【Facebook Ads】How to Address Issues of Excessive and Low-Quality Conversion Data in Facebook Ads

One challenge Meta advertising optimizers face is that conversion data often appears inflated and of lower quality compared to actual conversion metrics. This article will guide advertisers in identifying whether conversion results are overstated or skewed. Specific troubleshooting methods vary depending on the campaign objective.

Here, I’ll break down the process for three distinct advertising goals:

Sales

Leads

Top-of-Funnel Events (Link Clicks, Landing Page Views, ThruPlay, Post Engagement, etc.)

Sales

Two primary factors contribute to inflated sales conversion data:

– Conversion Event Setup Issues

– Misinterpretation of Attribution Results

Conversion Event Setup Issues

Attribution (how Meta assigns conversions to ads) critically depends on correctly implementing conversion event code. The reports we see depend entirely on whether our Meta conversion codes are configured correctly. Issues with inflated or deflated reporting data often trace back to this stage.

1. Is the event triggering at the intended time? Meta should not receive purchase notifications before the transaction is complete. If events trigger prematurely (e.g., during content views, cart additions, or checkout initiation), conversion errors occur.

You can troubleshoot this in several ways. First, locate the purchase event on the Data Sources page in Events Manager, then click “View Details.”

Click “Dynamic Sample” to see a summary of recent URLs triggering this event. Are any of these URLs incorrect?

Use the “Test Event” feature in the Event Manager to test the normal customer purchase journey from the landing page to completing the purchase. Verify that each step triggers the correct event.

2. Have events been deduplicated? If you send conversion events from both your website and offline (or CRM) sources, ensure they are deduplicated to prevent double counting. This is typically achieved using an event ID that performs a meta-match between the two events, allowing one to be ignored.

When viewing an event’s “More Details,” the “Event Deduplication” section will be displayed. If Meta detects any deduplication issues, you should see an alert in the “Diagnostics” section of the Event Manager.

Incorrect Attribution of Results

Even if all settings appear correct, you may notice discrepancies in your conversion results. Typically, you can check this by following these steps:

1. Compare attribution settings. In Ads Manager, click the “Columns” dropdown menu and select “Compare Attribution Settings.”

Then select all click and view attribution options (especially 1-day clicks, 7-day clicks, and 1-day views).

This will generate separate column reports for each attribution model.

If daily view-through conversions are excessively high, it indicates your ad conversion performance is being overstated. While view-through conversions aren’t entirely worthless, their journey likely represents only a minor part of the customer conversion process. The best metric for measuring an ad’s contribution to sales is daily click-through conversions.

A high volume of view-through conversions typically indicates the ad is being used for remarketing. Many people exposed to the ad may have made a purchase because they received an email that same day.

By comparing attribution settings, you can gain a more accurate understanding of the ad’s impact on different segments.

2. First-Conversion Report. We can choose to view the “First Conversion” report.

The 7-day click attribution window can capture multiple independent conversions from the same customer. For example, someone might click your ad and make a purchase on Monday, then make another purchase on Thursday. Both purchases would be reflected in your ad management tool’s reports. While technically the ad contributed to both purchases (the second wouldn’t have happened without the first), this does inflate sales data.

First-conversion attribution helps address similar issues as described above.

Leads

If the number of leads reported in your ad reports also appears inflated, consider these three potential causes:

1. Conversion event issues. If leads are generated through website forms, troubleshoot inflated sales data by reviewing all steps outlined in the sales objectives above. Conversion events may be triggered at the wrong stage, or events may not be deduplicated.

2. Attribution interpretation. The default attribution setting is “7-day click + 1-day view,” meaning the advertising platform reports all conversions occurring within 7 days of an ad click or within 1 day of an ad view.

For the vast majority of potential customers, a 1-day click attribution window is sufficient in most cases. Any other attribution model risks inflating conversion results. Use comparative attribution settings to distinguish 1-day click conversions and focus on observing “first conversion” data to avoid unnecessary data inflation.

3. Age as a Key Algorithm Signal. Meta’s ad delivery algorithms leverage various signals to deliver more desired conversions for advertisers. I’ve found age to be one of the most significant signals. You can use the “Segment” reporting feature to break down conversion results by age group.

The report will generate separate rows for each age group. Do you need to check whether the budget is concentrated on a specific age group? Is the audience in this age group your core audience?

Based on extensive experience with long-term performance advertising, I’ve observed that inexpensive leads often originate from audiences aged 65 and above. If you’re receiving a high volume of low-quality leads from this demographic, you should restrict your demographic targeting to avoid these subpar conversions.

Unfortunately, when Advantage+ Audience is enabled, the upper age limit is merely a system suggestion rather than a restriction. The algorithm remains free to allocate budgets as it sees fit. Meta’s ad delivery algorithm can override the age cap if it determines doing so will generate more conversions for the advertiser. If you encounter issues with age concentration, you must disable Advantage+ Audience and use the original audience targeting to set strict age limits. I recommend doing this only when absolutely necessary, as it restricts the ad algorithm.

4. Gender Concentration in Algorithms. The algorithm also leverages other signals to help optimizers acquire low-cost leads, such as gender signals. While I generally don’t consider gender to be a major issue, we can use the “Segment” feature to break down conversion reports by gender.

If your business primarily serves women but your ads are generating a high volume of low-quality leads from men, you need to address this issue immediately. Similar to age cap issues, gender is only suggested as an audience when Advantage+ Audience is enabled. To target a specific gender option, you must use native audience targeting and disable Advantage+ Audience.

5. Algorithm country/region concentration. If you target multiple countries/regions or use “Global” targeting, the algorithm may exploit this to generate low-cost leads. Use the “Segment” feature to break down conversion reports by country/region.

If you’re receiving a high volume of low-quality leads from unexpected countries or regions, you should consider restricting your geographic targeting.

6. Lead Quality Issues.

Remember, Meta’s ad delivery algorithm continuously learns from attributed conversions. We need to prevent low-quality leads from submitting forms. We can address this by making ad copy and creatives more targeted to attract the ideal audience (and deter the undesirable ones). We can also use Instant Forms with conditional logic to exclude submissions based on how users answer questions.

Top-of-Funnel Events

By “top-of-funnel events,” I refer to shallow-level events beyond purchases or leads, including link clicks, landing page views, ThruPlay views, post interactions, and so on. These can be manually selected within performance goals.

I would not recommend using these events, as they hold no value for the business. However, if you insist on running this ad, be aware that the seemingly too-good-to-be-true results likely conceal underlying issues.

1. Segment by ad placement. The algorithm’s sole objective is to maximize conversions for advertisers within their budget.

Use segmentation to break down conversion results by ad placement.

If you optimize for link clicks or target page views, Meta will likely use Audience Network placements to drive low-quality traffic.

If you optimize for ThruPlay views, a significant proportion of impressions are expected to be allocated to rewarded videos, which exchange virtual currency on third-party apps for rewarded views. The ad placements for rewarded videos are also Audience Network.

In either case, you’ll see a lot of worthless ad clicks or views.

2. Segment by age, gender, and country/region. We’ve already covered this in audience segmentation, so we won’t delve into it here. Using the “Segment” feature allows you to see how your ad budget is spent and which audience segments are performing conversion events.

The age concentration issue we observed in lead generation ads may also appear in top-of-funnel conversion event targeting. We need to disable Advantage+ Audience and set an upper age limit.

Gender typically isn’t a primary concern, but for advertisers targeting predominantly male or female audiences, the algorithm should recognize female/male preferences when optimizing purchase strategies. However, top-of-funnel event optimization—especially for engagement events—often exhibits bias because the algorithm prioritizes cheaper traffic. As with lead generation ads, exercise caution when selecting countries/regions for top-of-funnel event ads.

3. URL Parameters and Third-Party Reporting. Beyond tracking conversion outcomes, URL parameters provide additional insights not reported by ad management tools. Combine URL parameters with GA4 or other third-party reporting software to understand user engagement metrics on your website, including time spent, bounce rates, and conversion performance.