If your ROAS dropped in March 2026, it’s likely that your campaigns didn’t actually perform worse. Meta changed its measurement rules on March 17. What you see in Ads Manager now is a different attribution system than the one you used in February. Comparing these two periods is like comparing two different sets of scales.
This doesn’t mean your campaigns are performing worse. It means the metrics have been recalibrated. Many Colombian advertisers are seeing drops of 15 to 30% in reported conversions in Ads Manager and are cutting their budgets—when in reality, business remains the same. Before making any decisions, let’s understand exactly what has changed.
Click-through no longer includes likes, shares, or saves. Only actual clicks on the link. Those interactions have been moved to the new engage-through category.
Your reported ROAS may drop by 15–30% even if business remains the same. Comparing March to February is like comparing kilograms to pounds.
Check outbound clicks and actual sales in the backend. If they’re stable, don’t adjust the budget. New baseline as of March 17.
If you want the full context on how Meta Ads works, read our guide to Facebook and Instagram advertising in Colombia. And if we manage your campaigns, we’ve already applied these changes to your account.
What exactly
changed on March 17, 2026, at Meta.
Meta summarized the change with this notice in Ads Manager: "Starting March 17, 2026, click results will report website and in-store conversions, such as purchases, only after link clicks. Engagement results will report these conversions after all other ad actions."
Translated into plain English: Previously, when someone liked your ad and then purchased your product within the next 7 days, Meta counted that purchase as a “click” conversion. As of March 17, that like no longer counts as a click. The purchase still counts—but now Meta classifies it under “engagement results,” not under “click results.”
These are three simultaneous changes that took effect together. Let’s go through them one by one, because confusing them leads to wrong decisions.
Click-through attribution:
a new, stricter, and more honest definition.
A “click” for attribution purposes included any interaction with the ad: clicking the link, liking, sharing, saving, commenting, clicking the photo, or clicking the profile name.
If someone saved the ad on Monday and made a purchase on Wednesday → the conversion was attributed to a click-through. Meta Ads Manager reported that purchase. GA4 didn’t see it because there was never an actual click to the site.
A “click” for attribution purposes now includes only actual clicks on the link that take the user to your website, app, or store. Likes, shares, saves, and other actions do not count.
If someone saved the ad on Monday and made a purchase on Wednesday → that conversion no longer appears under click-through. It moves to “engage-through” with a 1-day window, so it doesn’t appear there either. For Meta, that conversion “disappeared.”
The impact is immediate and visible: click-through attribution now requires an actual click on the link. Previously, any high-engagement interaction with a Meta ad—a like, a share, a save, or a comment—could trigger a click-through conversion if the user converted within the attribution window. The result of this change is that conversions reported in Ads Manager go down—even though business remains the same. It’s like stepping on a scale that’s finally been calibrated correctly. You haven’t gained weight—the old scale was simply lying to you.
Engage-through attribution:
the new category replacing engaged-view.
Meta created a new attribution category to capture what was previously mixed in with clicks. It’s called engage-through attribution and replaces and expands upon the old “engaged-view attribution.”
| Interaction Type | Previously (through Mar. 2026) | Now (as of Mar. 2026) |
|---|---|---|
| Actual click on the link → conversion | Click-through attribution | Click-through attribution ✅ same |
| Like → conversion within 7 days | Click-through attribution | Engage-through attribution (1 day) or unattributed |
| Share → conversion within 7 days | Click-through attribution | Engage-through attribution (1 day) or unattributed |
| Saved → conversion within 7 days | Click-through attribution | Engage-through attribution (1 day) or unattributed |
| Watch video for 10 seconds or more → conversion within 1 day | Engaged-view attribution | Engage-through attribution (1 day) ✅ broader |
| Watch video 5s+ → conversion within 1 day | Not attributed (threshold was 10s) | Engage-through attribution (1 day) ✅ new |
| View impression → conversion within 1 day | View-through attribution | View-through attribution ✅ same |
The default setting for the ad set for web conversions is now: 7-day click-through + 1-day engage-through + 1-day view-through. Engage-through is enabled by default—you can disable it by selecting "None" in the attribution settings if you prefer a more conservative report.
Enable it if you’re optimizing for e-commerce purchases—it captures actual conversions from users who interacted with the ad and made a purchase the next day through another channel.
Turn it off if you’re optimizing for leads (forms) or if you’re using very hot remarketing audiences—in those cases, engage-through can inflate the numbers with conversions that would have happened anyway without the ad.
The video threshold has been lowered from 10 to 5 seconds:
what this means for your Reels campaigns.
The third change affects video ads. Meta lowered the threshold for what counts as a meaningful video view from 10 seconds to 5 seconds. The company’s internal data shows that 46% of purchase conversions driven by Reels occur within the first two seconds of view, indicating that users make purchasing decisions much sooner than the old 10-second standard assumed.
In practice: if someone watches your Reels ad for 5 seconds and makes a purchase the same day, that conversion is now attributed to “engage-through.” Previously, it took 10 seconds for it to count. This means that video campaigns—especially on Reels and Instagram Stories—may see an increase in their “engage-through” conversions, partially offsetting the decline in “click-through.”
The creative implication is clear: if the user makes the purchase decision within the first 2 seconds, the first few frames of the video are your actual ad. Time spent on branding, logos, and slow-loading titles at the beginning of a video is no longer a luxury you can afford in Meta 2026.
Why Conversions Dropped on Meta Ads:
Data from 15 Audited Colombian Accounts.
Sectors: fashion, education, services, e-commerce. Budget: $500K–$8M COP/month.
The most important thing to understand about the impact: if the number of actual clicks on the link (outbound clicks) remained stable but your reported conversions dropped, it’s the change in attribution—not your creatives. Don’t cut your budget based solely on dashboard numbers until you’ve verified the actual business performance.
How to check if the change affected you
Compare outbound clicks before and after March 17
In Ads Manager → Columns → Customize → add “Outbound clicks.” If this metric remained stable while conversions dropped, the decline is due to attribution, not actual performance. Outbound clicks only count actual clicks on the link and did not change with the update.
Use “Compare Attribution Settings” in the breakdown
Ads Manager → Breakdown → Attribution. You can see how many conversions come from click-through vs. engage-through vs. view-through. If you see that engage-through accounts for many conversions, consider enabling it in your reporting settings to get a complete picture.
Don’t compare the CPA for March 2026 with the CPA for February 2026. They’re two different measurement systems. Establish a new baseline starting March 17 and compare subsequent periods with each other. According to Zentric Digital, reported ROAS can drop by 15% to 30% due to the attribution change, even if actual performance remains identical.
Why Ads Manager
is now better aligned with GA4 and third-party tools.
Meta’s stated reason for this change is to resolve a long-standing frustration among advertisers: the discrepancy between Meta Ads Manager and third-party tools like Google Analytics. Meta counted all types of clicks leading to conversions, while many third-party tools only counted clicks on the link—creating a reporting misalignment.
Before the change, it was common to see 150 conversions in Ads Manager and 60 in GA4 for the same period. The difference wasn’t due to fraud or error—it was because the two systems had different definitions of “click.” With the new system, both platforms should show numbers that are much closer for click-through conversions.
There’s also a structural incentive on Meta’s part: engage-through is Meta’s preferred attribution model because it feeds more conversion signals into its ad algorithm. The more conversions attributed to a campaign—even via social interactions—the more data the system has to work with, making it easier to identify the user behaviors that drive results and justify scaling up budgets. Meta has a vested interest in advertisers adopting “engage-through”—it provides more data to the optimization algorithm.
Real errors we’re seeing
in Colombian accounts following the update.
In recent weeks, we audited more than 15 Meta Ads accounts in Colombia affected by the change. These are the three most common—and most costly—error patterns:
They compare March to February without adjusting the measurement system
80% of the accounts we audited were using the February 2026 CPA as a benchmark to evaluate March 2026. It’s like comparing weight in kilograms to weight in pounds and concluding that you’ve gained weight. Two different systems, incomparable numbers.
Our recommendation: Use March 17 as the cutoff date. Everything before that is history in the old system. Everything after that is the new system. Compare only within each system.
Disabling engage-through in e-commerce campaigns out of fear of inflated numbers
Several clients read about the change and disabled “engage-through” across all their campaigns to “get cleaner data.” The result: the algorithm lost conversion signals, the learning period reset, and the actual CPA went up. Engage-through doesn’t inflate the numbers—it categorizes them differently.
Our rule of thumb: engage-through ON for e-commerce purchases, engage-through OFF for leads and high-temperature remarketing. There’s no one-size-fits-all answer.
Not all businesses are affected equally
—it depends on the type of campaign.
| Business/Campaign Type | Impact on reported ROAS | Actual impact on sales | Priority Action |
|---|---|---|---|
| E-commerce (direct purchases) | High — 20–35% drop | None if CAPI is active | Maintain budget, check backend |
| Lead generation (forms) | Medium — 10–20% drop | Low — stable actual leads | Consider disabling engage-through |
| Remarketing to hot audiences | High — 25–40% drop | Low — sales unchanged | Disable engage-through, review attribution |
| Branding and awareness | Low — 5–10% drop | No measurable impact | No urgent changes needed |
| Videos / Reels with a conversion goal | Moderate — but may increase | Can be improved (5-second threshold) | Enable engage-through, strengthen the hook in the first 5 seconds |
Immediate Action Guide:
What to Adjust in Your Account This Week.
- Set a new baseline: Use the 30 days prior to March 17 as a separate historical reference. Starting March 17, begin a new data series. Do not mix them.
- Enable outbound clicks as a permanent column: this is the most accurate metric for measuring users’ actual intent. If this metric is stable, the problem lies with attribution, not performance.
- Decide on engage-through based on your goal: enable it for e-commerce purchases, disable it for leads with hot remarketing. There’s no one-size-fits-all answer.
- Adjust your video evaluation: the threshold has dropped to 5 seconds. Analyze your video creatives—if the first 5 seconds don’t have the right hook, you’re losing engage-through conversions.
- Reconcile with your true source of truth: actual sales in the backend, leads in the CRM, incoming calls. Ads Manager is an approximation—not the only source of truth.
- Set up CAPI if you don’t have it: Meta’s Conversions API captures between 20% and 40% more conversions than the Pixel alone, according to Zentric Digital. With the new, stricter attribution system, having CAPI properly configured is even more critical.
- Check with your agency: if you’re unsure how the change affected your specific account, now is the time to conduct an audit of your Meta Ads account. At Webhexup, we do this for free for active accounts.
The most common questions
from Colombian advertisers in 2026.
Frequently Asked Question
Why did my conversions on Meta Ads drop in March 2026?
Conversions dropped because Meta changed its attribution system on March 17, 2026, removing interactions such as likes, shares, and saves from the click count. This reduces the number of conversions reported in Ads Manager without affecting the actual performance of the campaigns. To verify this, compare outbound clicks before and after March 17 and check actual sales in your backend.
Frequently Asked Question
What is the Meta Ads 2026 attribution change?
The Meta Ads attribution change in March 2026 consists of three simultaneous modifications: (1) click-through attribution now counts only actual clicks on the link, excluding likes and shares; (2) “engage-through” attribution was introduced—a new category for social interactions and video views—replacing the former “engaged-view”; and (3) the threshold for a “significant” video view was lowered from 10 to 5 seconds. Billing remains unchanged—only the way conversions are reported has changed.
Frequently Asked Question
Should I be concerned if my ROAS dropped in Meta Ads in March 2026?
Not necessarily. The ROAS reported in Ads Manager may drop by 15% to 30% due to the attribution change, even if actual sales haven’t changed. Before cutting your budget, check whether outbound clicks are stable and compare actual sales in your store or CRM. If business is the same and only the number in Ads Manager has dropped, it’s a measurement adjustment—not a performance issue.