Insights
Why small CNC machine shops should win on organic search, not ads
By Shivam Singh ·
In brief
For a small CNC shop, Google Ads on the main service terms are a poor bet: a click on “cnc machining services” costs about $32.50, partly because national marketplaces bid on it, and manufacturing buyers prefer organic results to ads by 47% to 13%. A 10–50 person shop gets more from organic search, detailed capability pages and being quotable in AI answers, with ads kept to narrow, specific terms. Confidence is low to medium: the click prices come from one keyword tool and the buyer surveys are run by publishers and vendors.
The short answer
A small machine shop can’t outbid the online marketplaces on the terms buyers use most, and it doesn’t need to. The buyers who matter prefer organic results, read supplier websites before they send an RFQ and are starting to use AI summaries to build shortlists. All three reward a good website and specific content more than an ad budget.
I don’t have machine shop clients yet. This post is my reading of the click data, the buyer surveys and what shop owners have said in public about marketing.
What a click costs
| Search term | Searches a month (US) | Cost per click | Source |
|---|---|---|---|
| cnc machining services | 5,400 | $32.50 | seodata.dev, Jun 2026 |
| cnc machining online | 14,800 | $47.59 | seodata.dev, Jun 2026 |
| machining services | 1,900 | $32.60 | seodata.dev, Jun 2026 |
| cnc machine services | 1,600 | $39.99 | seodata.dev, Jun 2026 |
| custom cnc machining service | 170 | $14.65 | seodata.dev, Jun 2026 |
| cnc machining (broad, mostly informational) | 135,000 | $3.00 | seodata.dev |
| Industrial and commercial category average | – | $5.87 | WordStream/LocaliQ 2026 (vendor) |
Commercial-intent CNC clicks cost $15–48, with about $30 as my best estimate. Broad informational searches cost about $3, but the people typing them are mostly learning, not buying. I rate the estimate low to medium confidence, because it rests on one keyword tool plus a category average, and I found no shop’s real spend data.
To turn that into a lead cost, I used the category conversion rate from WordStream/LocaliQ’s industrial and commercial benchmark, 8.2%. At $30 a click, that is about $366 a lead. This is my arithmetic, not a measured figure, and probably optimistic, since the category average covers many easier sales than a custom machining job.
Who you’re bidding against
Search results for machining services are crowded with online marketplaces that bid nationally. Xometry reported 81,821 active buyers and $630 million in marketplace revenue for FY2025, about $7,700 per buyer per year, with 1,760 accounts spending over $50,000 a year. A business at that scale can pay $32 a click across the country. A 15-person shop in Toledo or Wichita bidding on the same term pays the same price for clicks from buyers it may not be able to serve.
I think that national bidding goes a long way toward explaining why service terms cost ten times the broad term. That is an inference; the data shows the price, not who is setting it.
What buyers say about ads
Gardner’s 2025 Industrial Buying Influence study found 47% of manufacturing buyers prefer organic results and 13% prefer paid ads. TREW and GlobalSpec found 51% of engineers avoid clicking paid search links. Across all the buyer surveys I combined, about 85% of technical buyers use search at some point while sourcing, with medium confidence.
AI answers are the fast-growing part. Gardner found preference for Google’s AI Overviews went from 8% of buyers in 2024 to 26% in 2025. AI summaries draw on organic pages, not ads, so this pushes in the same direction.
What shop owners say
CNCCookbook’s 2020 survey of machine shops found word of mouth and networking were the main sources of new work, ahead of online advertising, trade shows and websites. Only 55% of shops had a website, against 75% of the fastest-growing shops. Most of the group said they felt lost and in need of help with marketing, and they rated advertising somewhere between “somewhat effective” and “not so effective at all”. CNCCookbook sells software to shops, and the sample size wasn’t stated.
On the Practical Machinist forum, owners have been blunt. One wrote: “Paid ads did not work for us… SEO has helped increase impressions but it does not appear to be a big success factor.” Another said that “a good website is a much better investment than google ads, at least in the machine shop world,” and warned against letting an SEO person “swamp the site in gobbledygook.” A third couldn’t attribute a single sale to their website after three years. The forum blocked direct access, so these come from search-engine snippets of the threads. Treat them as indicative.
Read together, the complaints are about attribution, distrust of SEO vendors and low-quality traffic, not about search itself. Those are fixable problems.
How big the opportunity is
| Market | Machine shops | 10+ employees | Source |
|---|---|---|---|
| US | 17,156 establishments | 5,647 | Census County Business Patterns 2023 |
| UK | 9,135 enterprises | 1,795 | ONS 2026 |
| Canada | 4,180 establishments (NAICS 3327) | About 1,560 employers with 5–499 staff | ISED 2025 |
| EU-27 | 134,171 enterprises (2022) | Not published; I estimate 25,000–30,000 in the main manufacturing countries | Eurostat |
No survey measures what a CNC customer is worth. From proxies such as Xometry’s spend per buyer and Canadian shop revenues (average CAD 873,900 for small and mid-sized shops), a repeat production account is plausibly worth $20,000–$200,000 or more a year to a 10–50 person shop. That is low confidence, but even the low end makes a single new account worth a lot of content work.
Where ads still make sense
I wouldn’t switch ads off completely. Narrow, specific terms are much cheaper: “custom cnc machining service” is $14.65 a click, and I’d expect terms that name a material, process and place to be cheaper again, with buyers who know what they want, though I don’t have click data for those. A small budget on those terms, with tight geography, is a reasonable test. I’d keep away from the broad service terms unless the shop has capacity to serve buyers nationally.
What I’d do for a 15-person shop
- Write one page per capability that buyers actually search for: process, material, part type and industry served. Include tolerances, part sizes, quantities and certifications such as AS9100 or ITAR registration.
- Publish the equipment list. Engineers check it.
- Make the RFQ form accept CAD files and drawings, and keep it short.
- Ask every RFQ how they found you, and use call tracking. This answers the “couldn’t attribute one sale” problem directly.
- Set up a complete Google Business Profile, so buyers who search your name after a referral find a credible company.
- Write plainly. Clear answers to engineer questions get quoted in AI summaries; keyword-stuffed text doesn’t.
- Keep LinkedIn active. One owner on the Titans of CNC forum said cold calling never worked, but LinkedIn brought in a good job.
Sources
- seodata.dev, “cnc machining services” keyword data (US)
- WordStream, 2026 Google Ads benchmarks
- LocaliQ, search advertising benchmarks 2026
- CIMdata, Xometry full-year 2025 results
- Gardner, 10 things the Industrial Buying Influence data has taught us (2025)
- TREW Marketing and GlobalSpec, 2024 State of Marketing to Engineers (report copy on SlideShare)
- CNCCookbook, 2020 machine shop marketing survey
- Practical Machinist, “what marketing is working these days” thread
- Practical Machinist, “Google ads” thread
- Titans of CNC forum, “Getting business methods?”
- US Census, County Business Patterns 2023
- ONS, UK Business: activity, size and location 2026
- ISED Canadian Industry Statistics, NAICS 3327
- ISED Canadian Industry Statistics, financial performance (NAICS 33271)
- Eurostat, structural business statistics (sbs_ovw_act)