ID Interactive agency logo

Google Ads in e-commerce: finding the real value behind each conversion

Published: 01.09.2026

Last modified: 19.11.2026

This case illustrates a step that is too often overlooked in e-commerce: a shop may know perfectly well how many sales it makes, without knowing which of its acquisition sources is really responsible for them, let alone what value each one brings.

Google Ads

This was the situation of an e-retailer specialising in household appliances and bedding before our agency took over the management of its Google Ads account.

This case illustrates a step that is too often overlooked in e-commerce: before optimising bids for a value objective, make sure that value is reported correctly, with no duplicates or approximations.

The client

An e-retailer specialising in household appliances, with an online business complemented by a physical shop.

The challenge: counting sales, but ignoring their real advertising value

The account reported two conversion sources in parallel: a native Google Ads conversion meant to capture the exact value of each order, and an import from Google Analytics 4 which, in practice, carried almost all of the measured value. The client could compare sales volumes between organic and paid traffic, but the native Google Ads conversion reported almost no usable value, which completely skewed any value-based bidding. Store visit conversions, which were purely modelled, added further noise to the whole.

The strategy: make value measurement reliable before optimising bids

The first action was to fix the configuration of the native conversion so that it systematically used the actual value sent by the site rather than a default value, finally giving meaning to the figure reported in Google Ads. The conversion imported from Google Analytics was then reclassified as observation only rather than a primary action, to avoid double counting between the two sources and let the native measurement become the reference for bid management. Store visit conversions, considered too noisy for e-commerce management, were removed from the account goals. Once this foundation was reliable, the account was reorganised into several Performance Max campaigns with distinct logics: one focused on the product catalogue, one dedicated to remarketing to already-engaged audiences, and a third positioned against direct competitors, each with its own budget. A large account-level exclusion list was also built, excluding the major competing retailers as well as all searches with no purchase intent (after-sales service, instructions, faults, recruitment, reviews).

The results: ROAS doubled, without spending more

Google Ads, by campaign

Cost before

Cost after

Conversion value before

Conversion value after

ROAS before

ROAS after

Performance Max

€4,244

€4,128

€103,769

€139,120

x24.4

x33.7

Search

€609

€591

€5,039

€15,635

x8.3

x26.5

Account total

€4,853

€4,719

€108,808

€154,755

x22.4

x32.8

Over the period observed, the total account cost fell slightly (-2.8%), while the conversion value attributed to Google Ads rose by 42%, taking overall ROAS from x22.4 to x32.8, an improvement of 46% with no budget increase. This progress comes from two distinct levers, one on measurement, the other on campaign optimisation.

On measurement. Before our intervention, the value reported in Google Ads relied almost entirely on an imported secondary source, not designed to drive automated bidding, while the site's native conversion, the one Google Ads actually uses to optimise, reported almost no usable value. Once this native conversion was fixed to send the real value of each order, and the secondary source reclassified as observation only, the bidding algorithm finally had a reliable signal to work with.

On optimisation. This more reliable data made it possible to recalibrate target ROAS goals on a real basis rather than on a distorted value, to remove store visit conversions that added noise unrelated to the e-commerce goal, and to restructure the account into several Performance Max campaigns with distinct logics (product catalogue, remarketing, competitive positioning), each managed with its own budget and target ROAS. Search, in particular, illustrates the combined effect of both levers: with an almost identical budget (-3%), its conversion value tripled and its ROAS rose from x8.3 to x26.5, once bidding could rely on accurate data at last.

Google AdsGoogle Ads

What this says about the method

This case is a reminder that a conversion figure only makes sense if its source and value are reliable. Before managing bids on a value objective, as Performance Max naturally does, you need to make sure that a single source is the reference, that this source reflects the real value sent by the site, and that conversions with no real value are excluded from the calculation. It is a check we apply systematically whenever we take over an e-commerce account, before even touching budgets or bids.

Key figures

ID Interactive, our agency based in Vannes, increased overall ROAS from x22.4 to x32.8 (+46% performance) over a comparable analysis period and with a constant budget (-2.8% spend) for an e-retailer in household appliances and bedding, solving a problem of distorted conversion value measurement and skewed Google Ads bidding.

Conclusion

This result did not come from a budget increase, but from an overhaul of the measurement itself. By clarifying which conversion source is authoritative and eliminating duplicates, we enabled the account to finally reflect the real value generated by each euro invested, and therefore to manage bids against a genuinely reliable objective rather than a misleading volume. This is the rigorous checking we systematically apply at the start of every e-commerce account engagement.


Related case studies: rebuilding Google Ads conversion tracking, the Google Ads restructuring of a food supplement manufacturer and the Google Ads and Meta Ads campaigns of a figurine e-commerce shop.

Frequently asked questions

Parce qu'un nombre de conversions élevé peut cacher des doublons entre plusieurs sources de mesure, ou des actions sans valeur commerciale réelle. Une fois ces éléments retirés, le chiffre restant est plus bas, mais bien plus proche de la réalité.

Parce qu'un algorithme d'enchères basé sur la valeur ne peut optimiser correctement que s'il dispose d'un seul signal de référence. Laisser deux sources actives revient à lui donner des informations contradictoires sur ce qui doit être maximisé.

Cela dépend de l'objectif réel du compte. Si le pilotage vise avant tout les ventes en ligne, des conversions modélisées et peu fiables comme les visites en magasin ajoutent du bruit sans apporter de valeur exploitable pour les enchères.

Parce que ces trois audiences n'ont pas le même comportement d'achat ni le même niveau de maturité face à la marque. Les séparer permet d'allouer un budget et un discours différents à chacune, plutôt que de les mélanger dans un seul pilotage moyen.