Dmitriy К. Google Ad Expert

Smart shopping campaigns: strategy, configuration and analysis

The official release of Google Smart Shopping took place during Google Marketing Live 2018 and caused a stir among advertisers. Smart Shopping campaigns have been updated multiple times since launch; now they are used even by big advertisers with large budgets, often bringing great results.

Smart shopping campaigns continue the trend of recent years: shifting the focus from performing mechanical tasks to the opportunity to solve strategic problems; and maximum automation on all fronts, which we talk about regularly and hear about at conferences. And that is definitely a big plus - both for us specialists and for companies.

In the future neural networks will do the job that a human does now, but do it better, because a human still makes mistakes, and a neural network, if learned correctly, cannot make mistakes. It will consider all factors and make the most correct announcement in this case. Even if we don't see it yet, we have to fight for it.

What are Smart Shopping Campaigns?

The main difference between standard shopping campaigns and smart shopping campaigns is that smart shopping campaigns use machine learning for automation.

The task is the same: to relieve the advertiser of "manual" tasks when setting up and maintaining contextual advertising and to increase the effectiveness of campaigns thanks to an automated search for the optimal parameters.

Machine learning allows you to place bids based on the probability of purchase for each individual user. Thus, the system evaluates "external" (user) and "internal" (our, "advertising") factors.

External factors are search query, user location, Internet access device, etc. Internal factors are what indicates and influences the likelihood of a purchase by an advertiser i.e. audience lists, statistics of previous purchases, product prices and other data.

Google Smart Shopping settings

Smart campaign ad types and placements:

- advertisements in Google search publication;

- display ads on the Display Network (primarily used for remarketing);

- dynamic remarketing ads on the Display Network, YouTube and Gmail.

Ad formats in Google Smart Shopping campaigns:

- simple shopping ads for the search network;

- local assortment ad format;

- dynamic remarketing and dynamic customer acquisition;

- HTML5 declarations.

How Smart Shopping campaigns work

1. The advertiser adds the company logo, image and text content.

2. The advertiser chooses to display all or part of the products in the feed/

3. The system automatically creates advertisements from the offered materials

4. Relevant ad variations are displayed on Google networks if the system considers the user's search query to be relevant to the product.

As with any campaign, Google Ads determines relevance by comparing search queries and flow data. In CMS, the system evaluates the user's interaction with the site. For example, if the user has visited the site before and was interested in a certain product, he will see its advertisements. The interests are also taken into account, that is, the system will work both as a remarketing network and as a regular display network.

Benefits of smart campaigns

New Smart Shopping campaigns require minimal maintenance once set up. Thanks to two features: expanded reach across all Google networks and a new smart bidding strategy.

Extended Scope

Extending reach across Google Networks means that Smart Campaign ads will serve users beyond search and retail channels, but also on the Display Network, YouTube and Gmail. Displays are managed by Google itself: the system estimates the likelihood that showing ads will result in a conversion and decides when and where to show ads for Smart Shopping campaigns.

An extended range feature is an increased chance of conversion. Since Google analyzes user and advertiser statistics, it is more likely to show ads to potential customers, that is, people who are theoretically interested in buying your product.

In this case, the essence of machine learning and automation is to show the right product to the right user, in the right place, at the right time. Smart shopping campaigns seem to do the job.

Smart Bidding Strategy

There are many variables to consider when making important optimization decisions, combined with cues of intent and context.

The new shopping campaign type automatically optimizes goals and lets you take guessing games off the agenda to achieve your marketing goals, whether it's maximizing conversion value or maximizing conversion value with some return on ad spend (ROAS).

The systems will evaluate all available signals and use machine learning to predict the likelihood of a search query converting to get the most out of your network, product and audience budget.

Disadvantages of Smart Shopping campaigns

- Less control over the context in which ads appear;

- Smart Shopping campaigns do not report on search queries: we don't know how or why we got clicks;

- There is no audience targeting in Smart Shopping campaigns. Yes, you need to have at least a list of 100 users, but you can't decide which audience to target. Google will automatically select based on a machine learning algorithm;

- Smart Shopping campaigns have no device targeting, no bid adjustment, no scheduling: less control over when and where ads are shown;

- It’s impossible to evaluate the performance of different networks separately in Smart Shopping campaigns;

- The total budget is not available;

- The search impression percentage data is unavailable; Search Abs Top IS and percentage of clicks are unavailable as well;

- You cannot influence the percentage of lost impressions (budget and rating).

With Smart Shopping campaigns, you lose manual campaign management and Google's artificial intelligence is in charge. Can we trust it? We believe that only independent testing and performance analysis in your specific case can answer this question.

Bid Strategies

There are 2 smart bidding strategies for Smart Shopping campaigns. In the bidding process, they use "various factors, such as the device on which the fingerprint is taken, the location of the user, the time of day, the language of the interface and the operating system". These include the maximum conversion value and the return on ad spend target.

Maximum conversion value

While other automated bidding strategies focus on driving conversions, the new "Maximum Conversion Value" strategy focuses on driving revenue. The goal of this smart campaign strategy is to generate as much revenue as possible from a given budget; it is set by default.

This is a very interesting indicator of how Google is investing in automated bidding strategies and highlights its efforts to find new solutions for retailers. With Amazon and other e-commerce giants, the online shopping space is becoming increasingly competitive. And Google certainly won’t stay aside!

Target ROAS

In addition to a maximum conversion value strategy for smart campaigns, you can also set a target return on ad cost. Then the system will maximize the conversion value, taking into account the target return on advertising cost. In other words, the strategy is still focused on conversions, but gives you more control over costs. This strategy in Smart Campaigns is useful for advertisers who have strict ROI requirements.

Attract new customers

Earlier the maximum value per conversion allocation and the target return on ad spend were the only strategies available for Smart Shopping campaigns.

Now Google Ads began testing a new smart buying objective: new customer acquisition (NCA). NCA aims to attract new customers and optimize the costs of this attraction. You can set the customer acquisition cost. The new buyer's value will be used in the auction and listed in the settings.

The Total Conversion Value will equal the sum of the New Customer Value and the Purchase Value.

Which strategy to choose

We recommend that you first launch a Smart Shopping campaign with maximum conversion value and let it run for 2 weeks without making any changes. In this way, we allow transaction data to accumulate and the return of the advertising costs themselves. Once we have the right data, we can carefully test our target ROAS. To begin with, set the indicator a little higher than the real one - by about 20%. Next, it's important to run the Smart Shopping campaign for at least a week to gauge the results. If the smart campaign performs well and reaches the set level [Actual ROAS + 20%], you can test the increase by another 20%. And so on until your target level

However, it is important to understand that the choice between the maximum conversion value and the maximum conversion value with the target ROAS, can significantly affect the effectiveness of smart shopping. If the target ROAS plugin works to achieve a given ROAS value, the maximum conversion value strategy tends to increase ROI.

You can use simulations to understand how adjustments to your ROAS strategy affect the results of your Smart Shopping campaigns and make more informed decisions next time. They show what would happen if you set different rates or budgets. Simulators work for the strategy "Targeted ROAS", separately for the maximum conversion value is not available.

If there are a lot of different products in the range, they can start to compete with each other and it will increase the cost per click and then when it sells, the price stays the same, but the ROI goes down. Thus, the company makes less profit literally out of the blue. This problem can be solved by proper segmentation of goods.

Structure and segment implementation of Smart Shopping campaigns

Google recommends combining products in a single Smart Shopping ad and creating separate campaigns only if warranted by the specifics of your business.

Does that mean you can't create more than one Smart Shopping campaign per account? Of course not.

You can set up to 100 active or paused Smart Shopping campaigns and manage them in your Google Ads account. Technically, an advertiser can use any of them. In fact, you can create any number of smart campaigns and they'll work just fine IF the products in those campaigns don't overlap.

What does this mean? For example, we want to take 10 identical products and create 2 campaigns: one with a maximum conversion value strategy, the other with a target ROAS. Run them simultaneously to test the effectiveness of the strategies and the results of both campaigns. Great idea, isn't it? It’s not.

If you run 2 Smart Shopping campaigns with overlapping product sets, they will affect each other: one or both campaigns may not drive impressions or their performance may be affected. Either way, account performance will drop.

What happens if a Smart Shopping campaign has low impressions and clicks for a long time? Officially, Google does not disclose its methods for improving the quality of work, including actions against advertisers whose campaigns are performing poorly. But after many years of work, communication with technical support and Google presentations, we can imagine the scenario of how the system handles in such situations.

For the system, the absence of clicks or conversions in a Smart Shopping campaign indicates that the campaign is not relevant or useful to users. And since user satisfaction is Google's main objective, these indicators are the subject of the greatest attention.

If there are no clicks or conversions for a long time, Google drops the count to other advertisers who target similar keywords but show better ad results. This comes at the cost of a lower Quality Score, which means these accounts will have to bid higher in the auction to get a better placement. If a smart campaign works with a smart bidding strategy, the system itself will overestimate them.

There are two options:

1. if the performance of the video campaign returns over time with clicks and conversions so Google will gradually reduce the bids and adapt them to the level of the marke

2. if performance is kept at a minimal level, impressions will only be delivered where competition is weakest, i.e. on unpopular queries or on weak sites.

At the same time, campaigns whose product set does not even fully match can influence each other. In other words, if a product is found in two campaigns and the other products in them are different, this can already affect the work.

Therefore, Smart Shopping does not leave a convenient opportunity to test and compare campaign performance with different parameters and slightly complicates account structuring.

If you want to launch or order the maintenance of two advertising campaigns with different parameters for the same products, the most practical solution is to launch standard shopping campaigns. They don't use machine learning to the same extent as Smart Shopping campaigns, but Smart campaigns will have an edge.

And the last important nuance, which concerns the question "To segment or not to segment?" For multimedia ads, we normally specify a text and 1 image. Therefore, if we have a wide and varied assortment, adding all products to one campaign would be a mistake in terms of advertising relevance. This is why segmentation is so important.

Campaigns segmented by product potential

A general rule for segmentation in Google Smart Shopping is to include products in different campaigns if you want to use a target ad return strategy and set different ad return values.

One way to achieve this is to use what we classically call product potential segmentation. For example, the site sells laptop power supplies for $500 and laptops for $7000. These products obviously have different levels of marginality, which means that it is important for them to define different target ROAS.

In this case, it makes sense to divide the products into categories: 1 category = 1 Smart Shopping campaign

Segmentation also lets you set your own budget for each campaign, giving you a little more control in Smart Campaigns.

Also, pay attention to products in Smart Shopping campaigns: the better organized campaigns are, the more effectively they will work. If you find that only certain products perform well, highlight the most effective ones and launch Smart Shopping for them. For example, with one of our clients, we noticed that Smart Shopping only showed a few products. Other products flashed only occasionally or received no impressions at all: the campaign was in the plus. We've highlighted the most converted products over the past few months and launched a smart campaign just for them. This made it possible to quickly collect statistics for training the system.

Impact of Smart Shopping on other campaign types

Smart campaigns take priority over standard Shopping campaigns and Display Network remarketing campaigns.

If you are running only one Smart campaign, especially for a small range of the same type, we recommend pausing all standard Shopping and Display Reading campaigns to minimize interference with machine learning algorithms.

In other cases, depending on the business objectives, and especially for stores with a large assortment, it makes sense to use Smart Campaigns for only some of the products and advertise the rest of the positions using standard product campaigns, remarketing on the Display Network. The key is to provide the system with a set of products that does not repeat itself between smart shopping campaigns, shopping and dynamic re-advertising.

There are no such barriers and priorities between smart campaigns and search campaign

Preparing to set up Smart Shopping campaigns

Before setting up Smart Shopping campaigns, you need to make sure you've set up your Google Merchant Center account and product data feed.

Account must be eligible for Shopping campaigns:

- Google Merchant Center and Google Ads accounts are linked;

- it is possible to send updated data on the goods at least once every 30 days;

- You must have had at least 20 conversions from existing Shopping campaigns in the last 45 days and the value of those conversions will be credited to your account.

In new accounts, it is possible to immediately create a smart shopping campaign, without conversions. But it is important to understand that without historical data, the system simply cannot function properly. That's why we recommend running a standard campaign to collect an initial amount of data, then start buying smart.

- the site must work in general, opened on all common browsers;

- actual and target URLs must match;

- there is no flashy, attention-grabbing tricks (like "FREE" caps lock);

- all descriptions are accurate and reliable;

- compliance with data collection and use rules;

- advertise only products authorized by the system with authorized content;

- conversion tracking is set up and transaction value is displayed to understand which ads are driving sales;

- the conversion tracking tag is active;

- set the global tag required for basic Google Ads remarketing functions and pass parameters for dynamic remarketing;

- have a remarketing list with at least 100 active users to use dynamic remarketing to show personalized ads to those who have already visited the site. To keep your campaign active, you need to keep adding new people to your remarketing audience. If the list is not updated within 28 days, the campaign will not start until new users are added.

Configuring Google Smart Shopping

Google Smart Shopping campaigns are very easy to set up:

1. Create a new Google Ads campaign.

2. Set the goal as sales, it is recommended by Google.

3. Shopping campaign type.

4. The campaign subtype is Smart Shopping.

5. Bid strategy: Maximum conversion value or return on ad spend target

6. Daily Budget. A good starting point for budgeting for Smart Shopping campaigns is the daily spend history of Classic Shopping campaigns. A new campaign automatically maximizes conversion value within a given budget.

7. Configure a product group. We recommend that you start working with Smart Shopping from the tests and initially launch a campaign only for the selected product groups. This way, you can see how Smart Shopping works with a small inventory before making a big investment in setup and management.

To create a product group, you can select category, brand, item ID, product type, special tags. Be sure to exclude these products from your standard Shopping campaign so the Smart campaign priority doesn't affect your impressions in other ad accounts.

8. Upload your logo, ad display images and text content (short and long title, description, URL). This data will be used for display ads, YouTube and Gmail.

Video is also available in Smart Shopping campaigns. If you don't have your own videos, you can use those automatically generated from product data: titles, logos, company names and objects from feeds. Video length - no more than 30 seconds, aspect ratio - 16:9, 1:1, 4:3, 9:16. The videos will play to various in-app locations to increase reach.

All graphics must be of good quality. A company logo can be uploaded from a Merchant Center account; you should create responsive remarketing ads: display ads and YouTube ads for users who visited your site but did not show interest in certain products. If the user pays attention to a certain position, he will see an ad with exactly the information he is looking for.

Smart Shopping campaign analysis

Google recommends leaving Smart Shopping campaigns for 15 days before evaluating performance. These 15 days are necessary to adjust the system and start effective campaign optimization. In fact, this data collection interval may vary.

If you get a stable number of impressions, clicks and conversions within 15 days, you can give the system more time for analysis, 20 to 30 days.

If in the first 3 to 4 days you see a lot of views and clicks WITHOUT conversions, and from 5 to 6 days the views drop by 80% or more, it is likely that this campaign will no longer work. This is because Google doesn't see the conversions at first, views the campaign as low quality, and then doesn't give it a chance. If you see this situation in the reports, we recommend that you do not wait more than 10 days. Stop the campaign, optimize the product page, images, and other in-feed attributes, and relaunch the smart campaign.

Also, before evaluating the effectiveness of a campaign, make sure you have enough data. "Sufficiency" is a subjective concept. Typically, the normal amount of information for Smart Shopping analysis is accumulated 2 to 3 weeks after the training period.

Decision time analysis

For a correct interpretation of the statistics, it is important to understand how much time elapses between the moment the ad is seen (and clicked on it) until the actual conversion. In some companies, the time to make a purchase decision is short: in this case, the conversion takes place within a day of the click. In other cases, the waiting period can be extended up to 3 months.

In Google Ads, this is important because the time between impression and conversion will affect the data available in reports. For example, in a new campaign, you may see a decrease in conversions and decide that it is ineffective and should be disabled. Or maybe this campaign had a longer decision-making period and the volume would catch up in a matter of weeks.

For the same purpose, it is important to consider assisted conversions.

Network analysis

In Smart Shopping campaigns, there is no way to measure the performance of Shopping, YouTube, and Gmail separately. Ads on these networks work differently: display ads tend to have low click-through rates and low conversion rates, while Gmail and search ads perform much better in this regard.

As a result, metrics for Smart Shopping campaigns may show lower performance than regular Shopping campaigns. In fact, the statistics will contain data on advertisements in all networks. It is important that you and your client know this.

On the other hand, you may notice an improvement in performance when comparing a Smart Shopping campaign against all existing Shopping campaigns and Dynamic Remarketing campaigns due to more advanced algorithms.

Additionally, you can see where the Smart Shopping traffic is coming from.

Conclusion

Google Smart Shopping has been running since 2018. The essence of the campaign is based on machine learning, which takes into account external and internal factors for optimal bidding and displaying ads to the most interested users. This approach automates the work and improves the performance of advertising campaigns, since now their parameters are based on a much larger number of signals than a PPC specialist could process to set up a regular campaign. To do this, Google will use the Wide Coverage system in the networks and the Maximum Conversion Value bid strategy.

On the other hand, automation takes away some of the control over where and how ads are displayed. In particular, we cannot list negative keywords; use region/city or device-specific targeting; independently select a specific audience or enable bid adjustments.

How serious are these defects? It's worth looking at how Smart Shopping campaigns are performing on your projects. In our view, the benefits of smart shopping campaigns outweigh the loss of control – with proper setup, strategy selection and regular performance reviews, of course.

Smart Shopping includes SERP ads, display ads on the Display Network, and dynamic remarketing on the Display Network, YouTube, and Gmail. Do you want to order contextual advertising? Contact us!