Broad Match Is the Most Expensive Cheap Traffic You Can Buy

The cheapest click is usually the most expensive

Most accounts we inherit have the same instinct baked into them: get the cost per click down. It is the number the platform puts in front of you, it is the number that moves when you change match type, and it is the number that feels like efficiency.

It is also the number least connected to revenue.

Here is the same budget spent two ways. Same product, same landing page, same sales team. The only difference is match type.

Illustrative model, not a benchmark. The point is the shape of the arithmetic, not the figures — run it on your own CPCs, your own landing page conversion rate and your own qualification rate.

Broad match bought three and a bit times the traffic and two more enquiries. It also produced fewer than a third as many real sales conversations, at nearly four times the cost each.

Nothing in that table is a trick. Broad match genuinely is cheaper per click. It genuinely does produce more enquiries. Both of those facts are true and both of them are irrelevant, because neither is what you are buying. You are buying conversations with people who can become clients, and on that measure the expensive traffic was cheaper by a factor of nearly four.

What broad match actually buys

Broad match does not just match your keyword. In Google's own words, broad match ads "may show on searches that are related to your keyword, which can include searches that don't contain the direct meaning of your keywords". In practice that can mean the adjacent profession, the student, the job seeker, the competitor doing research, the person who wanted something that merely sounds like what you sell. Phrase and exact are not literal either — exact match now covers searches with "the same meaning or same intent" as the keyword — but they stay far closer to the words you chose.

Source: Google Ads Help, About keyword matching options.

We can put a number on that, because we audited it. Our audit of the existing Google Ads account of MLA Traffic, a VicRoads pre-qualified traffic management company in Victoria, scored it 35 out of 100, with roughly a third of spend going to traffic that could never convert. A Performance Max campaign had turned 174,000 impressions into a single conversion. Competitor brand keywords carried quality scores of 1 and 2, the bottom of Google's 1–10 scale, and produced nothing. A meaningful slice of budget went on job-seeker intent: people searching for traffic controller courses and traffic control jobs, whom an ad for commercial traffic management cannot convert. And every ad pointed at the homepage, because there were no service pages to point them at.

That last point matters more than it looks. Match type and landing page are the same decision viewed from two ends. A tightly matched query deserves a page that answers it; a broadly matched query has no page that could. The trades account teardown goes through the whole audit.

Why the higher CPC is the point, not the cost

Phrase and exact match cost more per click for a reason that is worth understanding rather than resenting.

Google prices a click on a blend of what competitors will pay and how relevant it believes your ad is: by its own definition, you pay "only what's minimally required to clear the Ad Rank thresholds and beat the Ad Rank of the competitor immediately below you", and Ad Rank weighs your bid alongside Google's auction-time view of your ad's quality. On an exact, high-intent commercial query, the businesses that sell what you sell are bidding, and they are bidding because that query converts. The price is high because the intent is high. You are not being penalised; you are being quoted the market rate for a buyer.

Broad match is cheap for the mirror-image reason. Much of what it serves you is barely contested, because few other advertisers want it.

Source: Google Ads Help, Actual cost-per-click (CPC): Definition.

The spread inside one programme can be enormous. In a multi-year enterprise lead generation programme for Nuance, one of our foundation clients, run across search, social, display and LinkedIn, cost per lead ranged from A$14.66 in the legal vertical and A$26.17 in law enforcement to A$57.53 in government and A$82.14 in finance: a 5.6× spread. The price of a click was no guide to it. Finance paid A$1.12 a click, less than half of government's A$2.53, and produced the most expensive leads of the four. A single blended target would have hidden every one of those differences.

The incentive built into the Partner badge

There is also a structural reason accounts drift toward broad match, and it has nothing to do with the accounts.

The Google Partner badge requires the agency's registered manager account, which covers the client accounts it manages, to hold a minimum 70% optimisation score, alongside a spend threshold and certifications. Optimisation score is Google's own metric, built around Google's own recommendations — and "Add broad match keywords" sits among them, as does "Turn on AI Max".

The mechanic is the interesting part. A pending recommendation holds your score down. Applying it lifts the score. And dismissing it lifts the score too: Google's own Partner FAQ says "You'll notice a score uplift by applying and dismissing recommendations."

So an agency can hold its badge while running phrase and exact only — but every broad-match recommendation it leaves unapplied counts against its score until someone dismisses it, deliberately, account by account, and Google says a dismissed recommendation can reappear if it becomes relevant again. The agency that does nothing leaves those recommendations weighing on its score. The agency that clicks apply gets the uplift straight away, and its client's budget is opened up to broader matching.

We are not going to tell you what Google intended by any of that. We will simply note which behaviour takes one click and which is a standing chore. The badge tells you an agency meets Google's spend, certification and optimisation-score thresholds. It does not tell you whether its clients' budgets are well spent.

Sources: Google Ads Help, How to become a Google Partner or Premier Partner · Google Partners FAQs · About optimisation score · Types of recommendations. Partners are not required to adopt every recommendation or reach 100%.

The other way budget disappears: paying for words you do not own

Match type is one half. The other half is which words you bid on at all, and the most expensive mistake there is subtler.

We audited an Australian B2B software company's account recently. Their own company name cost A$0.90 a click and produced enquiries at A$74 each. The name of one of their products, bought on exact match, cost A$12.55 a click and produced zero. Fourteen times the price, same account, same 90 days.

The product name is also a common English word. When we searched it in September 2026, page one was a mattress brand, a text editor and a kind of snail, and it is hard to see the company ever ranking for it. The A$12.55 was not the price of defending a brand; it was the price of buying a word whose search results belong to other people.

Across the account, 63% of search keyword spend went on bare product names, bid on without the company name, while the category terms an actual buyer types took 19% — and others like them sat paused.

The full teardown, including the two-word fix already running in the same account and the four checks to run on your own, is in when a product name isn't a brand keyword.

Organic owns what you are called. Paid buys what you do.

That is the whole division of labour, and the wasted search budgets we audit usually have it backwards.

• What you are already winning organically, do not rent. Bid a little on your own name to hold the result against competitors, cap it hard, and move on.
• What you can realistically win organically in six to twelve months, build. Specific, narrow, answerable questions where the competition is thin. BCS Broking, a specialist commercial broker whose site we rebuilt, relaunched on 29 April 2026. Checked on 27 July, it held the first organic result for "surety bonds broker australia construction mining", and Google's AI Overview for that search named it first among the brokers it cited — earned by having the most specific pages on a narrow subject, not bought.
• What you cannot win organically in the near term, buy. Contested category terms owned by directories and incumbents are exactly what a paid budget is for. That is not a failure of your SEO; it is the correct use of media.
• What you will never own, walk away from. Generic English words that happen to be your product names belong to whoever has been using them longer.

The same arithmetic, four verticals

The same shape turns up in each vertical we have modelled. What changes is the CPC, how much junk broad match drags in, and how far apart the two qualification rates sit — and each vertical leaks in its own particular way, which is why each one gets its own teardown:

• Commercial insurance broking — the word “insurance” is shared with personal lines, claimants, students and job seekers.
• B2B SaaS — cost per lead varied 5.6× between verticals in one enterprise software programme, so one blended target hides where the money goes.
• Trades and local services — in the illustrative model, broad and phrase-plus-exact produce the same number of enquiries, and phrase-plus-exact produces nearly five times the real sales conversations; plus the inherited account that audited at 35/100.
• Financial services — categories searched by people who will never open an account, and why the measured event belongs at the funded customer, not the application.

What we actually do

1. Phrase and exact only, to start. Broad match is not banned forever, but it is not how an account learns. It is what you might add once you have enough conversion data for the bidding to have something true to optimise toward. That now includes checking AI Max for Search, whose search term matching extends your keywords with broad match and keywordless technology. From September 2026, Google is automatically upgrading Search campaigns that use automatically created assets or the campaign-level broad match setting to AI Max, with search term matching switched on. With search term matching on, a phrase-and-exact campaign is no longer phrase and exact; Google lets you switch it off ad group by ad group.
2. One campaign per thing you sell, one page per campaign. If the ad makes a specific promise, the landing page has to keep it.
3. A negative keyword list built from the search terms report, monthly. Courses, jobs, salary, free, template, DIY, and every adjacent profession. This is where much of the waste lives. The report never shows everything: Google omits search terms without enough query activity, for privacy.
4. Brand separated from generic, and capped. Otherwise brand's cheap conversions flatter the terms that are losing money. If the account runs Performance Max, that means a brand exclusion on it too, which Google recommends when brand traffic is managed in its own campaign.
5. Measure to the qualified conversation, not the form fill. Feed the qualified stage back to the platform, through offline conversion import or its upgraded form, enhanced conversions for leads, so the bidding optimises toward clients rather than enquiries.
6. Decide organic and paid together, on one screen. The question "should we buy this" cannot be answered without knowing whether you can earn it.

Run this on your own account

Four checks, about an hour:

1. Sort your keywords by match type and compare cost per qualified enquiry, not cost per click and not cost per enquiry. If you cannot produce that number, that is the finding.
2. Google every product name on its own, incognito. If page one is full of things that are not you, you are not defending a brand — you are buying a word.
3. Compare the bare version against the qualified version — product name alone versus product name plus company name. If you have both running, the answer is already in your account.
4. Add up what share of spend goes on your own names. There is no benchmark for this, but our rule of thumb is that above roughly 40–50% you have a reporting problem as well as a media problem: your numbers look good because brand always looks good.

The line worth keeping: organic owns what you are called, paid buys what you do. Get it the wrong way round and you can spend most of a search budget on names that only look like brand, while the terms your actual buyers type sit paused.