Google Ads bid adjustments in 2026

With Smart Bidding, 28 to 40% of manual bid adjustments still in place are crushed or counterproductive: they inject noise into an algo that already weights 70 internal signals. The 2026 rule is simple: -100% to exclude a device, zone, or time slot — 0% everywhere else, unless an explicit Manual CPC case.

Bid adjustments are one of the oldest legacies of Google Ads. Device adjustment, geo adjustment, audience adjustment, dayparting: four levers available since the late 2000s that, in 2026, find themselves largely crushed by Smart Bidding. Should you still use them? The answer depends entirely on your bid strategy. This guide walks through the 4 types of adjustments, their behavior by bid strategy, the exact calculation formula, priority order, and the 3 cases where they remain legitimate in 2026. To go further on bidding mechanics, see our Smart Bidding Maximize vs Target CPA comparison.

What did the Smart Bidding era change for bid adjustments?

Before Smart Bidding — up until around 2018 — bid adjustments were the primary optimization lever on a Google Ads account. Under Manual CPC, the advertiser set a bid per keyword, then refined it via 4 successive modifiers: device (mobile, desktop, tablet), geo (zones and zip codes), audience (remarketing lists, In-Market, Affinity), dayparting (hour and day). Each segment multiplied the base bid, sometimes up to ±900%. It was laborious, manual, but the manager kept total granular control.

Smart Bidding — Target CPA, Target ROAS, Maximize Conversions, Maximize Conversion Value — reversed this logic. The Google algorithm uses more than 70 real-time signals (hour, device, geo, user history, query context, weather, seasonality, etc.) to calculate an optimal bid on each auction. In this world, nearly all manual bid adjustments are ignored or overridden. Official documentation on Google Ads support.

One notable exception that survives in 2026: the -100% modifier remains active on device and geo, because it's equivalent to total exclusion (functional equivalent of criteria exclusion). You can always cut an entire device, zone, or time slot under Smart Bidding. Everything else — +20% mobile, -15% on Alaska, +30% RLSA — is read by the algo as one signal among 70, automatically weighted, and rarely applied literally.

The strategic consequence: in 2026, the decision is no longer "what adjustment value" but "should I keep this adjustment at all". For a pure Manual CPC account — rare but still existing in heavily regulated sectors — adjustments retain all their relevance. For everything else, give way to segmentation by campaigns and asset groups.

Device adjustments: mobile vs desktop vs tablet

Google Ads has historically distinguished three device categories: mobile phone, desktop, tablet. Each accepts an adjustment between -100% (total exclusion) and +900% (nine times the base bid). Device adjustment is configured at the campaign or ad group level, and applies to every keyword it contains.

Under Manual CPC, typical values observed across our audits:

• B2C local retail: mobile +10 to +25% (strong local intent on smartphone), desktop -10 to 0%, tablet -30 to -50%.
• Generalist e-commerce: mobile +0 to +10% (observed median close to 0), desktop 0%, tablet -20 to -40%.
• B2B SaaS: mobile -10 to -30% (purchase mostly desktop), desktop +10 to +20%, tablet -50 to -80%.
• B2C gaming: desktop -20 to -40%, mobile +20 to +40%, tablet -30 to -60%.

Across the SteerAds sample 2025-2026, under Manual CPC, the median mobile device adjustment is -3 to +7% in e-commerce and -15 to -25% in B2B. These numbers reflect market duality: e-commerce trends toward device parity, B2B remains strongly desktop-dominant on the purchase act even when discovery happens on mobile.

Under Smart Bidding, the rule changes radically. Only use -100% to cut a device entirely. Two legitimate cases observed: a SaaS dashboard unusable on mobile (cut mobile to -100%), a native mobile game with no desktop version (cut desktop to -100%). Any intermediate value is ignored by the algo.

Caution signal before any device adjustment: first check your mobile conversion tracking. In 55 to 68% of audited accounts (by vertical), mobile conversions are under-reported — GA4 cross-device poorly configured, Enhanced Conversions disabled, CMP consent broken on mobile. An apparent mobile CVR of 2% can hide a real CVR of 3.5%. Before adjusting to -30%, audit the tracking — see our Google Ads audit checklist.

Geo adjustments: campaign level + audience

Geo adjustment plays out at three successive levels in Google Ads: the campaign (Locations section), the ad group (target restrictions inherited from the campaign), and location groups (location groups shared via the shared library). The finer the level, the more precise the adjustment, but the more conversion volume it requires to be statistically solid.

Typical values under Manual CPC:

• Major US metros (New York, Los Angeles, Chicago): +10 to +25% on premium e-com verticals.
• Sparse rural areas: -10 to -30% depending on purchasing power and delivery logistics.
• Outlying territories: -20 to -50% if delivery is limited or CPA historically high (shipping fees, delays).
• High-income zip codes (10021, 94027, 90210, etc.): +30 to +50% on luxury, real estate, premium services.
• Seasonal tourist hotspots: combined dayparting, +20 to +40% in high season only.

In our sector panel, the median urban/rural CPA gap is 30 to 41% by vertical — a significant gap that justifies geo granularity under Manual CPC. But under Smart Bidding, geo adjustment is crushed by the algo. The recommended solution: switch to segmentation by campaign structure. Create two separate campaigns "US Premium" and "US Standard" with differentiated budgets, Target CPA, and creatives, rather than applying a +25% on a single campaign that will be ignored.

Special legitimate case under Smart Bidding: geo adjustment -100% to exclude a non-delivered or non-relevant zone (e.g., a fresh-food e-commerce that doesn't ship to overseas territories). This exclusion stays active and clean. See the official geo-targeting documentation.

Audience adjustments: RLSA, Customer Match, interests

Audience adjustment applies to five major Google audience families: RLSA (Remarketing Lists for Search Ads — recent site visitors reused on Search), Customer Match (CRM emails/phones upload), In-Market (users with purchase intent in a sector), Affinity (long-term interests), and Similar Audiences / Lookalikes (though deprecated and merged into Customer Match in 2023-2024).

Healthy adjustment values under Manual CPC:

• RLSA 30-day visitors: +30 to +50% (strong return intent, typical CVR ×2 vs cold).
• Customer Match existing customers: +20 to +40% (re-engagement, cross-sell, upsell).
• Customer Match churned: +15 to +30% (win-back, CVR slightly lower than active).
• In-Market target sector: +10 to +20% (broad intent, to monitor).
• Affinity: 0 to +10% (too top-of-funnel to push hard, especially on Search).

In practice, the legitimate Customer Match adjustment observed lies between +15 and +30% — beyond that, you enter overbidding on an already-acquired public, which degrades marginal profitability.

Under Smart Bidding, the logic reverses. Audiences are no longer bid modifiers but input signals for the algo. Adding RLSA to a Smart Bidding campaign tells Google who your recent visitors are, without forcing a +30%. The algo will use them appropriately by context. Our rule: under Smart Bidding, attach all relevant audiences in Observation mode, leave at 0% adjustment, and let the algo work.

Observation mode lets you collect performance data by audience without modifying the bid — useful to decide later if a campaign split is needed. Documentation on Google Ads audiences.

Dayparting: hour and day of week

Dayparting (or ad scheduling) lets you adjust the bid by time slot (0-23h, quarter-hour or full hour) and by day of the week (Monday to Sunday). Possible combinations: 168 weekly cells (7 × 24), each accepting an adjustment from -100% to +900%. Accessible in the "Schedule" section at the campaign level.

Classic B2B SaaS example:

• Monday-Friday 9 AM-5 PM: +10% (peak B2B activity hours).
• Tuesday-Thursday 10 AM-12 PM and 2-4 PM: +15% (observed demo peaks).
• Monday-Friday 10 PM-7 AM: -50% (sales team absent, leads unqualifiable).
• Saturday all day: -70% (near-zero B2B traffic).
• Sunday all day: -100% (total cutoff, no qualifiable leads).

In our sector panel, dayparting is useful on 35 to 47% of B2B accounts (qualification cycles requiring sales presence during business hours), but useless on 72 to 84% of 24/7 e-commerce accounts. The reason is simple: e-com accepts orders at any hour, and Smart Bidding already measures conversion peaks by hour via its internal signals — a manual adjustment adds nothing.

Valid case under Smart Bidding: dayparting -100% to fully cut non-business hours when there's no one to qualify leads. Typically: a law firm that doesn't handle Sunday leads loses 12% of qualified leads if ads run 24/7 — better to cut. Dayparting also remains relevant for regulated sectors (gambling, alcohol) with legal time constraints. See Think with Google for hourly behavior trends by vertical.

How do you calculate the right bid adjustment (formula + example)?

The universal formula applied by our teams on all bid adjustment types is the following:

Concrete mobile device adjustment example:

• Account mean CVR over 30 days: 4.5%.
• Mobile segment CVR over 30 days: 3.2%.
• Application: (3.2 / 4.5 − 1) × 100 = (0.711 − 1) × 100 = −28.9%.
• Pragmatic rounding: apply -29% or -30% (Google accepts both).

Concrete RLSA audience example: mean CVR 4.5%, 30-day RLSA CVR 7.8%. Application: (7.8 / 4.5 − 1) × 100 = +73%. You'd apply +70% — but watch the profitability ceiling: beyond +50%, marginal profitability often drops because you're paying more for an already-probable conversion. Empirical rule: cap RLSA at +50% even if the formula suggests more.

This calibration methodology is detailed step by step in the HowTo block associated with this article (5 steps, 45 min of execution on a mature account). To automate this diagnostic across your entire account, a SteerAds audit detects under-calibrated or counterproductive adjustments within 72h, stratified by vertical.

In what order are bid adjustments prioritized?

Under Manual CPC, adjustments don't cancel each other, they multiply. Each modifier applies in cascade on the base bid, in a strict order defined by Google Ads: campaign → ad group → criteria overrides (location, audience, device, schedule). Mastering this order is essential to avoid aberrant stacking.

Typical B2C urban stacking example:

• Base keyword bid: $1.00.
• Mobile device adjustment: -20% → ×0.80.
• Downtown New York geo adjustment: +10% → ×1.10.
• 30-day RLSA audience: +30% → ×1.30.
• Tuesday 10 AM dayparting: +5% → ×1.05.
• Final bid: 1.00 × 0.80 × 1.10 × 1.30 × 1.05 = $1.201 (or +20% net).

This +20% net stack is perfectly coherent. The danger: unsimulated stacks giving +150% or -70% without anyone noticing. In practice, 28 to 40% of Manual CPC accounts show at least one aberrant stack silently degrading performance.

Mandatory simulation rule: before activating a 4th adjustment on a campaign, calculate the total product to verify it stays between 0.5× and 2× the base bid. Beyond that, either a rule is miscalibrated, or you need to switch to a dedicated campaign structure (e.g., isolated "Premium NYC Mobile RLSA" campaign). See also our Google Ads CPA reduction guide to go deeper.

Smart Bidding + bid adjustments: trap or lever?

Mostly, a trap. Smart Bidding is designed to optimize in real-time across more than 70 internal signals the advertiser doesn't even see. Imposing a manual +30% mobile without solid statistical data behind, that's injecting arbitrary noise into a system that already has its own measurements. The algo interprets it as a constraint, loses degrees of freedom, and often degrades performance instead of improving it.

The 2026 golden rule for Smart Bidding accounts:

• Device adjustment: only -100% to cut a device entirely (e.g., SaaS dashboard unusable on mobile).
• Geo adjustment: only -100% to exclude a non-delivered or non-targeted zone.
• Audience adjustment: 0% — attach in Observation mode, let the algo work.
• Dayparting: -100% only to cut hours without possible qualification (e.g., B2B weekend).
• Everything else: leave at 0% and trust Smart Bidding.

One point often stays fuzzy: why does Google still accept non-100% adjustment input under Smart Bidding if the algo ignores them? Pragmatic answer: compatibility with accounts that aren't 100% Smart Bidding (some ad groups or extensions are, others aren't), and historical backward compatibility. Draw no conclusion from the fact that input is possible — it guarantees nothing about application.

For edge cases — multi-catalog e-commerce, hybrid Manual/Smart accounts, regulated sectors — the tradeoff requires a detailed audit. Our free SteerAds audit detects within 72h counterproductive bid adjustments left in place after migration to Smart Bidding, typically on accounts that changed bid strategy without cleaning old modifiers. For continuous automation, our Auto-optimization module audits obsolete adjustments daily and alerts the team on aberrant stacking.

Going further: compare bidding strategies via our complete Performance Max guide and our Google Ads B2B SaaS strategy. Complementary analyses on Smart Bidding evolution are regularly published on Search Engine Land.

Sources

Official sources consulted for this guide:

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

FAQ