Broad Match Has Changed. Most Accounts Haven't Caught Up Yet

For most of Google Ads' history, broad match had a straightforward reputation: powerful in the right hands, dangerous in the wrong ones. It expanded your reach significantly, triggered on loosely related queries, and rewarded accounts that managed it carefully with cheap discovery volume. Accounts that didn't manage it carefully paid for a long tail of irrelevant traffic and learned their lesson the hard way. The solution was well understood — tight negative keyword management, regular search term reviews, a clear preference for phrase and exact match on anything high-stakes. Broad match was a tool you used selectively, with appropriate scepticism, and with one eye permanently on the search term report.

That playbook was built for a specific version of broad match — one where the matching algorithm operated primarily on keyword semantics and surface-level query relevance. Add "running shoes" on broad match and you'd get "trainers," "athletic footwear," "jogging shoes," and occasionally "shoes for running a business." The expansion was predictable enough to manage with a well-maintained negative list. The risks were real but bounded. Most experienced practitioners knew where the edges were.

That version of broad match no longer exists. What replaced it is more capable, more opaque, and more consequential — in both directions. Understanding the difference between the old model and the new one is not a technical detail. It's the foundation of every match type decision you make in a modern Google Ads account.

The changes to broad match didn't arrive in a single announcement — they accumulated across a series of updates between 2021 and 2024, each incrementally expanding what the algorithm could match and how it made matching decisions. Taken together, they represent a fundamental redesign of how broad match works, not an incremental tweak to the original model.

The cumulative effect of these changes is that broad match in 2024 operates more like a campaign-level intent signal than a keyword-level matching instruction. You're not telling Google "match queries that are semantically related to this keyword." You're telling Google "find queries where conversion probability is high given everything you know about this campaign, this ad, this landing page, and this audience." That's a fundamentally different instruction — and it requires a fundamentally different management approach.

The most important structural change in how broad match works is its integration with Smart Bidding — and most accounts are not fully accounting for this in how they manage the match type. The relationship is not additive. It's architectural: broad match and Smart Bidding are designed to function as a single system, with each component performing a role that only makes sense in the context of the other.

Broad match's job in the integrated system is reach — finding queries that represent genuine conversion intent across a wider surface area than phrase or exact match can cover. Smart Bidding's job is constraint — using conversion probability modelling to bid low or not at all on the broad match queries that represent low intent, and bid competitively on the ones that don't. In theory, the combination produces broad coverage with intelligent filtering. In practice, the quality of the filtering is entirely dependent on the quality of the conversion signal Smart Bidding is working from.

This is the mechanism that most explanations of the new broad match architecture underemphasise. The system works as advertised when Smart Bidding has rich, accurate conversion signal — 50 or more primary conversions per campaign per month, a clearly designated primary conversion action, and a stable account environment that allows the algorithm's model to accumulate meaningful history. When those conditions aren't met, the constraint layer of the system is operating on a weak or noisy model. Broad match's reach advantage remains; Smart Bidding's filtering advantage does not. The result is the waste profile that gives broad match its persistent bad reputation among practitioners who've been burned by it.

Google's official communication about broad match is enthusiastic and not wrong — it just describes the system at its best rather than in the conditions most accounts actually operate in. The risks are real, specific, and worth naming precisely rather than either dismissing them or generalising them into vague scepticism.

The first risk is signal-starved expansion. Accounts with fewer than 30 conversions per campaign per month using broad match are running the reach component of the system without the filtering component functioning reliably. In this dataset, these accounts averaged a 47% waste rate on broad match traffic — compared to 19% in accounts with 50 or more monthly conversions per campaign. The waste doesn't come from broad match being badly designed. It comes from using a system that requires good signal in conditions that don't provide it.

The second risk is search term opacity. Since 2020, Google has progressively reduced the visibility of search term data — queries that don't meet a privacy threshold are excluded from the search term report. In broad match campaigns, this means a meaningful share of your traffic is invisible to you. You can see that broad match is spending; you can't always see what it's spending on. This opacity makes the traditional negative keyword management approach — identify bad terms in the report, exclude them — structurally incomplete. You need a proactive negative infrastructure that blocks predictable waste patterns before the report shows them, because some of the waste will never appear in the report at all.

The third risk is competitive cannibalisation. Broad match's expanded intent inference means it can trigger on queries that should be served by a different campaign in the same account — a brand campaign, a competitor campaign, a higher-funnel awareness campaign with a different bid strategy. Without explicit campaign priority structures and negative lists managing traffic flow between campaigns, broad match can undermine your segmentation architecture from within.

Google's recommendation to switch to broad match applies more broadly than it should. It describes the optimal conditions for broad match performance without adequately specifying the conditions under which the advice is safe to follow. Here is a more honest framework.

The expansion of broad match's matching scope does not reduce the importance of negative keywords — it changes their role. In the old model, negatives were primarily reactive: you identified waste in the search term report and excluded it. In the new model, negatives serve a structural function that exists independently of what the search term report shows. They define the boundaries of the space broad match operates in. Without them, that space has no walls.

1. 1

   Account-level universal exclusions — Build a comprehensive account-level negative list before enabling broad match on any campaign. This list should cover universal irrelevance patterns: free, DIY, jobs, courses, certifications, wholesale, and any competitor or adjacent vertical terms you never want to trigger on. These run automatically across every campaign and require no per-campaign work.

2. 2

   Campaign-level intent boundaries — For each broad match campaign, define the intent space it should operate within and add negatives that prevent expansion outside it. A campaign targeting commercial intent queries should exclude informational intent signals — how, what, why, guide, definition. These aren't one-off exclusions; they're structural intent walls.

3. 3

   Cross-campaign traffic flow management — Map your campaigns by intent stage and add negatives between them to enforce clean traffic segmentation. Brand terms excluded from non-brand campaigns. Top-funnel terms excluded from bottom-funnel campaigns. This prevents broad match from collapsing your campaign architecture by routing traffic through the wrong bid strategy.

4. 4

   Invisible traffic pattern blocking — Because a meaningful share of broad match traffic won't appear in your search term report, build your negative list to block predictable waste patterns proactively rather than reactively. Single-word queries, highly generic category terms, and known low-conversion intent signals should be excluded pre-emptively — you don't need to see them in the report to know they'll appear.

The most common mistake in evaluating broad match performance is comparing it to phrase or exact match at the campaign level and concluding from a lower conversion rate that broad match is underperforming. Broad match operates on a different query set — it's reaching traffic that phrase and exact match wouldn't reach at all. The relevant question is not "does broad match convert as well as exact match?" It's "is the incremental traffic broad match reaches converting at a rate that justifies its cost — and is the waste rate within acceptable bounds?"

1. 1

   Measure incremental reach — Compare impression volume on broad match campaigns against what phrase and exact match equivalents would reach. If broad match isn't delivering meaningfully more reach, its primary value proposition is absent. Reach delta below 30% suggests your phrase and exact match coverage is already capturing most of the available intent.

2. 2

   Track waste rate independently — Export your search term report monthly and score the irrelevant traffic share as a percentage of total broad match spend. A waste rate under 20% in a well-managed account is achievable. Between 20–35% is acceptable if reach advantage is significant. Above 35% is a signal that negative infrastructure is insufficient or signal quality is too low for the system to filter effectively.

3. 3

   Monitor query diversity over time — Broad match's expansion should be producing a progressively wider and more relevant query set as Smart Bidding learns. If your search term report shows the same limited query set month after month, the algorithm may be under-exploring — often a signal that conversion volume is too low to drive confident expansion.

4. 4

   Compare CPA trajectory, not point-in-time CPA — Broad match campaigns with strong signal typically show improving CPA over the first 60–90 days as Smart Bidding's model matures. A broad match campaign that isn't showing CPA improvement over that window — despite adequate conversion volume — warrants a structural review of negative coverage and conversion signal quality.

5. 5

   Run a controlled reach expansion test before full adoption — Before switching existing phrase or exact campaigns to broad match, add broad match as an additional campaign running in parallel with a modest budget allocation. Measure waste rate, incremental reach, and CPA delta over 30 days. This gives you account-specific evidence rather than relying on Google's generalised benchmarks.

Broad match has genuinely improved. The integrated Smart Bidding architecture is more sophisticated than anything that existed in 2019, and the reach advantages in accounts with strong conversion signal are real and measurable. The mistake is not using broad match — it's using it as though the old rules still apply. The old playbook said: use broad match carefully and defensively, with maximum negative coverage and a preference for tighter match types wherever possible. The new playbook says: use broad match structurally, with the right signal quality, the right negative infrastructure, and a clear framework for measuring whether it's earning its place. Different instructions. Both versions of the same underlying principle — that reach without control is expensive, and control without reach leaves value on the table.