Conversion Optimization
Google Direct Offers: How to Use AI Discounts Without Destroying Margin
Updated: September 2026
Google Direct Offers lets eligible advertisers surface discounts, giveaways, coupons, and bundles within AI Mode and Search when a shopper appears ready to buy. The conversion opportunity is real, but an AI-selected deal can increase orders while reducing margin or subsidizing customers who would have purchased anyway. Guardrails must be defined before launch.
How Direct Offers works
Google says advertisers can provide eligible products, promotion options, and value limits. AI Brief can help describe the intended audience, while Gemini selects or constructs a relevant deal for the shopper’s context. Google has expanded the format beyond the AI Mode product viewer into the AI Mode answer.
Direct Offers can be managed through Google Ads for AI Max and Performance Max advertisers, complementing Merchant Center and Hotel Center workflows. Availability may vary by account and vertical.
The four risks to manage
Margin leakage
A higher conversion rate is not automatically more profitable. Discounts can reduce contribution margin, raise fulfillment pressure, and train customers to wait for a deal.
Non-incremental redemptions
A shopper who was already ready to buy may receive an unnecessary incentive. Last-click reporting can reward the offer even when it did not create the sale.
Inconsistent promises
The offer shown in Search must match pricing, eligibility, inventory, shipping, expiration, and checkout behavior. Contradictions damage trust quickly.
Brand erosion
Frequent price-led messaging can weaken a premium position. Bundles, loyalty benefits, gifts, or shipping incentives may protect brand value better than a blanket percentage discount.
Build an offer guardrail table
For each product or category, define:
- Minimum contribution margin after promotion
- Maximum discount or benefit value
- Eligible customer and product groups
- Inventory and fulfillment limits
- Geographic and channel restrictions
- Start, expiration, and blackout dates
- Stacking and coupon rules
- Refund and cancellation treatment
- Required legal or promotional language
Choose the incentive by customer problem
Use a discount when price is the real barrier. Use free shipping when delivery cost creates friction. Use a bundle when complementary products increase value. Use a gift when sampling can create future demand. Use loyalty pricing when retention matters more than immediate acquisition.
The offer should solve a specific obstacle rather than simply make the number smaller.
A measurement plan that protects profit
- Record baseline conversion rate, average order value, contribution margin, new-customer rate, and return rate.
- Separate exposed, redeemed, and completed orders where reporting allows.
- Compare promoted and non-promoted customer cohorts.
- Track repeat purchase and refund behavior after the initial order.
- Use a holdout or geo test when the offer represents meaningful spend.
- Scale only when incremental contribution profit improves.
Our guide to AI personalization testing explains how to validate relevance without damaging the experience. Retailers should also align offers with the product-page preparation checklist for AI-powered Shopping ads.
Frequently asked questions
Can Google create personalized bundles?
Google’s 2026 announcement says brands can provide promotions, eligible products, and guardrails, and Gemini can construct a relevant deal such as a product bundle.
Are Direct Offers available to every advertiser?
No. Availability, campaign support, country, and vertical eligibility can vary. Confirm access inside Google Ads, Merchant Center, or Hotel Center.
What is the most important success metric?
Incremental contribution profit is stronger than redemption rate or platform ROAS alone because it accounts for causality and promotion cost.
Offer scenarios worth testing first
Free shipping threshold
Test whether a threshold increases basket size enough to cover fulfillment cost. Compare average order value and margin, not only checkout completion.
Complementary bundle
Bundle products that solve one customer job and have reliable joint inventory. Measure attach rate, contribution profit, returns, and whether shoppers remove items before purchase.
New-customer benefit
Restrict an incentive to verified new customers when acquisition is the goal. Define how existing buyers, household members, and loyalty accounts are identified so the offer does not create frustration.
Value-added gift
A sample, service, or accessory may protect price integrity better than a discount. Confirm that the gift is relevant and that fulfillment cost remains within the guardrail.
Customer-experience QA before launch
Test the full journey from AI Mode through the landing page and checkout. Verify the displayed price, coupon behavior, bundle contents, inventory, mobile layout, taxes, shipping, expiration, and confirmation email. Customer support should receive the offer rules and a process for honoring legitimate edge cases.