GO TO WAR
Military.com — Events Manager Walkthrough
Source: events_manager_walkthrough.md

Events Manager Diagnostics Walkthrough

Tuesday morning, 15 minutes before the meeting. This is the doc you follow click-by-click.

The whole audit's biggest finding — the Purchase value pipe being broken — was validated indirectly via the Marketing API (Meta returns purchase_roas: "Not available" everywhere). The direct evidence lives only in Events Manager Diagnostics. This walkthrough gets you to it.

Why this matters

The API confirms: - ✅ Purchase events fire (~755/day) - ✅ EMQ on Purchase is excellent (9.1) - ✅ fbc parameter has 87% coverage on Purchase, 100% on CompleteRegistration - ❌ But purchase_roas: "Not available" everywhere

That last point means something in the value pipe is broken. The API tells us "value optimization isn't working" but not why. Events Manager Diagnostics tells us why — specifically, whether the value parameter being sent is malformed (Mo's "98% of price data" claim).

If you confirm value malformation Tuesday morning: - The root cause is locked in - The Auto Max Val decision becomes data-supported, not inference-supported - The Tuesday dev conversation becomes "fix this specific thing" not "investigate this hypothesis"

Step-by-step

Open Events Manager

  1. Go to business.facebook.com
  2. Sign in with the Iron Corp / Military.com BM access
  3. In the left sidebar, click Events Manager

If you have multiple BMs visible, select the one containing the LG Military Ads account (1034037341617540). The associated business ID for the lead-gen dataset is 551525685236215 — that may be a separate BM you need to switch to.

  1. In Events Manager, the left sidebar shows a list of Datasets / Pixels
  2. Click on Military Insurance (ID 332555186604235) - Not Military.com's Pixel (that's the content pixel) - Not MA Military Insurance (that's the dormant MediaAlpha one)

You should see the dataset overview page. Top of the page shows: event activity graph, today's volume, etc.

Open the Diagnostics tab

Top navigation of the dataset page shows tabs: Overview | Test Events | Diagnostics | Settings | History

Click Diagnostics.

This tab surfaces parameter-validation errors that the Marketing API doesn't expose.

What to look for (in priority order)

Priority 1: Purchase event value parameter

In the Diagnostics tab, look for an entry that says one of: - "Purchase event has formatting issues with the value parameter" - "Purchase event value parameter is missing or invalid" - "Recommend updating Purchase event value field to match expected format" - Or similar — Meta phrases vary

If you see one of these, click it to expand: - It will show how many events are affected (Mo's audit cited 98% — verify this) - It will show what the value parameter currently looks like (probably $0, $1, or a non-numeric string) - It will show what the recommended format is (numeric, typically decimal like 27.50)

Screenshot this for the master record. This is the single most important screenshot from Tuesday.

Priority 2: fbc / fbclid parameter

In the same Diagnostics view, look for an entry referencing fbc (Facebook Click ID) with words like: - "Truncated" - "Invalid format" - "Missing characters"

If you see this, click to expand: - Note how many ad sets / events are affected (Mo's audit cited 15 ad sets / $67,174 / 27% of CompleteRegistration) - Note the example fbc value Meta shows (it'll be a string starting with fb.1. — should have 3+ dots; truncated = fewer)

Priority 3: Anything else flagged

Scroll the full Diagnostics view. Things to also note: - Any CompleteRegistration parameter warnings - Any dedup warnings (browser + server fire same event with mismatched dedup keys) - Any PII format warnings (hashed values not in lowercase, etc.) - Any upload frequency warnings (events arriving too late)

Most of these will be clean given the EMQ scores are strong, but anything flagged is worth a screenshot for the dev conversation.

What "good" looks like (so you know if Mo's prior claim was wrong)

If the Purchase value parameter is NOT flagged in Diagnostics, then: - Mo's "98% of Purchase price data is malformed" finding has been resolved since the prior audit, OR - The diagnostic surfaces it differently in 2026 than it did when Mo first found it, OR - The issue is real but Meta's Diagnostics tab doesn't catch it (rare but possible)

In that case, the next check is: 1. Click into the value parameter detail under the Purchase event (in the Overview tab → Event Details) 2. Look at the distribution of values Meta has received 3. If 98%+ of values are the same number (e.g., everything is $1, $27.50, or $0), that's the flat-value problem even if Meta doesn't flag it 4. Real lead-payout data would show a distribution: $15 here, $32 there, $47 there, etc.

A flat distribution = broken value pipe regardless of whether the Diagnostics tab flags it.

What to do with what you find

If value malformation is confirmed:

Walk into the meeting with the screenshot. Say:

"Confirmed this morning — Purchase events fire fine, but the value parameter is malformed in 98% of cases. Meta's flagging it in Diagnostics. This is why ROAS isn't available, this is why Max Val is paying a 67% CPL premium for nothing. Here's the screenshot. Dev — what's the fix and how long?"

If fbc truncation is confirmed:

Same approach, second priority. Say:

"Second thing — fbc parameter is truncated on N ad sets representing M% of CompleteRegistration events. That's the click-ID Meta needs to attribute leads back to specific ads. Less urgent than value but real. Dev — likely a URL parameter handling bug. Worth scoping with the value fix?"

If both are confirmed and value malformation is severe:

The whole pitch is: the Auto Max Val decision becomes urgent, not speculative. Lean toward Option A (pause) unless dev commits to a 2-week value-fix.

If neither is flagged (Mo's prior audit overstated):

Don't pretend. Update the master_audit.md's Section 4 to mark the finding as "claimed but not reproducible in current Diagnostics — dev should still investigate root cause of ROAS unavailability."

Adjacent thing to check while you're in there

In the Datasets list, click on Military.com's Pixel (ID 587241028099131) and check:

This confirms the +23% cost/result lever (adding CAPI Gateway to the content pixel). Screenshot the "no CAPI" state — it's the artifact you walk into the dev conversation with for that ask.

Total time

Realistic total: 12-15 minutes. Do this before the 9 AM meeting and you walk in with screenshots that anchor the whole conversation.

What to NOT click on

The Diagnostics tab has a "Resolve" / "Apply Recommendation" button for some issues. Do NOT click anything that says "Fix" or "Apply" without dev review. Meta's auto-fixes for parameter issues sometimes restructure the event schema in ways that break downstream integrations (especially with QW/Delty postbacks). Read-only Tuesday. Decisions get made in the meeting, not in the dashboard.

← Back