How Should an HVAC Contractor Set Up Meta Ads Targeting Under Meta's New Campaign Structure?

Meta reworked campaign structure, and the audience section now splits into two layers that behave in opposite ways. One layer is a wall your ads cannot cross. The other is a hint the algorithm is free to ignore. Most contractors never learned which is which, because the split did not exist when they built their first campaign.

That distinction decides whether a tech gets dispatched outside your service area, or whether a renter in an apartment complex sees your system replacement offer. Below is how that structure maps onto a contracting business with a service radius, a seasonal demand curve, and leads that arrive by phone. An earlier article on this site, "Why Do HVAC Facebook Ads Reach People Who Will Never Call You?", covers which interests to pick and why cold reach misses. This one stays on where each setting sits in the new structure.

What is the difference between the controls section and the suggest an audience section?

Controls are hard boundaries Meta cannot cross. Suggest an audience is input Meta can follow or ignore. For a heating and cooling company that means your service area is enforced, while age ranges and interests are only hints. Put anything that would waste a truck roll in controls, and everything else in suggestions.

Ben Heath, in his 2026 Meta ads targeting tutorial, describes the split plainly: "anything that you select within controls is a hard boundary, a hard constraint" (3:58). About the other layer he says, "Anything underneath suggest an audience, Meta can choose to go with it or ignore it" (12:00).

For a residential and light commercial heating and cooling business, the sorting looks like this:

One item to check before launch. If your replacement offer promotes financing terms, the video notes that declaring a special ad category changes the targeting options available to you, and that failing to declare it can lead to an ad account being disabled (3:10). Verify against Meta's current policy whether your specific offer counts before you write the ad.

Can a contractor block Meta ads from reaching renters or people under 30?

No. The controls section offers a minimum age but no maximum, and the controls the video walks through contain no homeowner or renter setting at all. Property related signals sit in the suggestion layer, where Meta can ignore them. Hard age limits exist only if you switch the ad set to manual targeting.

Heath points out the missing ceiling directly: "You see there is no maximum age" (7:58). He explains the minimum exists so businesses avoid advertising to people they legally cannot advertise to, not as a targeting lever. The controls he walks through are location, minimum age, custom audience exclusions, and language. Nothing about property ownership appears among them.

That has a specific consequence for this trade. A full system replacement is a homeowner purchase, and every contractor's instinct is to fence out renters and much younger audiences. The platform does not offer that fence in controls. What you can do:

1. Search detailed targeting for property and home related options and add what you find. Whatever is in there is a suggestion, and Meta may override it.
2. Let the offer do the filtering. An ad written around replacing an aging system self selects. A renter scrolls past.
3. Check where your booked jobs actually came from before you narrow anything. If your recent replacement jobs skew to one age band and one set of ZIP codes, that is your evidence, not an assumption.
4. Only if the evidence is strong, use the manual switch to make age a hard limit, and accept the tradeoff below.

The tradeoff is visible in the interface. Heath opens the option to further limit reach and reads Meta's on screen warning: "limiting your audience may reduce performance potential outcome and current up to 7.2% lower cost per result based on their experimentation" (53:13). Meta also drops a campaign score as you restrict. Treat that score as the platform's opinion, not a verdict, but do not ignore it if your cost per booked call rises after you narrow.

One limit on manual mode is worth knowing before you switch. Interests and behaviors stay suggestions no matter what. As Heath puts it, "we can't turn that off as a suggestion even within manual" (56:25). You can harden age, location and custom audiences. You cannot harden interests.

On the location control, his rule is the one that matters most for field service: "what I'd recommend you select here as your location targeting are the areas that you can serve" (4:45). He also warns against overengineering it, saying "don't overthink your targeting. Just very specifically, where do you serve? What's your best markets? Start there, look to expand afterwards" (7:09). In the demo, entering a city defaults to that city plus a 25 mile radius for most places, and he adjusts it up or down by hand. The honest number for a contractor is whatever a tech can reach and still make the next appointment, which is a drive time question the map does not answer.

What should a contractor upload as custom audiences when most leads arrive by phone?

Upload your customer list from field service software, your email list, website visitors at the 180 day maximum, and Facebook and Instagram engagers at 365 days. Phone leads never fire a website conversion event, so your paid invoice history is the closest thing to purchase data Meta can read from a contracting business.

This is where the trade departs from the setup the video assumes. The default flow in the tutorial leans on a website that fires events like purchase or add to cart. Your best jobs end with a signature in a driveway, invisible to the pixel. Your customer export is the substitute.

What to build, in order of value to a contractor:

• Customer list from your field service or invoicing software. The video sets the requirements at a minimum of 100 records and at least one main identifier, with email recommended and more data fields improving the match rate.
• A customer value column. Meta recommends it and Heath does too. If your export can carry the invoice total per customer, include it. That is how Meta learns the difference between a customer who bought a full system and one who paid for a single part.
• Audience labels. The upload flow lets you tag a list as a qualified lead, a disqualified lead, an existing customer, high value or low value. As Heath puts it, "the more information you give about who and what these people are to your business, the better meta is going to be able to plug that back into their targeting" (25:26).
• All website visitors at 180 days. He uses the longest window on purpose: "the reason why I want to create the largest possible custom audience is because I still want to potentially get the conversion from someone who did visit my website 5 months ago" (19:58). If your own quote records show homeowners who priced a system in spring closing months later, that window matters more for you than for a retailer.
• Video viewers and page engagers. Anyone who watched three seconds counts, and the retention maximum is 365 days.

One thing to plan for. The video notes that pixel based and engagement audiences update themselves, while lists uploaded from external sources do not. Your customer export goes stale the moment you close the next job. Put a recurring reminder on the calendar to re-upload.

Should a contractor exclude past customers and maintenance plan members?

Usually no. Exclusions sit in the controls section, so they are permanent walls rather than hints. Heating and cooling customers buy again through renewals, second systems, and repairs that turn into replacements. The narrow case for an exclusion is a promotion priced below what recent buyers paid, and even that is a judgment call.

Heath's own position on exclusions is that they are overused. "by default, I don't like including custom audience exclusions" (42:55). His reasoning is that people who bought recently will often buy again, and that past customers who see your ads leave comments that work as social proof.

Applied to a contracting business, the repeat purchase argument is unusually strong:

• A maintenance agreement renews on a cycle. Excluding members means excluding the renewal.
• A homeowner who bought a furnace still has an air conditioner aging in the yard.
• A repair customer with a unit near the end of its life is your warmest replacement prospect, not a spent lead.
• Past customers in the comments of a local ad are the referral engine. Neighbors recognize names.

The one exclusion worth considering matches the case Heath describes, which is existing customers who paid a certain price seeing a promotion at a lower one. If you ran a spring changeout promotion at one price and a fall promotion below it, excluding the spring buyers is defensible. Decide that against your own renewal and complaint history, not a rule of thumb.

Do lookalike audiences still help a contractor that only serves a few counties?

Rarely as a first move. A lookalike is built as a percentage of Facebook and Instagram users in an entire country, while your service area is a hard boundary in controls. Most of that lookalike is unreachable for you anyway, and the video's position is that Meta now builds similar audiences automatically.

Two points from the video make this concrete. On what a percentage actually means: "this is the percent of Facebook and Instagram users in that country" (50:03). And on the current value of the tool: "Lookalike audiences are now much less valuable and the reason why is because Meta does the lookalike audience thing automatically in the background" (46:53).

The geographic math is our point, not his. A national lookalike is spread across a whole country. Your service area is a handful of counties, and the location control is a hard boundary, so everything outside it is unreachable regardless of how the lookalike was built. What is left is a thin slice. Heath does discuss raising the percentage, but his reason is country population size, not service area. He starts at 1% in a large country like the United States and goes higher in small populations such as Ireland.

There is still one case worth the effort. If Meta delivers plenty of form fills but your dispatcher says the calls are renters, warranty questions, and people outside the radius, a lookalike built from your paid invoice list gives the algorithm a cleaner definition of a real customer. Build it from your highest value jobs, not from all leads.

Is it better to run one ad set for repair and replacement, or separate ad sets?

One ad set, in most cases. Splitting repair and replacement across several ad sets divides a contractor budget into amounts too small for Meta to learn from. Consolidating puts warm and cold audiences together and concentrates conversion events. Split only when you need a hard boundary that suggestions cannot give you.

The consolidation logic in the video is explicit. "as much as possible, we're going to consolidate that down. Sometimes literally to one campaign, one adset" (38:57). And on why the budget concentration matters: "You're going to have way more conversions go through that one ad set, which is exactly the data meta needs to optimize" (39:46).

A contractor running a modest monthly budget across four ad sets, one for AC repair, one for furnace repair, one for replacement, and one for maintenance plans, is starving all four. Consolidate and the same money produces enough events for Meta to find a pattern. Heath's argument is that you then learn whether new creative worked in days rather than weeks, because the spend is concentrated.

The exception is real, and it is structural. If you switch to manual targeting to enforce a hard limit, warm and cold audiences can no longer share an ad set. As Heath explains, "They need to be in separate adsets" (55:38). That is the cost of the hard boundary. You get control and you lose consolidation.

On leaving everything open, the video's condition is worth holding onto: "open targeting has been one of the best performing targeting options across the board for a number of years now," followed in the same passage by "The most important thing is that you have conversion data. Brand new ad accounts tend to struggle a bit" (58:46). A contractor with a new ad account and no uploaded customer list has neither. That is the case for starting with suggestions rather than nothing.

Which tools help a contractor manage Meta ads targeting?

No single tool covers targeting setup, list management, creative, and follow up. Most contractors end up using two or three together. The table below compares what each one does, what it addresses for a company with a service radius and phone leads, and how much advertising knowledge it assumes.

Disclosure: SaleADS.ai is the product of the company that publishes this site. It is listed as one option among several, not as a recommendation over the others, and the row above is not an endorsement. No pricing is shown for any tool in this table because we do not have current, verified figures for all of them.

Where does this information come from?

The platform mechanics come from a 61 minute Meta ads targeting tutorial by Ben Heath, recorded inside an example ad account using a sales campaign and his own agency's marketing services as the example business. He never mentions HVAC, contracting, or any home service trade. Everything about contracting below is our application.

Source: "The ONLY Meta Ads Targeting Tutorial You Need in 2026" by Ben Heath, https://www.youtube.com/watch?v=-CsBl3tv-X0

Taken from the video:

• The controls versus suggest an audience split, and that controls are hard boundaries while suggestions can be ignored
• Location targeting set to the areas you can serve, and the 25 mile radius applied by default to most locations in the demo
• The absence of a maximum age in the controls section, and that the minimum exists for legal restrictions
• That interests and behaviors remain suggestions even in manual setup
• Custom audience setup, the 180 day website window, the 365 day engagement window, the 100 record minimum, the customer value column, and audience labels
• That uploaded lists do not refresh while pixel and engagement audiences do
• The position that custom audience exclusions are usually not worth using
• That lookalike percentages are a share of Facebook and Instagram users in a country, that lookalikes are less valuable now, and that the percentage goes up in smaller country populations
• Consolidation into fewer campaigns and ad sets, the campaign score, and Meta's on screen warning about limiting reach
• That warm and cold audiences must be separated once you switch to manual targeting
• The condition that open targeting depends on having conversion data
• The special ad category declaration and the account risk of not declaring

Not taken from the video: Everything specific to this trade. The drive time reading of the service radius, the phone lead problem and using invoice history as a conversion data substitute, the replacement versus service call difference, maintenance plan renewals as an argument against exclusions, the county level reading of lookalike reach, the aging second system in a past customer's yard, and the financing and special ad category check for a replacement offer. Those are our application of the video's method to contracting, and none of them are claims Ben Heath made.

No industry statistics, benchmarks, conversion rates, ticket prices, or equipment ages appear in this article. Any threshold suggested here is something to verify against your own booked jobs, your own radius, and your own invoice totals.