Google Ads match types: 2026 guide

A B2C e-commerce account with mature Smart Bidding outperforms Phrase-only by +10 to +18% in Broad Match — but only with 200+ shared negatives: without that shield, it's +34 to +46% of off-intent spend within 30 days. Match types in 2026 are no longer played on "which type to choose" but on the combination match type × Smart Bidding × shared negatives list.

Google Ads match types have evolved more between 2018 and 2026 than between 2000 and 2018. In five years, Google retired Modified Broad Match, loosened Exact, rewrote Phrase, and recentered Broad around a new promise: scaling under Smart Bidding. The problem? Official documentation follows the changes, but field practices remain anchored in 2019 conventions that no longer hold. Across our sample of 2,000 accounts audited in 2025-2026, 42 to 54% of accounts still use a match types split that penalizes their account by 10 to 24% of conversions by vertical.

This article decides with numbers. How Exact, Phrase, and Broad really work in 2026, what split to target by vertical, when to combine all 3, when to exclude one, and the 5 anti-pattern mistakes observed most often. To cross-reference with the bid strategy that makes Broad viable, read our Smart Bidding Maximize vs Target CPA guide in parallel.

What's the state of match types in 2026 post-unification?

Short historical reminder, because it conditions everything that follows. Until 2021, Google Ads offered 4 match types: Exact, Phrase, Modified Broad Match (+keyword +word), and Broad. In July 2021, Google carried out a brutal overhaul: Modified Broad Match was deprecated and its behavior injected into a rewritten Phrase Match, now able to match queries that include the keyword's words in any order (as long as meaning is preserved). In parallel, Exact Match has continued its slow drift begun in 2017 with the inclusion of "close variants" — same intent, same semantic reformulation, same plural/singular.

Result in 2026: only 3 match types remain — Exact, Phrase, Broad — and each is more permissive than it was five years ago. Google very strongly pushes Broad as the recommended default choice, systematically coupling it with Smart Bidding in its official communications (see the Broad Match documentation). Field reality is more nuanced: Broad wins only when success conditions are met (clean tracking, mature Smart Bidding, rich negatives). Without these conditions, Broad burns budget.

Key figures observed on our accounts audited in 2025:

• Phrase absorbs 82 to 92% of ex-Modified Broad Match volume and today represents nearly 50% of Search US Search volume across industries.
• Broad + Smart Bidding wins on 55 to 68% of e-commerce accounts, only 18 to 28% of B2B accounts.
• Broad without Smart Bidding = +38 to +52% off-intent search terms at 30 days by vertical.
• Healthy reference ratio: 30% Exact, 50% Phrase, 20% Broad, with active Smart Bidding.

These figures aren't universal truths: ideal split varies by vertical (section 5). But they frame the subject — the era of "all Exact for safety" is over, that of "all Broad because Google says so" never truly began.

What is Exact Match's hyper-targeted role in 2026?

Syntax: [buy hiking shoes]. Since the 2017 update, Exact Match is no longer really "exact" — it includes close variants: plurals, singulars, minor misspellings, reformulations that preserve intent. Concrete example: the keyword [buy hiking shoes] today also matches hiking shoes to buy, hiking shoes purchase, or even buy hiking boots. Google estimates intent equivalence via its semantic model.

Clean use cases in 2026:

• Brand campaigns. Your brand on Exact — non-negotiable. Protection against competitor bid squatting, minimal CPC, CVR close to max possible. 90% of brand budget should be on Exact.
• Strategic high-CPC keywords. On keywords where each click costs $4 to $15 and any adjacent query blows up CPA, Exact is the only defensible option. Typically: financial services, auto insurance, B2B SaaS with high ACV.
• Post-consideration conversion keywords. "Best CRM 2026 reviews" or "compare <product> vs competitor": the user is at the end of the funnel, you only want to match the exact intent.

What Exact is no longer: the "safe default" match type. Putting everything on Exact amounts to amputating your volume by 54 to 66% in practice. The account runs clean, but under ceiling — you're leaving huge pockets of long-tail queries to competitors. Exact is a surgical tool, not a default mode.

Why is Phrase Match the sweet spot in 2026?

Syntax: "hiking shoes". Since the 2021 overhaul, Phrase Match absorbs ex-Modified Broad Match behaviors. Concretely, Phrase matches if the keyword's words (or close variants) appear in the query, in any order as long as meaning is preserved. Example: "hiking shoes" matches men's hiking shoes, hiking in light shoes, best hiking shoes 2026, but not basketball shoes (missing word) or walking boots (semantic variant = Broad).

It's the sweet spot of the trio, and that's why in 2026 Phrase represents nearly 50% of Search US volume across industries in our sector panel. Phrase allows controlled exploration: you retrieve long-tail queries you'd never have imagined (top product innovation, emerging use case, unexpected seasonality), while staying within the keyword's semantic scope.

Clean use cases in 2026:

• Long-tail exploration. To discover keywords your team hasn't anticipated, Phrase is the ideal tool — clean capture of adjacent queries, CVR often 65 to 85% of Exact CVR.
• Generic ad groups. "CRM solution," "invoicing software," "project management tool": these query heads are too broad for Exact, too precise for Broad — Phrase captures qualified volume without massive drift.
• Accounts without mature Smart Bidding. If you're still on Manual CPC or Target CPA without enough conversions for the algo to exit learning, Phrase is your safety ceiling — Broad would be too risky here.

To go further on the Quality Score discipline that underpins Phrase performance, see our complete Quality Score guide.

Is Broad Match a good bet under Smart Bidding?

Syntax: hiking shoes without quotes or brackets. Broad is not a "broad" match type — it's an intent match type. It no longer just tries to match words, it matches the topic Google infers from the query, crossed with user context: search history, audiences, purchase signal, seasonality, hour, device. Concrete example: the keyword hiking shoes can match Rockies walking boots, best pair for 3-day trek, or even hiking backpack if the user profile is strongly outdoor shopper.

It's both its strength and its risk. Its scaling power is unbeatable: when it works, Broad unlocks volume pockets Phrase will never touch. Its semantic drift is real: without clean Smart Bidding and without negatives, Broad puts 34 to 46% of your spend on off-intent queries in 30 days by vertical. In practice, Broad without Smart Bidding = +38 to +52% off-intent search terms within the first week.

The 3 invisible prerequisites to activate Broad without getting burned:

1. Mature Smart Bidding: Target CPA or Target ROAS calibrated on 50+ conversions/month, stable for 30 days.
2. Server tracking + Enhanced Conversions: the algorithm needs reliable signals, otherwise it optimizes on noise.
3. Shared list of 200 to 500 negatives minimum, with weekly Search Terms review.

Under these conditions, Broad becomes a powerful scaling tool — in our internal analysis, 55 to 68% of e-commerce accounts run with a Broad + Smart Bidding dominant in 2026, versus only 18 to 28% of B2B accounts. Official Broad documentation is available on ads.google.com.

Recommended split by account type

Let's get concrete. Here's the median match types split observed on top-quartile performance accounts by vertical — so both what works and what's reproducible. Figures from our aggregated 90-day data, stratified by vertical.

Table reading: these ratios represent the optimal starting point, not an engraved truth. A mature e-commerce on advanced Smart Bidding can push up to 50% Broad if tracking and negatives follow; a B2B SaaS with 90+ day cycle can bring Broad down to 0%. The healthy reference ratio — 30% Exact, 50% Phrase, 20% Broad — is a midpoint for accounts without yet strong conviction on their vertical.

Watch the brand campaign: it must be isolated in its own campaign with its own budget, never mixed with generic. For B2B SaaS specifics, see our Google Ads B2B SaaS strategy, and for e-commerce details, the e-commerce 2026 playbook.

Query sculpting and negatives: the key to success

There's no viable Broad without negatives, and no profitable Phrase without regular audit. Query sculpting — the art of using match types, negatives, and ad group priority to route each query to the right ad group — is the invisible skeleton of high-performing accounts in 2026.

Target composition of a shared account negatives list:

• 40% competitor / tools / comparison negatives — rival brand names, "vs" comparisons, mentions of alternative solutions.
• 30% intent negatives — free, definition, reviews, jobs, internship, tutorial, training, wikipedia.
• 20% adjacent product negatives — neighboring categories you don't serve (e.g., if you sell CRMs, exclude "ERP").
• 10% geo negatives if relevant — cities or countries outside your delivery zone.

Operational floor: 200 negatives before activating Broad at scale. Healthy target: 400 to 600 negatives split between a shared account list + per-ad group lists. Beyond 1,000, marginal value drops — better to spend time on weekly Search Terms review to feed the list in real time.

Second essential mechanism: ad group priority via cross-negatives. Structure your ad groups in 3 tiers: High (strategic Exacts), Medium (exploration Phrase), Low (scaling Broad). Inject as negatives in Medium and Low the Exacts from the higher tier, and as negatives in Low the Phrases from Medium. Result: each query goes to the most precise ad group that matches it — you get Exact CVR on Exact queries, Phrase CVR on Phrase queries, and Broad only captures truly exploratory queries.

For a structured Search Terms review process, see the Google Ads audit checklist that integrates the negatives review into the weekly routine.

When to combine all 3 vs when to exclude one

3 scenarios dominate match types decisions in our audits.

Case 1 — Combine all 3 (common configuration)

The most frequent configuration, optimal when prerequisites are met. Structure: Exact on core keywords (top 10 strategic keywords, brand, high conversion), Phrase for controlled exploration of long-tails, Broad for scaling under Smart Bidding. Typical split in e-commerce: 20 / 40 / 40; in mature lead gen: 30 / 50 / 20. The 3 ad groups are separated, each with its own ads and own bid level.

Case 2 — Exclude Broad

3 situations justify complete Broad exclusion:

• High-ACV B2B SaaS (> $5k/year) with long cycle — Broad dilutes qualified audience, CVR collapses.
• Brand defense — on your brand, you only want to match branded queries, Exact is more than enough.
• Monthly budget < $2k — Smart Bidding never cleanly exits learning, Broad is too unstable.

Case 3 — Exact only

Rare but sometimes relevant configuration: pure brand campaign, or ad group on ultra-competitive keywords with mediocre Quality Score. In this latter case, opening Phrase or Broad would blow up average CPC without gaining useful volume — better to stay on Exact and work the Quality Score before expanding. See the CPA reduction guide for the complete method.

What are the 5 classic match types mistakes?

The 5 anti-patterns below represent 73 to 85% of match types underperformance cases observed in our audits. None is complicated to fix — you just have to detect them.

1. All Broad without Smart Bidding or negatives. The textbook case, observed on 15 to 23% of audited accounts. Result: 34 to 46% of spend wasted on off-intent queries in 30 days by vertical, CPA doubling, CVR collapsing. Broad requires mature Smart Bidding + 200 negatives minimum — without these 2 pillars, go back to Phrase.
2. Pure Exact across the whole account. The extreme opposite, observed on 8 to 14% of accounts — often from advertisers traumatized by a bad past Broad. Result: -54 to -66% potential volume, the account runs clean but under ceiling. Open at least 40% Phrase to recover qualified long-tails.
3. Mixing match types in the same ad group. Documented Google bug: when Exact, Phrase, and Broad coexist in the same ad group, the algorithm underperforms Exacts in favor of Broads (especially under Max Conversions). Exact CVR loses 10 to 18%. Absolute rule: 1 ad group = 1 match type only.
4. Negatives never audited. The shared list dating from 2022 that hasn't been touched since. Queries evolve, seasonality too, new competitors appear. Without weekly review, the list gradually loses its protective value — average observed drift: 2 to 3% additional off-intent spend each quarter.
5. Broad without Smart bid strategy. Manual CPC + Broad = the worst imaginable combo. The algorithm can't compensate for Broad's semantic drift without dynamic bidding per query. Result: Broad overbids on non-converting queries and underbids on converting ones. If you insist on Manual CPC, stay on strict Phrase.

To detect these 5 anti-patterns on your account without manual audit, launch a free SteerAds audit: it scans the match types split, negatives coverage, and bid strategy coherence in 72h, flags ad group mixes, and proposes a prioritized correction plan. For advanced accounts requiring continuous management, our Auto-optimization module adjusts match types split and enriches shared negatives lists every 24h. To go further on the PMax ecosystem that coexists with these Search match types, see our complete Performance Max guide.

Useful source to benchmark the evolution of match types practices in the industry in 2026: Search Engine Land regularly publishes qualitative analyses, and Think with Google documents Google's official view on search intent evolution.

Sources

Official sources consulted for this guide:

• support.google.com
• support.google.com
• support.google.com

FAQ