TL;DR: Retargeting should support a specific purchase decision and add value for the store. Separate carts, product interest and repeat buyers; inspect events and the offer; read attributed ROAS alongside margin and additional effects. There is no universal percentage of “wasted” budget that establishes your result.
The campaign returns visitors and reports good ROAS. Some were already planning to buy: they check a size, wait for delivery or reopen a cart. You see sales in the ad report without yet knowing how many the advertising added.
Canem Errant examines the route from product view to paid order. Our position: a cheap attributed retargeting purchase is insufficient reason to increase spend. That decision needs store context, costs after returns and investigation of the additional effect.
Why should the existing retargeting structure be reviewed?
“All visitors, 30 days” combines different situations: a recent product comparison, an incidental homepage visit and a completed purchase. One message need not suit them all. A broad audience can still be a useful starting structure or comparison group when data is limited.
Available events are not every customer action
User choice, browser restrictions and the integration affect which events reach advertising. App Tracking Transparency and website cookie consent are different questions. A general iOS opt-out rate cannot establish your store’s missing-order percentage.
Compare available events with actual orders. Check what you intend to measure, when it is transmitted and under which conditions. Server-side tracking can complement an existing transmission without removing those conditions. Our tracking guide explains the architecture.
Campaigns can reach the same people
An Advantage+ Sales campaign can cover different purchase stages. A dedicated visitor audience does not mean only that campaign reaches people familiar with the store. Meta’s public audience course covers Advantage+ and overlap.
Review each campaign’s role and currently available constraints. Overlap proves neither payment twice for one impression nor an automatic CPM increase. It may prompt simpler structure and measurement, but the economic conclusion requires more than two similar audience names.
The offer and material may miss the actual objection
Another ad impression cannot fix missing stock or unexpected delivery cost. Investigate why buyers do not complete and whether the material answers their question. There is no established universal fourth- or eighth-impression point beyond which all purchase probability disappears.
Which audiences and windows could you investigate?
Start with the person’s action and the next useful message. These five groups form an inventory, not a requirement for five ad sets. Limited data may justify a simpler setup.
| Audience | Example window and question |
|---|---|
| Cart without confirmed purchase | Seven or 14 days, for example: explain sizing, delivery or completion; exclude current purchases where correctly available. |
| Viewed a product | Seven or 14 days: explain features, demonstrate use or provide a relevant comparison. |
| Actively explored the store | For example seven days with a measurable interest signal. Three pages or two minutes can be clues, not proven purchase intent. |
| Already purchased | For example seven to 60 days when a repeat offer or cross-sell is appropriate; consider delivery, returns and use of the previous product. |
| All visitors in a month | A simple starting or comparison audience; evaluate composition and additional effects instead of banning it by default. |
Interior doors for a renovation have a different purchase cycle from a low-cost accessory. Study time from interest to payment. Day eight after a view does not automatically mean zero interest. Even a 180-day audience needs evaluation against its purpose; its name alone does not establish waste.
Consider overlap and updates. A missing Purchase event does not guarantee no purchase happened; the event may be absent. Exclusions need correct available data and permissible use. Collecting more personal data simply to enlarge the audience is not a general solution.
How should you control ad frequency?
Frequency describes average impressions per reached person over the selected period. An average does not mean everyone received that number. Read it alongside audience size, period, spend, clicks and paid orders.
Check the available controls first
Frequency controls depend on buying type, objective and campaign settings. Inspect the actual account before promising “one impression daily.” Meta discusses auction, reservation and frequency in its buying and delivery course.
A budget limit is not a strict personal frequency cap. An engagement filter does not establish that a person saw the ad exactly five times either. Specific account controls have not been verified here.
Respond to causes and outcomes
Carts, product viewers and repeat buyers may need different messages and contact intervals. Compare equal periods to see whether more contact accompanies better orders or higher costs. Three to five impressions weekly is not a universal profitability limit.
Check budget relative to audience, new material, purchase cycle and campaign composition. Choose a bounded change and observe its result. One current average frequency value cannot establish fatigue or waste by itself.
Catalog ads, static ads or UGC?
A catalog ad can show relevant products when catalog and events agree. An image or video can explain material, sizing, conditions or use. UGC needs accurate statements and advertising rights. None automatically wins for an audience segment.
| Criterion | What to inspect in the format |
|---|---|
| Unfinished cart | Does a product reminder help, or is an answer about delivery and sizing missing? |
| Product interest | Is the specific item enough, or does the person need a demonstration? |
| Repeat purchase | Does an additional product fit the previous order and its delivery? |
| Message control | What can the catalog template communicate, and what needs separate material? |
| Preparation | Catalog: prices, images, stock, links and matching IDs. Separate material: production and updates. |
| Assortment size | Automation and manual production have different costs; compare the actual task. |
Pay particular attention to product IDs, current price and destination. A credible delivery promise is more useful than invented urgency or “free shipping today” that the store does not provide. Our creative testing system helps compare materials; catalog ads are not restricted to cart audiences.
What do Canem Errant’s internal cases show?
In our ongoing engagement with Czech online retailer Erkado, work combines server-side event tracking, segmented campaigns across three distinct goals (e-commerce sales, B2B leads, and showroom hiring), and mobile checkout UX adjustments guided by Clarity observations. This demonstrates that reliable tracking infrastructure establishes data clarity, but works in concert with campaign management rather than driving isolated revenue leaps alone.
In our Fashion demonstration framework, structured alignment between creative messaging, on-page sizing guidance, and transparent checkout terms is modeled end-to-end. This highlights that trust elements and cohesive ad angles work in concert, rather than acting as isolated performance guarantees.
Internal records are neither independent event exports nor controlled format experiments. They explain why measurement and advertising deserve joint investigation; they do not promise your retargeting result. Our CPA guide explains the economic assessment.
How should you combine retargeting and Advantage+ Sales?
Choose a clear role and comparison method:
- Simple starting structure: one sales campaign with relevant material when data and budget are limited. It makes the overall result easier to assess.
- Separate retargeting task: a bounded cart test with its own message when there are enough data, materials and a measurable hypothesis.
- Controlled effect comparison: an appropriate control group when volume and available tools permit investigation of additional purchases.
A €5,000 monthly threshold does not decide the choice by itself. Check current audience settings and treatment of existing customers. Website visitors are not automatically previous buyers. Older existing-customer budget parameters may be absent or have a different scope; inspect the current interface rather than following an old instruction. Our Advantage+ guide supports preparation without confirming private account features.
Lookalikes also serve a different purpose from returning to known visitors. A similar audience does not mean everyone has previously visited your store. Organize by task rather than a prescribed campaign name.
Which metrics support a budget decision?
ROAS and store outcome: The account attributes revenue using its reporting rules. Reconcile period, payment, cancellations and returns. High revenue can leave little margin after discounts and delivery.
Additional effect and iROAS: Incremental revenue is the difference against a comparable situation without the advertising effect; iROAS relates that revenue to the relevant ad spend. A reliable estimate requires suitable design and data volume. Meta’s incrementality course describes different methods and prerequisites. A simple pause can mix seasonality or stock changes with advertising effects.
Frequency and overlap: Review audience size, contact development and available measurement. A frequency of four or 30% overlap does not prove loss. A particular menu path is not a guaranteed current feature. Our Meta KPI guide supports the combined reading of purchase costs, order value and the measurement basis.
What should work in CAPI and event quality?
Check event, time, amount, currency and relation to a confirmed order. Pixel and CAPI can transmit the same purchase together when deduplication works. EMQ 8.7 does not mean exactly 87% of all orders were captured or matched, nor a 2.7-times-larger audience.
Matching, coverage, freshness and deduplication are separate questions. Server transmission must reflect user choice and does not automatically recover all unavailable data. Review the existing integration before adding infrastructure. See CAPI setup and CAPI versus Pixel.
What could a three-task plan look like?
Main campaign: evaluate offer and materials for the available assortment; choose appropriate targeting and delivery using current account options.
Optional cart campaign: answer a separate question about an unfinished order. Seven or 14 days can be test windows, not a proven universal maximum. Compare a relevant product reminder and explanatory material where useful.
Optional repeat purchase/cross-sell: offer a relevant follow-up after the previous order arrives and another purchase makes sense. Seven to 60 days, or exclusion of the first seven days, are planning hypotheses rather than requirements.
Budget example: Out of €1,000, €750 could go to the main task, €150 to carts and €100 to repeat purchase: €1,000 in total. If the third task is not useful, it needs no allocation. Instead of independently combining percentage ranges, your plan needs a total, role and condition for revision. Three campaigns are not always better than one.
What should you investigate this week?
- Map audiences by action, window, volume, purchases and next message. Mark groups that can usefully be combined.
- Compare frequency, materials and paid-order costs over the same period. Choose one question for a bounded test.
- Inspect events and catalog: confirmed purchase, deduplication, price, stock and link. A new creative cannot replace a fix to payment or incorrect product data.
Frequently Asked Questions
How much retargeting spend pays for purchases that would happen anyway?
You cannot establish that for a store from a universal percentage. Attributed and additional purchases differ. An appropriate control group and comparable data help; with limited volume, uncertainty about the effect remains explicit.
Which retargeting segment should be reviewed first?
A cart often provides a clear question. First inspect its volume, event and purchase obstacle. Product views may suit a longer decision cycle; a previous order may inform repeat purchase. No audience is best for every store.
Are Dynamic Product Ads or UGC better?
Catalog ads can remind someone of a current item, while UGC and separate images can explain its features. Compare them for the same task, with cleared rights and accurate availability. The internal cases do not establish a universal format winner.
How can I avoid unproductive overlap with Advantage+?
Check roles, current audience and customer settings, available exclusions and store outcome. Keep a separate campaign when it has a measurable distinct task. Overlap is not automatic evidence of paying twice.
Does retargeting always need a new server-side setup?
It needs correct available events and a permissible data process. CAPI can complement the Pixel without guaranteeing an audience or sales. Inspect the existing integration and specific fault before commissioning more infrastructure.
Bring campaign structure, spend, revenue after returns and available events. We can discuss the retargeting question worth investigating; further diagnosis and implementation are agreed separately. Discuss the task →
