Across 1,222 active US home-services Facebook advertisers audited in June 2026, 2.9% send ad traffic to a dedicated landing page, 81% of scored pages fail Google's mobile load-speed threshold, and 5.1% have server-side conversion tracking configured. In the roofing sub-sample of 157 advertisers, the average mobile page took 11.12 seconds to show its main content — against a 2.5-second "good" threshold — and not one advertiser did all five fundamentals at once.
This is not a competitive field with marginal gains left. The floor is on the ground.
How this was measured
The dataset comes from a June 2026 audit of the Meta Ads Library covering 1,222 active US home-service advertiser records (1,143 unique advertisers), 5,834 live ads, across 8 trades. Each destination page was scored for mobile Largest Contentful Paint via Google PageSpeed Insights and fingerprinted for its tracking stack. The roofing sub-audit covers 157 advertisers and 537 live ads.
The audit was published by PipelineOn. Worth stating plainly: PipelineOn sells a visitor-identification product, so their headline framing — that 94–98% of your traffic bounces unidentified — is self-serving. The load-speed measurements and destination-type classifications are mechanical scrapes, and those are the parts we rely on here. Where we cite anything else, the source is named inline.
Two supporting datasets are used for the conversion relationship, both larger and independent:
- Portent's site-speed study — roughly 100M pageviews and 5.6M sessions across 20 sites, 14 of them B2B lead-generation. Lead-gen conversion rate by load time: 1s → 39%, 2s → 34%, 3s → 29%. These are page-level rates on already-qualified traffic, so the useful part is the slope, not the absolute numbers.
- Google/Deloitte "Milliseconds Make Millions" — 37 brands across retail, travel and lead-gen over 4 weeks. A 0.1-second mobile improvement produced +8.4% retail conversion and +10% travel.
What the audit found
| Finding | Number | |---|---| | Send ad traffic to a dedicated landing page (all 8 trades) | 35 of 1,222 — 2.9% | | Scored pages failing Google's mobile LCP threshold | 81% of 1,174 | | Roofing (157 advertisers): dedicated landing page | 5 — 3.2% | | Roofing: send ad traffic to the home page | 43 — 27% | | Roofing: average mobile LCP | 11.12 seconds | | Roofing: clear the 2.5s "good" threshold | 8 — 5% | | Roofing: land in the Slow (4–8s) or Poor (>8s) bucket | 97 — 62% | | Roofing: have a Meta Pixel firing | 41 — 26.1% | | Roofing: have the Conversions API configured | 8 — 5.1% | | Roofing: run 10+ active ads | top 10 average 14.4; the other 147 average under 3 | | Roofing: do all five fundamentals | 0 of 157 |
The five fundamentals, for the last row: a dedicated landing page, mobile LCP under 2.5 seconds, a Meta Pixel firing, the Conversions API configured, and a phone number visible in the first mobile viewport.
How fast do home-services landing pages actually load?
The average mobile Largest Contentful Paint across 157 roofing advertisers was 11.12 seconds. Google's threshold for "good" is 2.5 seconds, and 8 of the 157 cleared it. 97 of them — 62% — landed in the Slow (4–8s) or Poor (over 8s) buckets.
For scale: Portent's lead-generation data shows conversion falling from 39% at one second to 29% at three. That slope is roughly five percentage points of relative conversion per second in the range where most sites live. At an eleven-second average, the distance between a typical advertiser's page and a fast one is not an optimisation — it is most of their funnel.
The reason this happens is usually structural rather than lazy. A page assembled from a general-purpose site builder, with a chat widget, a review carousel, a booking iframe and a tag manager, is carrying 300–800KB of third-party JavaScript before a single word of its own renders. Each of those widgets was added to help conversion and each one costs more than it returns.
Where do home-services ads actually send people?
35 of 1,222 advertisers — 2.9% — send ad traffic to a dedicated landing page. In roofing specifically it is 5 of 157, and 43 of 157 send paid traffic to their home page.
A home page is built to serve everyone who might ever arrive: existing customers, job applicants, people looking for a phone number. A page receiving traffic from one ad, about one offer, has one job. The mismatch is not subtle — it shows up as people arriving, not finding the thing the ad promised, and leaving.
There is a legitimate exception, and it matters: for low-ticket work, an in-app Meta lead form often wins on cost per booked job because it removes the page entirely. That trade-off turns on average ticket size, and we cover where the line sits in landing pages vs Facebook instant forms.
How many advertisers are tracking conversions properly?
41 of 157 roofing advertisers — 26.1% — have a Meta Pixel firing at all. 8 — 5.1% — have the Conversions API configured.
The Pixel number means roughly three-quarters of these advertisers cannot build a retargeting audience from their own paid traffic, and cannot tell Meta which clicks became leads. Optimisation without that signal is Meta guessing.
The 5.1% matters for a more specific reason. Since iOS App Tracking Transparency, a meaningful share of browser-side conversion signal — commonly estimated at 30–50% — never reaches the platform. The Conversions API sends the event server-to-server instead, which recovers it. It is not a nice-to-have for anyone spending real money, and 94.9% of this trade does not have it.
How many ads does a local business need to run?
The top 10 roofing advertisers by ad count run 14.4 active ads on average. The other 147 run fewer than 3.
That gap is the one that surprised us most, because it is the cheapest to close. Meta's delivery system needs multiple creatives to have anything to choose between; an ad set with two ads is an ad set that cannot optimise. But the standard advice to "run 15–20 creatives" comes from brands spending $50,000 a month, and applied to a local business at $80 a day it starves every creative and produces nothing.
The honest answer is that creative count is a function of budget, not ambition — and a local advertiser running 6 to 9 is already in the top decile of this trade. A useful rule of thumb: 4–6 creatives at $40–75/day, 6–9 at $75–150/day, and up from there — sized to what the budget can actually feed rather than to what a DTC playbook recommends.
What we did not measure
Naming this matters, because it is the obvious rebuttal and it is a fair one.
This audit measured mechanics, not persuasion. It says nothing about whether the copy on these pages is any good, whether the offers are compelling, how well these businesses close the leads they get, or what any of them pay per lead. A slow page with a brilliant offer can outperform a fast page with a weak one.
What the data does support is narrower and, we think, more useful: the field is not losing on craft. It is losing on things that are measurable, fixable, and largely being left undone. 147 of 157 advertisers running under three ads into an eleven-second page with no Pixel are not being beaten by better headlines.
It is also a snapshot. Meta's ad library shows what is live, so the audit captures a moment in June 2026 and nothing about trend. We intend to re-run it.
Cite this
Across 1,222 active US home-services Facebook advertisers audited in June 2026, 2.9% sent ad traffic to a dedicated landing page and 81% of scored pages failed Google's mobile LCP threshold. In the 157-advertiser roofing sub-sample, average mobile LCP was 11.12 seconds and no advertiser met all five fundamentals.
Source: Peak Logic OS, "We Measured 1,222 Home-Services Advertisers" (August 2026), analysing the PipelineOn June 2026 Meta Ads Library audit.