A US custom metal parts manufacturer we have worked with for 15 years gets a steady stream of quote requests from Google search, and has for most of that time. That flow is what a job shop is really buying when it starts Google Ads; the ads are simply the fastest way to get in front of the same buyers, and it takes about 90 days for a campaign to settle.
Get those 90 days right and you end up with something most shops never have: quote requests that don’t depend on word of mouth or on two big customers.
Short answer: In the first 30 days a manufacturing Google Ads campaign mostly buys data, not customers: you learn which searches bring buyers and which bring job seekers and hobbyists. Days 31 to 60 are for cleaning the campaign and usually bring the first qualified quote requests (RFQs). By day 90 you have enough numbers to decide whether to scale, hold or stop. Shutting the campaign down in week three, before any of that has happened, is the most expensive mistake a shop can make.
Table of Contents
Why 90 Days, Not 30
Most job shops that try Google Ads have never advertised before. Their customers came from referrals, former employers, and one or two OEMs that grew over the years. Scroll through listings of machine shops for sale and the phrase “relies only on word of mouth for marketing” shows up again and again.
That history shapes expectations. An owner who has never paid for a lead expects the ads to behave like a sales rep on commission: money out, orders in. Google Ads behaves more like a new machine. It needs setup, calibration and a few weeks of scrap before it runs clean.
For this guide we went through what shop owners write on machining forums and in business listings, and combined it with what we see in campaigns for manufacturers. Two numbers explain why the effort is worth it.
A Thomas survey of industrial buyers found that 71% vet fewer than five suppliers before deciding; the survey is from 2019, but the mechanics have not changed. According to a 2025 Sopro survey, nine in ten B2B buyers use online channels as their primary way to find new suppliers. The shortlist is short, and it is built online.
Before Day One: Four Things That Decide the Campaign
Most failed manufacturing campaigns were lost before the first click, in the setup. Four things decide whether the 90 days produce data or just spend.
Conversion tracking on the quote form. Conversion tracking is the counter that tells Google which clicks turned into a quote request. Without it the campaign is a machine with no gauges: it runs, but nobody knows what it produced. Track the form submission, and track phone calls if your buyers call.
A landing page per service, not the homepage. Someone searching “sheet metal fabrication services” should land on a page about sheet metal fabrication: capabilities, materials, tolerances, certifications, and a quote form. The homepage tries to talk to everyone and converts nobody. The structure of a page that converts is covered in Landing Pages for Manufacturers and What a Manufacturer’s Homepage Needs.
Keywords per niche, plus a negative keyword list. Build a separate ad group for each service you actually want work in: CNC machining, sheet metal, welding, coating. Negative keywords are the words that stop your ad from showing. Without them, “jobs”, “salary”, “training”, “for sale” and “DIY” will eat a third of the budget. The starter list below is where every manufacturing campaign should begin; Keyword Research for Manufacturers shows how to build the positive side.
| Category | Starter negatives | Why they cost you |
|---|---|---|
| Job seekers | jobs, careers, salary, hiring, apprenticeship, resume | The largest single source of wasted clicks on “cnc machining” |
| Learners | course, training, certificate, tutorial, how to | People learning the trade, not buying parts |
| Equipment buyers | for sale, used, price, lathe, mill, router | Want a machine, not a part |
| Hobby and one-off | diy, hobby, home, garage, etsy | One piece, no drawing, no repeat |
| Software | cad, cam, software, download, free | Tool users, not buyers |
Manual bidding to start. Google will push you toward automated bidding, where its system sets the price per click for you. Automation needs conversion data to work, and a new shop account has none. Start with manual cost per click, which means you set the ceiling yourself, and let the automation earn its way in later.
Days 1–30: You Are Buying Data, Not Customers
The main job in month one is the search terms report, the list of exact phrases people typed before they clicked your ad. Google matches keywords loosely, so “cnc machining services” will also show your ad for “cnc machining jobs near me” until you tell it not to.
Expect a large share of the early clicks to be irrelevant, even with a good starting list. Every irrelevant phrase goes into the negatives, and by the end of the month that list is two or three times longer than what you started with. That is the calibration, and you pay for it once.
A quote request can come in month one, and often does, but it is not the measure of success yet. Impressions, click-through rate (the share of people who see the ad and click it) and the share of relevant search terms tell you whether the campaign is pointed at buyers. Cost per quote request is a month-three number.
“I spent $500 on Google Ads when I started because I didn’t know any better. $500 is nothing. You have to spend a lot more before you can even tell if the campaign is working.”
Owner, CNC job shop, forum discussion, 2024
The owner behind that comment was right about the budget and wrong about the timing. Three weeks of a small budget shows you the noise, not the signal.
Days 31–60: Cleaning Up and the First Real Quote Requests
Month two is when the campaign starts to look like a campaign. Ad groups that brought clicks but no quote requests get paused. Keywords get tightened from broad matching to phrase or exact matching, so the ad only shows for searches that contain your phrase, and the ad copy is rewritten with the words buyers actually typed.
This is also when the first qualified quote requests usually arrive. A qualified request has a company email domain, a drawing or a part description attached, a quantity, a material and a deadline. An unqualified one is a hobbyist with one part and no drawing, or a student with a school project.
“A machine shop doesn’t want the general public calling. It’s always the fix-my-lawnmower type of job.”
Shop owner, forum discussion, 2025
You can filter the second kind before it reaches your inbox. Put “production quantities” and “OEM suppliers” in the ad copy, make quantity and material required fields on the form, and let the negative list handle the rest. A shop that does this in month two spends month three reading drawings instead of deleting emails.
Days 61–90: Enough Data to Make Decisions
By the third month you can see which services convert, which regions send drawings and which send noise. Budget moves toward the ad groups that produce quote requests and away from the ones that only produce clicks.
This is also the point to consider Smart Bidding, Google’s automated bidding that sets the price per click based on how likely a click is to convert. Google’s own guidance says you typically need at least 15 conversions in a 30-day period for it to work, and recommends at least 30 conversions in the last 30 days before judging a Target CPA strategy. Below that, stay on manual bidding; the automation would be guessing with your money.
By day 90 three outcomes are on the table: scale the ad groups that win, hold the budget and keep cleaning, or stop because the numbers say the niche is too small for paid search. All three are legitimate decisions. Deciding in week three is not, because the data does not exist yet.
| Period | What you do | What you can see | What you cannot judge yet |
|---|---|---|---|
| Days 1–30 | Add negatives, read search terms, test ad copy | Impressions, click-through rate, share of relevant searches, maybe a first quote | Cost per quote request |
| Days 31–60 | Pause losers, tighten match types, rewrite ads, filter the form | First qualified quotes, falling waste | Return on the spend |
| Days 61–90 | Shift budget to winners, consider Smart Bidding | Cost per quote request by service and region | Nothing: this is decision time |
The Numbers: Budget, Cost per Click and Cost per Quote (US, UK, Canada)
Two sources put numbers on all of this. The first is search demand and estimated cost per click for the core service searches, pulled from Ahrefs in September 2026. The second is the LocaliQ and WordStream 2026 benchmark report, built on 13,474 US search campaigns, which puts the Industrial & Commercial category at $5.87 per click, a 6.57% click-through rate, an 8.20% conversion rate and $75.19 per lead.
| Search phrase | US: searches / est. CPC | UK: searches / est. CPC | Canada: searches / est. CPC |
|---|---|---|---|
| cnc machining services | 3,100 / $0.60 | 1,100 / $4.50 | 250 / $1.50 |
| sheet metal fabrication services | 1,100 / $3.50 | 200 / $1.10 | 60 / no data |
| welding services | 600 / $1.30 | 450 / $1.10 | 200 / $1.40 |
| powder coating services | 350 / $1.30 | 300 / $1.30 | 150 / $1.20 |
Monthly searches and estimated cost per click in USD, Ahrefs, September 2026. Estimates like these tend to sit below what you actually pay once competitors bid on the same phrases, so plan with the benchmark and treat the estimate as the floor.
Run the math and the budget question answers itself. At the benchmark cost per click, $500 a month buys about 85 clicks and, at an 8% conversion rate, around seven leads. Job shops rarely hit that rate, because a quote request with a drawing attached is a bigger ask than a phone call to a plumber; 3 to 4% is a more careful planning figure, which means two or three quote requests a month.
That is a test of one service in one region, not a campaign that can learn. At two or three conversions a month you reach Google’s 15-conversion threshold in about five months, long after the 90 days are over. Ten dollars a day is the same story, smaller. A budget of $1,000 to $1,500 a month is where a campaign covering two or three services collects enough data inside 90 days to make the decisions above.
Cost per quote request is the number that matters, and it only stabilizes in month three: highest in month one, lower as negatives accumulate, and by day 90 a figure you can compare with what a trade show booth or a marketplace commission costs you per qualified inquiry.
Why Shops Quit in Week Three (and What It Costs)
The week-three exit usually goes like this. A shop puts $500 in, sees 80 clicks and a search terms report full of job seekers, gets one form fill from a student, and shuts it down as a waste. Every dollar of that was spent on calibration; the negative list it produced was the first useful asset the campaign built, and it gets thrown away with the campaign.
In our campaigns for manufacturers the pattern is consistent: the first qualified quote request shows up in month two, and the first order from it often lands four to eight weeks later, because a buyer who sends a drawing still has to collect quotes from two other shops and get sign-off internally. A shop that judges the campaign at day 21 is judging it before the buyer has finished comparing.
The situation that brings most shops to paid search looks the same. A twelve-person shop has three customers that make up most of its revenue. One moves production, capacity drops by a third, and the owner realizes there is no pipeline because there has never been one.
The fix is not complicated: a landing page per service, a campaign on the searches buyers actually use, and 90 days of cleaning. The first quote requests arrive in month two, and by month three the owner holds a number he never had before: what a qualified quote request costs, which he can now compare against the price of a booth at FABTECH.
When Google Ads Is Not the Right First Move
Three situations call for something else before paid search.
Nobody can answer a quote request within a day. Buyers who search are comparing shops in parallel. A request that waits three days goes to the shop that answered in three hours. Fix the response process before you pay for requests.
There is no landing page. Sending paid traffic to a homepage built in 2015 wastes most of it. Build the service page first, then advertise it.
Nobody searches for what you do. The Canada column above shows some services get 60 searches a month or fewer. For a niche that small, SEO for manufacturers and direct outreach to the OEMs you can name do more than paid search; how the two get sequenced is on the services page.
The Bottom Line
Google Ads works for a job shop only after 90 days of cleaning and learning, and the most expensive mistake is turning it off in week three, before the campaign has learned anything. Set it up right, budget for the learning, and by day 90 you will hold a number most shops never have: what a qualified quote request costs you, and whether you want more of them.