CSS Incrementality: How to Measure It, Read the Data, and Make the Case
Can you prove that your CSS revenue is truly incremental?
Incrementality is the core value of the CSS model. A strong partner expands Shopping coverage, captures search demand that the merchant's own campaigns miss, and generates sales that would not have occurred otherwise.
The question is how to demonstrate that in the data. This article explains which reports, metrics, and tests make the case, and how to present the results internally.
Where the CSS incrementality question comes from
CSS incrementality is a question of budget efficiency. E-commerce teams need to know which activities generate genuinely additional sales, and where spending produces the highest output per euro.
The question is especially relevant for CSS because the partner promotes the merchant's own products in Google Shopping, often alongside the merchant's existing campaigns and, in some cases, other CSS partners. From the outside, this can look like several routes into the same demand. It also raises an understandable concern: will adding a CSS partner cause the merchant to compete against itself and drive up CPC?
No, Google deduplicates bids at the merchant level, so only the highest-ranking bid enters the auction. The merchant’s own bids and CSS partner bids are treated as coming from the same business rather than as competitors. Read more about CSS cannibalisation vs deduplication here.
Once this concern is addressed, the incrementality question remains: is the CSS partner expanding Shopping coverage, or reporting sales that the merchant's own campaigns could have captured anyway?
This is part of a wider shift in how performance marketing is measured. Channel dashboards report on attributed performance: orders, revenue, and ROAS associated with a campaign or partner. Finance teams increasingly want the next layer: how much of that result was additional to what the business would have achieved without the activity.
Measuring CSS incrementality as a framework
CSS incrementality becomes more tangible when different types of evidence are assessed together. No single metric gives the full picture.
Three methods draw on reporting data: auction overlap, query-level analysis, and new customer rate. Each looks at a different dimension of the partner's activity, and together they show how much of the reported performance came from demand that the merchant was less likely to reach independently.
The fourth method, geo-experiments, adds the causal layer. Rather than interpreting reporting data, it tests what actually changes when CSS activity is temporarily removed from a sub-region.
Below, we break down what each method shows, what data it requires, and how it feeds into budget decisions.
1. Auction overlap analysis
What it shows: what share of CSS impressions came from auctions your own campaigns did not cover.
Auction overlap is the most direct way to assess incremental Shopping coverage. It measures how often CSS impressions appeared in auctions already reached by your own campaigns. The remaining share is your incremental auction coverage:
Incremental auction coverage = 100% − overlap rate
If the overlap rate is 25%, then 75% of the partner's impressions appeared outside the merchant's own campaign presence. In general, a low overlap rate is a good indicator of a high level of incremental performance being delivered by your CSS partner.
What to ask: ask your CSS partner what percentage of impressions appeared in auctions where your own campaigns were absent. For a more useful view, ask for the same breakdown by market, category, and outcome, so the data shows where additional visibility is converting into qualified traffic and sales.
How to read it: lower overlap generally means stronger incremental coverage. The target is not zero, though: some shared auction presence is normal when both sides promote the same catalogue in one channel. What matters is whether the non-overlap share is meaningful and whether it leads to clicks and orders.
One caveat on scope: impression share and overlap are only calculated on the products your CSS partner actually bid on and won impressions for, not your full Shopping feed. A partner running a narrow, high-bid strategy on a small set of products can post very high impression share on that slice while driving little overall revenue. Read these metrics alongside traffic quality, conversions, revenue, and total incremental value, never on their own.
2. Query-level analysis
What it shows: which search terms triggered CSS impressions, clicks and orders.
Query-level analysis goes deeper than campaign-level reporting. Because Shopping campaigns match queries through product feed data rather than keywords, the searches a CSS partner captures are not always predictable in advance. The report shows the actual demand the algorithm reached.
What to ask: ask your CSS partner for a search term report that connects queries to impressions, clicks, products shown, and resulting orders, with enough detail to compare CSS activity against your own campaign presence, ideally by market, category and product group. Read it for net incremental orders and revenue across the account, not query-by-query visibility. Before cutting anything, including brand terms, check whether it's a defending position you'd otherwise lose to marketplaces and other advertisers.
How to read it: start with the type of queries the partner is capturing and where they sit relative to your own activity. Incremental value shows up in two ways. The first is reach you didn't have: long-tail searches, product variants, lower-priority categories, or markets where your own campaigns run thin. The second is added strength in auctions you already contest, including brand queries where marketplaces and other advertisers bid against you. Both grow the sales connected to your catalogue.
Where those clicks came from matters as much as whether they converted. If the report shows converting queries that your own setup had limited or no presence on, the partner is likely expanding the search opportunities connected to your catalogue.
A weaker signal is when most clicks and orders come from queries your own campaigns already dominate, with high impression share and little headroom left to win. Even here, the partner can add value: covering moments your bidding pulls back, holding share when competitors push, and defending a position you'd otherwise concede.
3. New customer rate
What it shows: what share of CSS orders came from customers who had not purchased from the merchant before.
New customer rate adds a customer-level view to the incrementality question. If a meaningful share of CSS orders comes from first-time buyers, the partner may be reaching shoppers who were not converting through the merchant's existing channels.
What to ask: ask your internal analytics team what percentage of CSS orders came from first-time customers. Compare this with your own Shopping campaigns, paid search, affiliate activity and other acquisition channels.
How to read it: a higher new customer rate is a positive signal, especially when the acquisition cost is competitive with other paid channels. It suggests the CSS partner may be reaching shoppers your existing activity was less likely to convert.
A new customer is not automatically an incremental buyer, since that shopper may still have found the merchant through another route. New customer rate should be treated as a quality signal, not as a standalone proof of incrementality.
4. Geo-experiments
What it shows: how sales change when CSS activity is active in one set of regions and held out in another.
Geo-experiments test incrementality by comparing similar regions under different CSS conditions. One group continues with CSS activity; the other is paused. The sales difference provides the clearest view of the impact the partner has.
Test and control regions need to be matched before the test begins: comparable in basket value, sales volume, conversion rate, and audience composition. Without that baseline, the result has no reference point.
The test also needs a defined timeframe and a success metric (sales lift, conversion lift, or cost per incremental sale). Avoid periods where promotions, stock issues, or seasonal spikes could distort the comparison.
What to ask: ask your CSS partner whether they can support a geo-incrementality test and help define clean test and control regions.
How to read it: compare the CSS-active regions against the held-out regions, controlling for how each group normally performs. If both groups typically grow at a similar pace but the CSS-active regions outpace the control during the test period, the difference is a direct estimate of incremental sales impact.
Getting started with measuring incrementality
You don't need a perfect measurement framework to begin. Start by reading four things together, and the picture builds from there.
- Reported performance is the base layer: confirmed orders, revenue and CPA from the partner's tracking. This tells you what the activity generated, but not yet how much of it was additional.
- Incrementality evidence is where you test that. Look at auction overlap and incremental coverage, converting search terms in areas your own campaigns reached thinly, and geo-experiment lift where you've run it. Read this in the spirit of the earlier sections: overlap is not lost value. Because Google deduplicates at the merchant level, a partner active in auctions you already contest is winning placements on your behalf, not bidding up your own CPCs. The question is how much they add by covering pullbacks, holding share against competitors, and catching the gaps inside queries you already run, not whether any overlap exists at all.
- Sales quality shows what that activity is worth over time: new customer rate, first-order value, and what each new customer costs to acquire. A partner bringing in first-time buyers is doing different work than one re-converting your existing demand.
- Cost is the last layer: commission paid, cost per sale, and cost per incremental sale. The final figure needs the incrementality evidence above to calculate, and it's the truest measure of what you paid for genuine growth.
- Read together, these answer the core questions about any CSS partner: what was reported, how much was additional, and at what cost. For the full picture, hold the results up against your other acquisition channels.
Shoparize Managed Ads: incremental Shopping Revenue at a pay-per-sale model
Measuring CSS incrementality requires a partner who makes the underlying data available. Shoparize Managed Ads is built on that premise.
Shoparize is a Google Premium CSS Partner. Through Managed Ads, we help merchants promote their products across Google Shopping, Bing Shopping and product placements on Shoparize.com on a performance-based model. The merchant sets the CPA, pays commission on confirmed sales only, and retains full ownership of customer data and traffic.
We work closely with merchants to grow incremental coverage and sales, with clear reporting throughout the process. More than 25,000 merchants across 21 European markets use Shoparize to expand product visibility and sales. Get started with Shoparize Managed Ads today.
