Google Ads launched Performance Max campaigns in April 2020. In the summer of 2022, the Mountain View firm permanently replaced Smart Shopping campaigns with Performance Max. A question then arises: are the results better when comparing Performance Max vs Shopping?
To answer it, I offer you a feedback based on the case study of an e-merchant who made the migration. And for all the details on what Performance Max is, and all the associated best practices, be sure to read this article:
The foundation, satisfying and profitable Shopping campaigns
11,000+ transactions and 400% profitability for Starter Shopping campaigns
An e-merchant has been using standard Google Shopping campaigns coupled with automatic bidding strategies, which he has set up alone without the help of sea agency, for more than 2 years. He is satisfied with his results, seeing that his Shopping campaigns are profitable (400% ROAS) and account for 1/3 of his e-commerce revenue generated by Google Ads.
So much so that when the Performance Max campaigns are available, he decides not to make the switch. However, Performance Max will definitely replace Shopping campaigns at one time or another: lying in the sand is not the right approach…
The advertiser, who belongs to the “small/mid size market” category, has an interesting volume of data (still far from the large e-commerce sites which spend millions). We are talking here about:
- 11.5k transactions per year
- 400k€ of turnover generated by Google Ads each year
- for a level of expenditure of €78k over the last twelve months.
The advertiser uses standard Shopping campaigns, coupled with automatic bidding strategies.
Will the test: Performance Max (vs. Shopping) improve profitability, without losing volume?
After taking over his account, the audit concluded that not testing Performance Max would be a missed opportunity. With application, all the good practices that I share here are implemented.
Standard Shopping campaigns will be replaced by Performance Max. To draw conclusions, it is decided that the test will last up to 100k€ of turnover generated by the new campaigns (i.e. approximately 3 months). At such a level, there will be a greater chance that the test will be statistically significant. During this test, no intervention will be made on the offer, the site and its UX, or in the Merchant Center flow.
The decision is made: a Max Performance vs Shopping test is launched, until the next 100k€ of income.
Implementation of the Performance Max vs Shopping test
Performance Max targeting based on the 4 personas defined by the business
Our retailer markets products aimed at 4 different personas. These are transcribed in applied to 4 different Performance Max campaigns.
Each campaign has 3 groups of elements, which correspond to 3 audience signals related to the targeted persona:
- a custom segment “Searched on Google XXXX”.
- an “In-Market for product XXXX” segment.
- a list of customers, which corresponds to the persona targeted by the campaign.
A fine adaptation of the elements for maximum customization for each persona
Once the audience signals were created for each Performance Max campaign, the groups of elements were enriched with creative elements (flows, text, images, videos) that were as personalized as possible in relation to the targeted personas:
- For each of the campaigns, the copywriting and products from the Merchant Center feed have been specifically selected for the targeted persona.
- The images used have been personalized to best reflect the lifestyle of each persona, in a very distinct way.
- YouTube videos promoting the products have also been differentiated for each target.
A target ROAS bidding strategy, aligned with Shopping’s historical performance
As far as bidding is concerned, a target ROAS has been indicated in each Performance Max campaign. The latter remained aligned with the historical performance of Shopping campaigns. It is the “normalized” ROAS, over the 2 years of distribution of the Shopping campaigns, which served as the basis for defining the level of profitability to be entered in the auction strategy.
It is noted however that the advertiser, in 2 years, has tested several different bidding strategies. Also, the ROAS of Shopping campaigns – calculated over the entire broadcast period – inevitably experienced ups and downs, over the different bidding strategies used.
Performance Max vs Shopping test results
Comparative results between Performance Max vs Shopping
Clicks / Impressions
Conversions / Clicks
Revenue / Clicks
Income / Conversion
Revenue / Cost
Comments on the Performance Max vs Shopping results
For this advertiser, we can see that the migration of Shopping campaigns to Performance Max is beneficial on almost all of the metrics.
- The ads had a better click-through rate, which implies that they were more attractive than with the standard Shopping campaign.
- In terms of conversion rate, it’s a four-digit increase that proves that Performance Max has been able to generate much more ready-to-purchase traffic.
- Average baskets are also on the rise, and the combined effect with the conversion rate has made it possible to multiply by 4 the monetization of traffic, with revenue per click increasing by +365%.
- Between the ability to attract more clicks, to convert them better, and to higher average baskets, profitability has also experienced a favorable evolution, by +48%.
- Only the Performance Max CPCs proved to be twice as expensive as the CPCs obtained by this advertiser on Shopping.
Moreover, these good performances have a direct impact on the business, since the Performance Max vs Shopping test enabled the advertiser to beat its sales record in a single month, almost reaching the €300k mark.
Why did Performance Max beat Shopping Ads for this advertiser?
Accelerated spending, for rapid learning
By switching to Performance Max, ad serving has accelerated. The system constantly sought to spend the entire daily budget allocated to it (and the campaign was not limited by the budget).
Fortunately for the advertiser, expenses have increased, but so has profitability! By leaving enough budget for the campaign, the algorithm was able to learn faster and without constraints. Thus, he was able to achieve satisfied results quickly.
Couples [audience signals / creative elements] worked strategically
In addition to the unconstrained budget, Performance Max was able to draw inspiration from rich audience signals, carefully thought out upstream, and personalized creative elements for each audience. The characteristics & behaviors of each advertiser’s target persona have been carefully transcribed, and 100% of the texts, videos and images have been personalized accordingly.
In other words, Performance Max’s machine learning was able to rely on audience signals & quality assets. This allowed him to test many combinations [audience/ad], which certainly contributed to the rapid achievement of satisfactory results,
What was the type of audience signal that generated the best results?
- Contrary to what one might think, the groups of Performance Max elements fed with first-party customer lists were not the most efficient.
- Item groups with “Searched on Google XXXX” custom segments had the best profitability.
As can be seen in the table above, the CPC Performance Max has more than doubled compared to the costs per click that the advertiser knew on Shopping.
Thus, the system decided on its own to have the advertiser participate in more premium auctions, which enabled it to obtain better positions (particularly in the Shopping carousel), on Internet users whom the system deemed “super-qualified”.
This approach ultimately paid off, as a more expensive click attracted visitors who converted 40 times better!
What reservations can we have about this Performance Max vs Shopping test?
The multi-network aspect has nothing to do with it: Performance Max has above all “optimized” what the advertiser was doing on Shopping.
When a Performance Max campaign generates more sales compared to a Shopping campaign, the hypothesis that immediately comes to mind is that the additional sales must have come from the display, discovery, youtube networks. included in Performance Max and which do not necessarily exist on Shopping. This is not what happened for this advertiser.
By exploring the Shopping MC ID report, we see that the Performance Max campaigns remained very focused on the Shopping format which weighed for:
- 92% of clicks,
- 96% of conversions,
- 97% of e-commerce revenue generated by the campaign
Apart from Shopping, we can see that:
- Performance Max campaigns only generated 1,480 YouTube views.
- Only 8.7% of impressions were made on the display & discovery network (this represents 0.3% of campaign clicks).
- The search network with the DSA function only accounted for 7.7% of clicks and 2.2% of sales.
Although the multi-network aspect still made it possible to generate 4% of incremental turnover (half of the additional coming from the DSA component of Performance Max and the other half from the Display network), its performance proved lower than that Performance Max had on the Shopping format. The multi-network aspect affected overall profitability:
- the ROAS Performance Max for the Shopping format only was 600%,
- but the overall campaign ended at a ROAS of 590%, showing that networks other than Shopping had lower returns.
Has the machine voluntarily lowered its presence on other networks to achieve the expected performance? This is possible… In parallel with the Performance Max vs Shopping test, it was found that the performance on the search network was far from equaling the Shopping Ads results (certainly because the product is very visual, so it may be logical that the text ads work less well). But is that the real reason why Performance Max voluntarily limited the broadcast on this network? This, Google Ads does not tell us…
The impact of the brand in this success cannot be quantified
During the test period, the advertiser carried out a show which certainly had a positive impact on the awareness of its brand. Brand queries also increased sharply over the test period, which could also suggest that the improvement in profitability observed during the Performance Max vs. Shopping test is also explained by a sharp increase in demand for “ready to convert,” who types the brand directly into Google and sees a Performance Max ad.
Unfortunately, this is an assumption that we cannot verify in a Performance Max world, because Google does not provide the sufficient level of detail to carry out the analysis.
However, on a volume of 100k€ of turnover, one can think that this does not explain the entirety of the success.
- Indeed, on standard Shopping campaigns, brand queries already accounted for only 2.6% of revenue.
- By imagining that this share doubles, it still represents only 5% of income, which seems insufficient to explain the improvement in profitability in its entirety.
Google Ads launched Performance Max campaigns in April 2020 and permanently replaced Smart Shopping campaigns with Performance Max in summer 2022. The objective of this case study was to determine if results were better when comparing Performance Max vs Shopping.
The e-tailer in question had been using standard Google Shopping campaigns coupled with automated bidding strategies for over 2 years and was happy with their results, with 400% profitability and about 1/3 of their e-commerce revenue generated by Google Ads.
To conduct the test, the standard Shopping campaigns were replaced by Performance Max and the test lasted until a turnover of €100,000 was generated by the new campaigns (approximately 3 months). During this time, no action was taken on the Merchant Center offer, site, or feed. Performance Max campaigns were based on 4 different personas and each campaign included 3 groups of elements corresponding to 3 audience signals related to the targeted persona.
Of course, this test is not necessarily representative of all advertisers. But the latter showed a +48% increase in profitability with Performance Max vs Shopping. In conclusion, the test showed that Performance Max made it possible to improve the profitability of the e-merchant without loss of volume, thanks to better targeting of audiences and optimization of bids.