Smart Bidding

Let artificial intelligence control auction bids

Manual keyword bid management has long been ineffective. Modern ad auctions on Yandex and Google take place in milliseconds. Real-time machine learning algorithms analyze over 100 signals for each user before displaying an ad: operating system, search query history, time of day, precise geolocation, weather conditions, and solvency.

We set up automated ad strategies ("Maximize Conversions", "Target ROAS") so that they work for your actual sales, not just clicks. We train algorithms on clean data, connect offline conversion transfers from your CRM, and build portfolio strategies for an even budget distribution.

  • Target CPA Auto-Strategies Setup (pay-per-conversion)
  • Target ROAS optimization for e-commerce stores
  • Offline Conversions Integration (transferring payments from CRM to ad accounts)
  • Designing goal values for training algorithm calibration
  • Conducting clean A/B tests of autostrategies (Yandex Experiments, Google Drafts)

-30%

Average cost per lead (CPA) reduction with proper Smart Bidding training

+20%

Conversion volume growth while maintaining the current monthly advertising budget

100+

Signals are processed by the neural network in fractions of a second during each bid auction

24/7

Continuous budget optimization and filtering out non-converting traffic

Garbage in — garbage out

Machine learning algorithms are smart, but they do not know your business. If you train a model on Metrica auto-goals (like clicking a button), it will find thousands of users who love to click but buy nothing. We build a funnel of quality signals.

Our approach

How we make automated strategies generate profit

We do not blindly rely on automation; we manage the training signals of the algorithms.

Goal values (Value Rules)

AI must understand the difference between a cheap newsletter subscription and a purchase worth 50,000 ₽. We assign a weight (value) to each goal on the website based on real profit.

The algorithm automatically optimizes the advertising budget in favor of audience groups making the most expensive purchases, maximizing ROMI.

End-to-end offline conversions

Often a website lead is just the initial contact. The deal is closed by a manager over the phone 3 days later. We set up API data transmission from your CRM back to Yandex and Google ad accounts.

The neural network is trained to find clients who pay invoices, not just leave test leads.

Micro and macro conversions

If you have few sales on the website, the algorithm won't have enough data for training. We build a hierarchy of signals: we train the model on frequent micro-actions (adding to cart, viewing 3 pages, clicking on contacts).

This gives the algorithm the necessary critical mass of data for stable bid optimization even with a small budget.

Process

Smart ads setup stages

Sequential process of analytics calibration, test launch, and driving campaigns to profitability.

01

Web Analytics Audit

We check proper operation of Yandex Metrica and Google Analytics 4. We clean up duplicate goals, remove junk auto-goals, and set up UTM tagging.

02

E-commerce Data Setup

We enable e-commerce tracking: transferring cart value, order contents, brands, and categories of purchased goods to dashboard accounts.

03

CRM integration via API

We write scripts for sending offline events (invoice payment, successful measurement, completed deal) from CRM to Yandex Metrika and Google Ads via Measurement Protocol.

04

Designing and launching strategies

We group campaigns into portfolios to accumulate statistics. We set starting limits for Target CPA or advertising cost share (ROAS/CRR).

05

A/B testing of strategies

We launch a controlled split test: one part of the audience goes to the auto-strategy, the second to the control group. We verify the actual cost per lead.

06

Scaling and control

We smoothly increase budgets without retraining the strategy, track creative burnout, and redistribute limits in favor of profitable combinations.

Our stack

AI Optimization Tools

We use built-in ML algorithms of advertising systems and external analytical platforms.

Yandex Direct Smart Bidding

Autostrategies for optimizing conversions and profitability in Yandex Advertising Network (YAN) and Search. The algorithm learns from the behavior of millions of Russian users.

Google Ads Smart Bidding

Target CPA, Target ROAS, and Maximize Conversions strategies based on Google neural networks. Works effectively in Performance Max campaigns and YouTube Ads.

GTM Server-Side & Looker Studio

Server-side tracking tools (for protection against ad blockers) and interactive Looker Studio dashboards for monitoring the training progress of autostrategies.

Prices

Smart Bidding Setup Pricing

The price depends on the number of ad accounts, volume of website conversions, and CRM integration.

Features Audit and Setup Preparation of analytics and launch of strategies 65 000 ₽ Timeline: from 7 days Order Popular ROAS Optimization E-commerce and CRM Offline Conversions Integration 110 000 ₽ Timeframe: from 15 days Order AI Ad Management Comprehensive launch and management of split tests from 180 000 ₽ Timeframe: from 30 days Discuss
Audit and Cleanup of Analytics Goals Full Full Full
Conversion value setup (Value Rules) Basic setup Dynamic (e-commerce) Integrated with CRM Margin
Transfer of offline payments from CRM via API Webhooks/API Setup Full-fledged queued connector
Conducting clean A/B tests 1 split test in Yandex Regular testing of strategies and creatives
Looker Studio dashboard by ROMI Basic Custom Looker dashboard
Management and support period 14 days after setup 30 days of management and optimization 60 days of full management
FAQ

FAQ about AI advertising and bidding

Didn't find the answer you need? Leave a request — our lead PPC specialist will conduct an express audit of your current ad campaigns and point out errors in automated strategies.

  • What to do if an autostrategy suddenly stops generating leads?

    A sharp drop in leads with Smart Bidding enabled ("learning broke") usually happens for three reasons: 1) a sharp reduction in budget (the algorithm lacks funds to buy enough conversions); 2) an excessively low Target CPA, preventing the bot from winning auctions against competitors; 3) a technical analytics failure (goals broke on the site, and the algorithm assumed there were no more conversions). We perform diagnostics, temporarily loosen restriction parameters, or restart learning based on a portfolio strategy.
  • What minimum budget and number of conversions are needed for training?

    For high-quality ad campaign training in Yandex or Google, 10–15 target events per week per strategy are required (ideally 20–30+). If your weekly advertising budget equals the cost of a single conversion, the algorithm won't be able to learn. In such cases, we combine campaigns, expand geo-targeting, or switch model training to micro-conversions (e.g., "click on phone number" or "click to messenger"), which occur 5-10 times more often.
  • What is the difference between click optimization and conversion optimization strategies?

    Click optimization buys the cheapest possible traffic without being tied to actions on the site — you will get many visitors, but almost no one will buy. Conversion optimization analyzes user behavior history and raises bids only for those people whose profile matches your buyers. The cost per click may increase, but the final cost per lead (CPA) will decrease.
  • Do you need a marketer if bids are set automatically?

    Yes, absolutely. Automation manages bids in the auction, but it cannot: write engaging ad headlines, draw banners, analyze competitors' value propositions, configure end-to-end data transfer from CRM, and set proper financial limits. Without a marketer's oversight, auto-strategies can easily waste budget on non-target audiences. We act as "pilots" guiding the autopilot of advertising systems.
Effective PPC

Want to reduce lead cost from Yandex or Google Ads?

Submit a request and provide your website URL. Our specialist will perform an express analysis of current ad campaigns, evaluate conversion tracking setup quality, and show the potential for launching Smart Bidding.

Launch AI ads