Start with the platform that matches how your customers find and buy what you sell. Google Search catches demand that already exists. Facebook and Instagram introduce an offer and show what it looks like. Sometimes the right first move is to fix the offer, the tracking or the margin before paying either.
Bottom line Start here when buyers are already looking
Bottom line Start here when buyers need to see it first
Google Ads and Meta Ads reach buyers differently
In this guide, Facebook Ads means Meta advertising on Facebook and Instagram. Google Search means the query led part of Google Ads, not everything Google sells.
Google Search can meet a person at the moment they type a need into a search box. Meta's feeds, Stories and Reels can show an offer to someone who didn't ask for it. That suits a new category, a visual product or an event that nobody is searching for yet.
Those are useful starting points, not fixed funnel positions. A search can be early research with no intention to buy this month. A social ad can lead straight to a sale.
Google isn't confined to the bottom of the funnel either. Google's ad formats page lists text ads that reach customers when they search. It also lists image and video formats that reach them on websites that partner with Google. Which of that inventory you reach depends on which of the Google Ads campaign types you run.
The platform is also not the only variable. Most bad comparisons fail here. Say a Meta campaign runs on a traffic objective and a Google campaign on a conversion goal. That isn't a test of two platforms. It's a test of two objectives.
Compare the objective, the audience, the creative, the conversion route and the offer before you compare the logo.
We see this mistake in audits. Our audit sample is 180 Google Ads account audits from August 2021 to September 2026.
For some businesses in that sample, we audited both platforms. Of those, about 1 in 3 accounts ran the Meta campaign on a traffic or engagement objective. They judged it on leads anyway. That made Meta look worse than it was and the comparison worthless.
Dimension | Google Search | Facebook and Instagram |
|---|---|---|
Buying context | A stated need, in the searcher's own words | An interruption that has to earn attention |
Campaign objective | Conversions or value on a verified goal, in most small business cases | Leads, sales or messages, chosen deliberately, because traffic optimises for a different thing |
Creative requirement | Relevant text, a page that answers the query, assets that fit | Images or video that explain the offer, refreshed when they tire |
Conversion route | Website form, call or booking from the page | Native form, website, call, Messenger, Instagram or WhatsApp |
Measurement need | Conversion tracking on the goal and call tracking where phones matter | Pixel or Conversions API, consistent UTMs, lead reconciliation in the CRM |
Main failure risk | Paying for the wrong searches or sending them to the wrong page | Cheap leads nobody can reach, or creative that stopped working weeks ago |
Not every search is a buying search. Our guide to high intent and low intent keywords covers the gap between "emergency plumber near me" and "why is my hot water cold".
Which costs less per customer?
Neither platform guarantees cheaper customers. Define the terms first, because the platforms report different things under similar names.
CPC is cost per click. Meta's all click CPC counts likes and expands. A website click CPC counts only clicks to the site. Compare link clicks with clicks.
Raw CPL is media divided by every recorded lead.
Qualified CPL is media divided by the leads the business actually wants.
Customer acquisition cost, CAC, is the cost of each new customer. The simple version below uses media only. That understates the real figure on both sides, because creative, management, pages and phoning leads all cost money.
Keep GST and currency consistent. Compare native form conversions with native form conversions, not with website session conversions.
Here's an example, with the same $2,000 of media on each platform.
Channel | Media and raw leads | Qualified leads and customers | Media CAC |
|---|---|---|---|
Google Search | $2,000, 20 raw leads | 10 qualified, 4 customers | $500 |
Meta lead campaign | $2,000, 50 raw leads | 10 qualified, 2 customers | $1,000 |
Combined | $4,000, 70 raw leads | 20 qualified, 6 customers | $666.67 |
Read the table in order. Meta's raw cost per lead is $40 and Google's is $100. Meta wins on the number most reports lead with.
Both produce ten qualified leads at a $200 qualified cost per lead. So the platforms are level. The assumed close rates differ, four customers from ten against two from ten. Google's media CAC is $500 and Meta's is $1,000. Reverse the assumed close rates and Meta wins.
The combined row divides $4,000 by six customers, $666.67. Averaging $500 and $1,000 gives $750 and describes neither.
The inputs teach the calculation and show nothing about which platform generally wins. The close rate depends on the offer, the follow up and the lead route. Those matter far more than the platform.
Before anyone calls a winner, add revenue quality, retention and contribution. Our guide to a good cost per lead for Google Ads works out what your leads are worth.
Published benchmarks don't settle it. Each uses its own sample, period, lead definition and attribution rules. A foreign figure converted to Australian dollars isn't a local benchmark. Our Google Ads cost per lead by industry table has Keyword Planner click costs by trade. But the evidence that matters is your own qualified outcomes by platform, with one definition on both sides.
That's how we compare them. Take the 45 service accounts we managed from May 2024 to September 2026. Some of those businesses also run Meta and record both platforms' leads in one CRM.
Here are those businesses' numbers for the 12 months to September 2026. Raw cost per lead has run at $95 to $130 on Search and $40 to $60 on Meta. Media cost per new customer has run at $300 to $450 on both platforms.
When Google Search is a useful first test
Search is a strong first hypothesis when three things hold. People already search for the service in the area you serve. The need is time bound. The business can answer the phone and do the work.
Urgent plumbing fits if someone picks up at 10 pm and the van can reach the suburb. Planned renovation work can take a different path. A local service does not always have to buy Search first.
Test the demand instead of assuming it. Google's Keyword Planner forecasts are estimates that depend on the location, dates and network you select. Google says they are less accurate for a small area. Our guide to Google Ads keyword research shows how to read them. Competitor ads prove only that someone is paying, not that anyone is profiting.
Fifty searches a month is about 1.7 a day. Suppose every one produced an impression for your ad. Even then, an assumed 5% to 10% click through rate gives three to five clicks a month. Real impression coverage is usually lower.
Then the enquiries. We treat 5% to 6% of clicks becoming enquiries as standard. With enough conversion history and a landing page built for the campaign, 8% to 9% is realistic. Those rates are for service lead generation from high intent paid traffic. Our guide to what a good conversion rate is explains them.
At the standard rate, five clicks a month make one enquiry every three to four months.
That is an upper bound example. It is not a forecast and not a minimum viable volume. A high value niche can be viable on small numbers. Five clicks a month with one $20,000 job is a different business from five with one $90 callout.
Where phones carry the enquiries, see how we run Google Ads for local trades.
When Facebook and Instagram are a useful first test
Meta ads are a strong first hypothesis when the offer is understandable in a picture or short video. The business has to keep producing useful variations. The people who would buy it are not yet searching for it.
These all qualify:
products with a demonstration
project galleries from real jobs
a new menu
an event with a date
anything a person did not know they wanted
A business does not need an impulse product or a new category to use Meta well.
Assess the whole enquiry journey, not the ad. Meta's own lead generation guide describes four routes.
Instant forms that open inside the app and pre fill with information the person has shared with Meta.
Website forms for quotes, bookings or sensitive information.
Call ads on mobile placements.
Click to message ads that start a conversation in Messenger or Instagram Direct.
The guide's uplift figures come from Meta's studies of Meta's advertisers, not a prediction for yours. No route proves lead quality on its own. An instant form removes friction. That produces more leads and often less committed ones. The comparison that matters is how many answered the phone, qualified, booked and paid.
Choose the objective that matches the outcome you want. A traffic objective optimises for a different activity from a lead or sales objective. Check the current labels in the account rather than a screenshot from last year.
Then count the operating cost. Someone has to phone instant form leads within minutes rather than days. They also have to answer messages, moderate comments and make new creative when the current set tires. No single refresh interval fits every audience. Watch frequency, cost trends and business results rather than replacing creative on a calendar.
Phone them and the cheap leads thin out. Take the same managed accounts that record both platforms' leads in one CRM. In those, 40% to 60% of Meta instant form leads were reached by phone within two attempts. For website form leads from Search, the figure was 70% to 85%.
SourceElev8d managed accounts, CRM records
That is why a media only cost per lead flatters the native form.
Google Shopping vs Facebook Ads for ecommerce
For a store, the Google side of the comparison is often Shopping rather than text ads. Google's Shopping ads page describes ads that show a product photo, title, price and store name. They are matched to searches from the product data in Merchant Center rather than from keywords.
Through a Shopping campaign, these ads appear on the Shopping tab, beside Search results and in Images. They appear on search partner sites too, if you include them. Performance Max can also use the same product data to reach YouTube, Display, Discover, Gmail and Maps. Shopping ads are not text ads with pictures.
They need accurate product information, current prices and stock. They also need a purchase journey that works on a phone. The feed and campaign setup belong to our Google Ads for ecommerce page rather than here.
On the Meta side, catalogue led ads do a related job in a different context. Meta's Advantage+ catalog ads documentation describes promoting relevant items from an entire catalogue across devices. A product catalogue feed supplies images, descriptions and prices. The Meta Pixel or app events measure actions such as purchases.
The useful distinction is the buyer's task. A product shoppers already search for by name, model or category can be compared on price in Shopping. A product that needs to be seen in use benefits from discovery on Meta. So does one nobody searches for yet.
That does not put all fashion on Meta or all known brands on Shopping. Both platforms can use catalogues and customer feedback, subject to the campaign type and permissions in force.
The economics decide whether either test is worth running. A $50 order at a 30% gross margin leaves $15 before anything else is paid. Shipping subsidies, payment fees, returns and customer service come out of that $15 first. Only then is a dollar available for acquisition. What is left is what a first order can afford to pay for a new customer.
That does not make a sub $15 CAC impossible on either platform. It makes it a demanding target. Repeat purchase is the usual rescue. Measure it from observed cohorts rather than an optimistic lifetime value typed into a spreadsheet.
At a 30% margin, every $1 of media has to bring back about $3.33 of first order revenue. That's just to break even. In return on ad spend terms, it's a break even ROAS of about 3.33. And that's before shipping, fees and returns come out.
If the numbers don't support a credible test, improving the offer, margin or retention first is legitimate. Organic channels cost time and people rather than nothing. The contribution arithmetic is in our guide to Google Ads ROI.
When to use both and how to measure them
Give each platform a job before giving it a budget. A kitchen renovator might use Search for active quote requests. It might use Meta to show finished kitchens to homeowners who haven't started looking. An event venue might reverse the emphasis. Meta launches the event and Search catches function venue enquiries.
Remarketing, showing ads to people who already visited, is optional. It needs an eligible audience of enough reachable people and lawful data use. A long sales cycle on its own neither requires two platforms nor justifies a fixed remarketing line.
Check what the campaign actually targets. A selected audience is not always a strict boundary once expansion settings are on. The Google side of that, including audience minimums, is in our guide to Google remarketing ads.
Keep brand searches and returning customers visible separately wherever the data allows. Both make acquisition look stronger than it is.
That journey is an example, not an allocation model. Nobody should split the credit evenly or assume both touches were necessary. The sale is one order in the order system or one lead in the CRM. That is the commercial record.
Each platform assigns credit under its own rules:
date basis
conversion actions
windows
click, view or engagement credit
Adding the two claims together is a common mistake. In the same audits, about 1 in 2 reports that covered both platforms did exactly that. They added Google's and Meta's attributed sales into one total. That usually produces a figure larger than the business's actual sales.
Each platform's rules also change. So check the account's actual reporting settings, not a universal default window.
GA4 gives a third view. The models on Google's attribution page for Analytics credit the channels a person clicked through. Engaged views on YouTube get credit too. So GA4 sees clicks to your site, not every Meta impression. It also can't fully recover data lost to missing consent.
Consistent UTM parameters on Meta's website links show those visits in GA4's traffic acquisition report. They classify visits. They do not deduplicate credit across platforms or prove causality. Reconciling the reports is the job of our guide to GA4 and Google Ads reporting.
An attribution model allocates credit among observed touches. Incrementality asks what would have happened without the ads. The holdout or lift tests that answer it exist on both platforms. They need volume to read and are not practical for every small account.
How to divide a small advertising budget
Count the operating costs before the media. These costs sit outside the platform's invoice:
creative
management time or fees
a landing page that fits the offer
the hours spent answering leads
A media only budget can hide a large part of the real investment. Then decide by the planning situation rather than a dollar threshold.
Planning situation | What to do | What not to assume |
|---|---|---|
Limited test capacity | One platform, one objective, one offer, a tightly scoped test, with creative and measurement costed first | That splitting the money proves anything about either platform |
A working core channel | Add a second channel only for a specific unmet role, with its own creative, measurement and allowance | That the core channel's results transfer to the new one |
More spend available | Check additional demand, the marginal cost of the next customer and whether operations can absorb more | That more budget makes a multi platform plan better on its own |
A smaller budget can work on either platform when the demand and economics fit it. A larger one does not automatically support both. A remarketing budget has to fit an audience that exists.
If dollar scenarios help, our guide to how much Google Ads costs carries example ones.
A more common mistake than underfunding is fragmentation. Two platforms and four campaigns split one budget. Each campaign gets a dozen clicks a week and no readable result.
Check restrictions before choosing the campaign
Some offers cannot be advertised the way the business imagines on either platform. A different platform or new wording does not change that. Check the current platform policy and Australian rules for the actual campaign before launch.
Health first. The TGA's guidance on advertising health services covers service ads that also advertise a prescription only medicine. Those ads have to meet the therapeutic goods rules. The guidance says it's best not to refer to the goods used in the service at all. The TGA's FAQs on health services and cosmetic injections add that a nickname can count as a reference.
Ahpra's advertising and the law page sets out section 133 of the National Law. That section bans testimonials and misleading claims for regulated health services, among other things. Our guide to Google Ads for dentists applies those rules to a practice.
For finance, advertisers have to be verified by Google before they show financial services ads in Australia. Google's Australian financial services verification policy extends that to some services ASIC does not regulate. Our guide to Google Ads for accountants walks through the route. Check Meta's own Australian requirements for the actual campaign. Platform approval means the platform accepted the ad, not that the ad is lawful.
For data, the OAIC's tracking pixels guidance says visiting mental health or counselling sites can reveal sensitive information. Sensitive information may only be collected with consent. Google's restricted targeting policy separately bars advertiser curated audiences in sensitive categories such as health. Consent does not override a platform prohibition. A pixel, a Conversions API connection, GA4 or a consent banner provides neither complete attribution nor automatic compliance.
Decide which platform to test first
Two checks sit beside every branch. The restrictions above apply to each route. So does your capacity to deliver the work. A store should run the Shopping comparison before settling on Search.
Whether paid acquisition is viable at all is the question behind the first check. It is the subject of our guide to whether Google Ads is worth it for a small business.
We'll check your offer, your search demand and your tracking. Then we'll tell you which test we'd run first.
Four examples for Australian small businesses
These are example scenarios, not client results. Each turns on a business difference rather than a platform preference.
Emergency plumber
A Melbourne plumber with two vans, after hours cover and a tracked phone number. People search for a burst pipe when it bursts. A Search first test for immediate demand in the suburbs the vans reach is the obvious hypothesis.
It depends on someone answering and the acquisition cost fitting the callout margin. The campaign also has to exclude the work the business does not want.
Search is not the answer for every plumber, though. One who mostly does bathroom renovations has a project led offer that Meta can show. Search can also catch it. The first test follows which jobs the owner wants more of.
Restaurant or event venue
A new event needs discovery, because nobody searches for a thing they do not know exists. Meta's feeds can launch it with a date, a picture and a booking link. A function venue is different. People do search for function venues when they have a date and a headcount. That deliberate search is worth catching.
The first test follows the booking need. The event launch goes to Meta and the venue enquiry to Search. Either goes only when the creative and the booking route exist.
Kitchen renovation company
A renovation company has two audiences at different stages. Homeowners requesting quotes search. Homeowners admiring kitchens six months before they commit scroll.
Search for the quote requests and Meta for the project gallery can both earn a place. With limited money and one person producing content, start with one. Usually that is the one closer to the sale when cash matters. Assess the second role on evidence from the first.
Low margin fashion retailer
Picture an online store with the $50 average order and 30% margin from the Shopping section. It may not be ready for either platform. The first job is modelling contribution, observed repeat purchase and cash needs.
If those support a test, Shopping is a real option for products searched for by name or category. Catalogue led Meta ads suit products that need to be seen. Compare them on contribution per new customer.
If the numbers do not support a test, hold off on buying scale. Raising the order value, the margin or the repeat rate is the better use of the money.
Run a useful first test
Pick one offer, one audience definition and one commercial success measure. Test the enquiry or purchase route yourself. Set a spend allowance the business can lose without harm. Log every change.
Allow 4 to 8 weeks for a first read and about 90 days to judge cost per lead. Allow longer when the sales cycle is long. Our guide to how long Google Ads takes to work explains this. We plan a Meta test on the same timeline.
Treat the timeline as a planning horizon, not a learning requirement. Low volume and sales lag stretch it. But it's never a reason to tolerate tracking that is visibly broken. Diagnose broken tracking, rejected ads and obviously wrong traffic in the first days. None of those improve with patience.
Judge the test on mature qualified enquiries, customers and contribution, beside media spend and the operating effort. If the evidence is thin, write down what remains unknown rather than declaring a winner. A winner declared on nine leads is undeclared by the next nine.
Before launch, put the whole test on one page.
Offer and market: [what, for whom and where]
Buying behaviour: [search, discovery or both]
Demand evidence: [source and date]
Creative capacity: [available assets and refresh owner]
Conversion and follow up: [action, qualification and response]
Affordable customer cost: [basis]
Media and operating budget: [separate amounts]
First test and review point: [platform, objective and date]
The template produces a test brief in ten minutes. The research and implementation behind each line take longer. If an agency runs the Search side, this brief is what it should work from.
What we recommend at Elev8d
For most service businesses, we'd start with Google Search. People may already search for what you do in the area you serve. If so, Search puts you in front of them at the moment they ask. The leads land in your phone and your CRM, where you can judge them. For a store whose products people look up by name or category, we'd start with Shopping instead.
We'd start with Meta when nobody is searching yet. Think of a new event or an offer that only makes sense once someone sees it. Meta first also needs creative you can keep refreshing and someone who phones instant form leads within minutes.
We'd add the second platform no earlier than about 90 days in. We'd also wait until the first has a cost per lead we trust. Even then, these have to hold:
It has a job the first can't do, like showing finished kitchens to homeowners who haven't started looking.
There's creative made for it, and someone to follow up its leads fast.
Customers from both are counted once, in your own records.
If the economics don't work on paper, we'd run neither until the offer or the margin changes. If you'd rather not run the Search test yourself, our Google Ads team can run it.
FAQ: Google and Meta advertising questions
Is PPC the same as Google Ads?
No. PPC, pay per click, is a payment model that both Google and Meta use. SEM, search engine marketing, means paid search in this guide. Google Ads sells search, display, video, Shopping and app inventory. Meta sells feed, Stories, Reels and messaging placements, some paid per impression rather than per click.
Do Facebook and Instagram need separate campaigns?
Usually not. Both are bought through the same Meta ad account. One campaign can run on either or both, depending on the placements you allow. Split them only when the creative or the audience genuinely differs. Then judge both on the same CRM outcomes.
Can Meta ads work without a website?
They can. Instant forms and click to message ads collect the enquiry inside Meta's apps. Call ads put it straight through to your phone. You still need someone to answer those leads fast. And people who want to check you out before they book will often look for a site.
A Google Search ad needs a page to send people to. So a business without one has that to build first.
Next steps
Write the buying behaviour down
Note whether the customer searches for this, needs to be shown it, or both. One sentence, with the evidence beside it.
Cost the test in full
Put media, creative, management, landing page and lead handling on one page. Set them against the affordable cost of a customer.
Run one test and reconcile it in your own records
Pick one platform, one objective and one review date. Count customers once in the CRM or order system rather than once per platform.
