The short version
Pull up your own ad for a minute and read it like someone who's never heard of you. They've got a tired roof or a fence that's falling over, and they've got one question: what do I get if I say yes?
So we went through the ads. On the morning of 2 November 2026 we used Meta's Ad Library to go through up to 10 active ads from each of 50 Australian trade businesses, 10 each from searches for roofing, concreting, fencing, painter and commercial cleaning. That came to 162 ads aimed at customers.
18 of the 50 had at least one ad that asked for an enquiry and said what you'd get: a named thing or a fixed scope, a stated time, or a guarantee with terms. One ad out of as many as 10 was enough to count, so that's the generous reading. On the first ad the Library showed for each business, it's 15.
The other 32 stopped short. At their best, 18 named the service and stopped, 10 offered a free quote and nothing more, 1 offered a deal, and 3 didn't have an ad that asked for anything.
Asking isn't the problem. 47 of 50 asked for an enquiry in at least one ad, and 34 of 50 used the word quote or estimate to do it. What's usually missing is the other half of the exchange.
There's a second gap. 92 of the 162 ads stated nothing that would turn untrue if a competitor ran them under its own logo. In fairness to the trades, half of these businesses, 25 of 50, had at least one ad that did. Only 10 had an offer and a fact like that in the same ad.
If you change three things this week, make it these. Name one thing the customer gets after they respond. Put it in every ad that asks, not only one of them. And put a fact in the same ad that only your business could state. Each finding is below with the fix for it, and the whole check is at the end so you can run it on your own ads tonight.
This shows what was being offered on one morning. It doesn't show which ads worked, because Meta doesn't show that to anyone looking in from outside.
Up to 10 distinct active ads per business, read in Meta's Ad Library on 2 November 2026. Each business sits on the best rung any of its asking ads reached. Video ads were coded from the ad's text and first frame only, so businesses with only video ads are a floor.
What counted as saying what you'd get
The rules were written down before we opened a single advertiser, and the close calls they didn't settle got ruled on in the review covered in the method section. An ad only had to clear one of three bars, and it only got judged on this if it asked for an enquiry.
A named thing or a fixed scope: a stranger could say what turns up and what they're left with, beyond the name of the trade and the quote itself. In the ads we looked through, that looked like a roof inspection with photos of the damage, another included with the quote, temporary fence hire that takes in delivery, set-up and pick-up, and a compliance certificate after a commercial kitchen clean.
A stated time: a number or a named day for the job or the reply. An estimate inside an hour counted, and so did a concrete quote back within two days. "Fast turnaround" didn't.
A guarantee with terms: a 10 year workmanship warranty counted. "Satisfaction guaranteed" didn't, and neither did "fully licensed and insured".
Three things that sound like offers didn't count. "Supply and install" is the name of the job. A list of services tells me what you do, not what I get. And a story about a job you did for someone else can be a great ad, but it doesn't promise the reader anything, so it counted for the swap test further down and not for this.
A discount didn't count either. It got its own code and sat below the 18, because a lower price doesn't say what turns up. An ad with a discount and a warranty counted for the warranty. Only 1 of the 50 had a deal as its best offer, and nothing here is an argument for discounting. Leading on price has its own problem, which we wrote about in advertising for price shoppers. An ad with no ask wasn't marked down at all. Some ads are there so people know your name, and that's a fair job for an ad.
The customer often gets a trade name, not an offer
What a customer sees: a service and an invitation to get in touch. Lines like "for all your roofing needs", then a phone number or a button.
What we found: 18 of 50 reached a concrete offer in at least one ad. 18 named the service and stopped, 10 had a free quote or inspection and nothing more, 3 had no asking ad and 1 had a deal. If you only count a named thing or a fixed scope, leaving out times and guarantees, it's 14 of 50.
A free quote is a good thing to offer, and we'd keep it. It's also what the next business is offering, so it doesn't answer the question on its own.
What to fix: write one sentence about what happens after the customer responds. Name the scope, the deliverable, the response time or the guarantee terms. Here's an example we've made up for a roofer: "Free roof report with photos of every fault, back to you within 48 hours." Only promise what you can keep in your busiest week.
The first ad does a little worse than the best ad
What a customer sees: one ad at a time. The Library's order isn't necessarily the order anybody meets them in, so "first ad" here means first shown in the Library.
What we found: 15 businesses had a concrete offer in the first ad shown, and 18 had one somewhere in their first 10. Across all 162 ads, 67 were concrete, 44 named the service and stopped, 38 were a free quote or inspection and nothing more, 10 didn't ask for anything and 3 were deals. Those 67 are unevenly spread, because a few businesses run the same offer across many ads.
What to fix: go through your whole active set, not your favourite ad. Some of these businesses had a good offer in one ad and left it out of the others. If you've got an offer you stand behind, it can go in every ad that asks.
Each ad sits on the best rung it reached. A few businesses run the same offer across many ads, so this isn't a share of businesses.
Asking is common. Explaining the exchange isn't
What a customer sees: a call, quote, booking or contact request in almost every advertiser's set.
What we found: 47 of 50 businesses had at least one asking ad, 34 of 50 used the word quote or estimate in one, and 152 of the 162 ads asked for an enquiry. 18 businesses said what the customer would get.
So nobody needs more asking. The ask is well covered.
What to fix: keep the route to contact and put the value next to it. Don't write another version of "contact us". Write "contact us for this".
Most ads could run under another logo, though half the businesses have one that couldn't
What a customer sees: claims that could usually sit under another business's logo and stay true, and the odd ad stating a fact only that business could state.
The test is whether the words would turn untrue if another business in the same trade and town ran them. An ad passes if it states a team member by name, a specific job with what and where, a review count or rating, years trading, or a named accreditation with the body named.
What we found: 92 of 162 ads failed it. 25 of 50 businesses had at least one ad that passed, so this one splits down the middle. Only 10 had one ad that was both a concrete offer and a pass.
Most of the passes rest on a star rating, a person's name or years of experience, not a real job. Naming a place didn't get an ad over the line, because every competitor in that town can say the same thing. 28 of 50 named a place in at least one ad.
What to fix: add a fact that belongs to your business. A named team member, a specific completed job, a review figure, your years trading or a named accreditation. Then put it in the same ad as the offer, which is what only 10 of the 50 had done.
The gap turned up in every search
What a customer sees: uneven offer detail across trades, with no one search supplying most of the passes.
What we found: at least one concrete offer turned up for 6 of 10 businesses in the painter search, 4 of 10 in roofing, 3 of 10 in fencing, 3 of 10 in commercial cleaning and 2 of 10 in concreting. No search supplied more than half of the 18.
Don't read that as painters beating concreters. Each search is 10 businesses, in an order Meta doesn't explain, and that's nowhere near enough to rank a trade.
What to fix: use this as a check on your ad, not a league table. Your competitors' ads are in the same Library, so read the ones in your trade and town and see what they're saying you'd get.
Fix the ad first, then the first screen
What a customer sees: where there was a website to go to, its first screen usually repeated at least the service, while the ad itself often had no concrete offer in it.
What we found: we could read 18 external website first screens from the ads, and 17 repeated the first ad's offer or at least its service. That's supporting context and no more, because it only covers the businesses with a website to open. We also checked Google's Ads Transparency Center by domain only. It showed creatives for 11 businesses, 9 of them shown in the 30 days before, and we read 17 text creatives. That's too few to say anything about Google ads, and none of it feeds the 18.
What to fix: make the exchange clear in the ad before you rely on the next click to do the work. Then repeat the same promise on the first screen, so the person who tapped can see they're in the right place.
How the 50 were picked, and what got read
The rules were written down before any search ran. Five searches in Meta's Ad Library, not signed in, set to Australia, all ad types, active ads only: roofing, concreting, fencing, painter and commercial cleaning. Each results list was saved before we opened a single advertiser. We walked each list in order and took the first 10 Australian businesses running their own ads for their own trade, deciding from the search card alone. For commercial cleaning, the card had to sell to a business buyer.
329 cards were walked to get to the 50. The 279 skipped were 109 suppliers or retailers, 70 outside the trade, 47 businesses from our earlier samples, 18 training providers, 13 marketing, software or lead-selling businesses, 11 not shown to be Australian, 6 recruitment ads and 5 franchises. Current Echo clients would have been skipped too, and none came up.
For each business we read the first 10 distinct ads in the order the Library showed them. Repeats and job ads were left out, and most businesses had fewer than 10 running, which is how it came to 162. 21 of the 50 had one customer ad to read, and 5 had the full 10, so a business with more ads had more chances to reach the 18. An ad's words included its headline and description, its first carousel card, and any artwork readable in the saved capture. Words on a shirt or a ute didn't count. Video was coded from the ad's text and the saved first frame, so nothing said or shown later in a video counted. No form, chat or call was started.
A first pass coded the ads with text patterns and individual corrections. A second pass then read all 162 one at a time with the saved images, because that first pass had missed headlines, words on artwork and everyday lines like "over 20 years". That review changed 97 ad-level codes on 27 businesses and ruled on close calls the written rules hadn't settled, like when a list of services counts as a fixed scope and whether years of experience count as years trading. The swap test moved the most, and every figure here comes from after that review. Ten businesses had also been coded blind a second time, and against the final codes that pass differs on 1 of 10 for the headline and 1 of 10 for the swap test.
What this can't tell you
It can't tell you which ads worked. The Library doesn't show results or who an ad was aimed at, so we can't say an offer caused anything, and we're not saying that. This is what active ads said on one captured morning, and Meta doesn't explain the order it lists them in. It isn't a census of trade advertising.
The 18 rests on some close calls. Other fair readings of those calls run from 16 to 22, all of them under half. It'd be 16 if a "same-day service available" line and one crew's scope both fell, and 22 if finished jobs and supply-and-install counted as offers, which we've said they don't. The 14 above is the narrower measure, named things and fixed scopes only.
11 of the 50 only had video ads among those we read, and anything said or shown after the first frame is outside the evidence, so treat their counts as a floor. One graphic was too small to read in the saved capture and might have added a yes. One of the 18 could be read as a supplier, not an installer. Five showed no Australian place in their ads, so we checked their public pages, and all five are Australian.
How much weight to put on this
Treat this as an honest, informal read. It isn't peer-reviewed research and nobody outside Echo has checked it. We've tried to show the working the way a journal paper would, so you can decide for yourself how far to trust it. We don't publish the business names or the raw sheet, so you can't re-mark it yourself. It's 50 businesses, picked by a written rule from five Ad Library searches, and it's one morning's ads, 2 November 2026. They aren't a random sample. They're the first that qualified, in an order Meta doesn't explain, so we wouldn't stretch these numbers to every Australian trade.
You should also know how the judging worked. All of it was done inside Echo, to rules written before the first search, and the second pass made more rulings on close calls after seeing the first pass's codes. Those rulings changed which businesses made the 18. A first pass coded the ads, a second went through every ad against the saved images, and a separate check recomputed every figure from the data and checked the wording. The second pass and that check used the same tool, so they aren't fully independent of each other. A few of the calls are judgement, too: the headline is 18 of 50, and other fair readings run from 16 to 22.
It can't show which ads brought in work. An ad that says what you get might do no better than one that doesn't, and nothing here can settle that. The fixes we've suggested are our view of what should help, not something this data proves.
And we've got a stake in the answer. Echo sells marketing to businesses like these, ads included, and there are invitations to talk to us in this piece. A finding that trade ads could say more suits us, so weigh it with that in mind. We also wrote the rules for what counts as saying what you'd get, and they reflect how we think an ad should work. Someone with a different view of good advertising could set different bars and get a different number. The examples pictured are our own clients, chosen by us and not drawn from the sample.
It's also a look from the outside at one moment. We can't see what these businesses say on the phone, what their ads said last month or how busy they are. Some of them might have all the work they want, and an ad that says very little might be doing its job fine.
The check you can run tonight
You don't need anyone's help for this. Ten minutes and a pen will do it.
- Open your ads
Go to Meta's Ad Library, set it to Australia and search your page name. Open your last three active ads, or the first three it shows.
- Circle what the customer gets
Circle the exact words that say what a customer gets after responding. A named thing, a time with a number in it, or a guarantee with terms. 18 of 50 had those words in at least one ad.
- Cross out what doesn't count
"Supply and install", a list of services, "quality workmanship". If a free quote is all that's left, you're with the 10 of 50 who stopped there.
- Cover your logo
Could a competitor in the same trade and town run every remaining word unchanged? 92 of 162 ads could. A town name doesn't get you out of it.
- Check the whole set
If one ad has the offer and the others don't, copy it across. Then see whether any one ad has the offer and a fact about you. 10 of 50 did.
- Tap through
Follow the ad to where it lands. Does the first screen repeat the promise, or does the customer have to take it on trust?
One more thing
If you can't circle an offer or point to a fact that belongs to you, that's the next edit. It's one sentence, and it doesn't need a new campaign, a new photo or a lower price.
Most of the 50 have already done the harder part, which is getting in front of people and asking. Telling them what's on the other side is the cheap part, and it's what someone choosing between you and two others is trying to work out. Nobody rings to say they went with the other business because its ad told them more, which is why we wrote the job you didn't get asked to quote.
What should a trade ad offer besides a free quote?
One thing the customer gets after they respond: a named deliverable or fixed scope, a stated time for the job or the reply, or a guarantee with terms. Of 50 Australian trade businesses whose active Meta ads we looked through, 18 had at least one asking ad that did this, out of up to 10 ads each.
Is a free quote a good enough offer in a Facebook ad?
Keep it, but don't stop there. It's what the next business is offering too, so it doesn't tell someone why they'd pick you. Of the 50 trade businesses we looked through, 10 had a free quote or inspection and nothing more as their best offer, and 34 used the word quote or estimate in an asking ad.
Do I need a discount to have a good offer?
No. A discount says what it costs, not what turns up, so we coded deals separately and left them out of the count of ads that said what you'd get. Only 1 of the 50 businesses had a deal as its best offer. A named deliverable, a time you can keep or a warranty with a term all work without touching your price.
How can I tell if a competitor could run my ad?
Cover your logo and ask whether any remaining word would be untrue under another business in your trade and town. A team member by name, a specific job with what and where, a review figure, years trading or a named accreditation will pass. A town name alone won't. 92 of the 162 ads we looked through failed this, and 25 of the 50 businesses had at least one ad that passed.
How were the 50 businesses chosen?
From five Meta Ad Library searches on 2 November 2026: roofing, concreting, fencing, painter and commercial cleaning, set to Australia, active ads only and not signed in. We took the first 10 Australian businesses per search running their own ads for their own trade, by a rule written beforehand. Suppliers, franchises, recruitment and training ads were skipped and logged. No business in the sample is named here.
