each miss against the drawings:

• Rated partitions, 120 SF short: the desk left a rated corridor wall off its count. That is the desk's error.

• Ceiling tile, 130 SF over: the desk counted a room whose ceiling the specification leaves open and listed it as an assumption. That is a flagged difference of reading.

• Wall base, 130 LF over: you missed two rooms. That is your error, and the desk was right.

Judge the desk on three things: the misses a live review would have caught without your own numbers, whether it flagged what it assumed, and the review hours in R.

Where the method stops working

One job tests one drawing type and one set of trades. It does not test turnaround, or how the desk handles an addendum two days before the due time. Run a second back-test on a different job type. With no closed job you may share, use a low-stakes live bid with slack in the schedule and keep your own count to compare. Make that first live job one with a written delivery date.

Your next step: run the numbers on last quarter

Start with what you already hold: your bid log and timesheets. Run the worksheet on your last quarter before deciding, because in-house vs outsourced estimating is an hours question first and a price question second. If it says keep it in house, keep it. When the break-even is close, run the back-test on one bid you have already priced before changing anything. With a bid due this week and no hours left, send out that one bid and log R.

Frequently asked questions