Taking over a Google Ads account with broken tracking

harper-local

Recently took over a Google Ads account for a company, and I'm smack in the middle of a tracking nightmare. The data is all over the place, and I'm honestly not sure if we've been bleeding money for years or if the tracking has just been garbage the whole time.

The setup:
New to performance marketing, but I know enough to spot when something's off. Account's been running for years - Search and Performance Max (now paused). We're in peak season, so scaling should be the name of the game.

The mess:
Google Ads rates campaigns as 'very Good' across the board. Traffic's coming in. But conversions? Either dead low or wildly inconsistent. The numbers in Ads Manager don't line up with what we see in Analytics - and when they briefly matched last week (after I raised hell), it looked realistic. Now both look awful again.

ROAS has been sitting at 0.27 for years. That's either tracking broken from day one, or the campaigns have been hemorrhaging cash forever. Neither is good.

What I've done:
Flagged it to the developer. They tinkered with event tracking (purchase events etc.). For a few days, data looked sane. Now it's busted again.

Situation now:
Paused the PMax because we can't trust anything. But we're supposed to be ramping up for peak season. Feels like scaling blind.

Questions running through my head:
What's the most bulletproof way to validate end-to-end whether conversion tracking is actually working? Common reasons for tracking working temporarily then breaking again? And should I keep campaigns paused until tracking is spotless, or cautiously restart with a low budget?

I've had a few folks suggest bringing in a specialist for a day or two - makes sense, because debugging tracking blind as a beginner could take weeks. Others pointed out that the 'fixed then broke' pattern usually means the developer patched a symptom, not the root cause, or there's a conflict between multiple tracking implementations (Google Ads tag + GA4 both firing purchase events = double counting or weird gaps).

For validation, I'm starting with Tag Assistant and GTM Preview mode to watch purchase events fire on a real test transaction. Then I'll check if the conversion action in Google Ads pulls from GA4 or a direct tag - mixing those causes the exact inconsistency I'm seeing. Also looking at the 'all conversions' vs 'conversions' column in Ads Manager, a big gap means something's being filtered out.

Honestly, I wouldn't keep Search paused during peak season over a tracking headache. Run it on limited budget, use manual backend data (order counts, revenue) as a sanity check, and fix tracking in parallel. Scaling blind is bad, but going dark during peak season is worse.

PMax being paused? Could be a blessing depending on your industry.

This is a common mess when inheriting accounts - tracking is often half-broken or inconsistent. Don't try to optimise anything yet. Validate tracking from scratch: check what conversion actions exist, see their source (GA4 import vs direct tags), run real test conversions and confirm they show up, make sure events fire only once on the correct step (thank-you page). Many accounts have duplicate tags, outdated GA4 imports, or conversions firing in the wrong place.

Until that's clean, campaigns optimise on bad signals. Then you can touch the campaigns. Otherwise, you're guessing.

logan-lifecycle

Pausing PMax was the right call for now. If conversion tracking is unreliable, Google's optimisation becomes a guessing game-especially with automated campaigns. A ROAS of 0.27 over several years plus inconsistent reporting usually points to broken attribution, duplicate or missing purchase events, consent-layer issues, or mismatched values being sent back to Ads.

First thing I'd do is validate the entire conversion path manually. Fire up Tag Assistant, GTM preview mode, GA4 DebugView, and run test purchases end-to-end, checking whether the same conversion fires correctly across GTM, GA4, and Google Ads. Look for double firing, delayed firing, cross-domain problems, cookie consent interference, or devs changing checkout flows without updating the tracking setup.

Seen this sort of thing before-campaigns looked terrible simply because the platform wasn't receiving clean signals. The real shift came when I stopped obsessing over surface-level metrics and focused on whether tracking actually reflected genuine customer behaviour. Once the signal quality improves, optimisation decisions become much clearer.

ethan-pr

That sounds like a nightmare to walk into, but I've been there too. First thing I'd do is fire up Tag Assistant in Chrome and manually step through the conversion journey-place an order yourself or trigger the lead form to see exactly what fires. If you're seeing clicks come through but no conversion data, it often points to a GTM container issue or a tag that's simply not present on the post-click page. Take it funnel step by step, the root cause usually reveals itself quickly. No need to panic-just methodically check what triggers.

24fire

Tracking is the foundation, and too many accounts are built on sand these days. My advice: kill PMAX until you've proven the funnel end-to-end. Keep a lean Search budget if the season matters, but don't let platform scores lull you into false confidence. Verify a single conversion - order ID, value, thank-you page, consent mode - through Tag Assistant, DebugView, and the conversion action itself. One thorough test beats any automated diagnostic.

If the numbers worked for a week then fell apart, odds are something changed on your end: a dev push, a checkout tweak, a consent banner update, or a plugin conflict. Google's own docs point to Tag Assistant for this exact forensic work. Tracking breaks are rarely mysterious - just easy to ignore until the data is worthless.

trendspotter

That "free" label on GA4 and Looker Studio looks absolutely stunning on paper - pristine, like a white marble countertop. But pull back the gloss and there's a full workshop of hidden labour underneath. Event design, custom dimensions, data layer, then those DAX-adjacent calculated fields, and if you dare connect BigQuery you're suddenly writing SQL. I keep coming out at roughly twelve hours of learning curve before the first revenue dashboard even glimmers into view. not a complaint - just a line item that deserves a spot in any build estimate, honestly.

emmareach

They're are endless variables here. Couple of additional things to be aware of that I didn't see mentioned here.

Sounds like you have both Google Ads conversions and GA4 events tracking purchases. Generally speaking Google Ads should always be higher due to attribution rules, the fact that the Google Ads tag is siloed (it doesn't know or care if you leave and come back through a different channel). However you should compare your conversion windows and make sure you're using the same windows in each platform.

The other thing worth noting is where conversions get attributed in each platform. Google Ads is date of click. Meaning if the user clicks on an ad on Monday and buys on Wednesday the purchase is attributed to Monday. GA4 is date of conversion, so that same purchase gets attributed to Wednesday.

I didn't see any mention of CMPs. If you're website recently added a cookie partner and has overly strict blocking rules you would absolutely be seeing little to no volume. Especially if you didn't modify your Google tags for consent mode. (Easy to do if you are using GTM.)

Worth checking in GA4 before you call in the calvary is device category, operating system and browser level performance. Look for significant differences to see if the issue is isolated or global.

Last thing to compare is your GA4 data against your internal sales system. This will tell you how much data you're actually missing. 3-6% data loss is normal nowadays.

zoecreative

ROAS of 0.27 for years is absolute madness. That's either a broken tracking setup or a slow-motion cash bonfire. The "works then breaks" pattern almost always points to duplicate events or conflicting tags fighting each other.

Quick way to sanity-check it: fire up GTM preview mode, run a test purchase, and see if the event fires exactly once. Cross-reference GA4 with your native Ads tags - they often double-count if you haven't deduplicated. On mobile apps, something like AppsFlyer beats patching broken web tracking hands down. Don't touch a single budget until the data behaves.

communityfirst

worked temporarily then broke again' usually means tracking was broken again during a code update.. even if roas is low but campaign ratings are good, the problem may be in tracking, not performance. you can first use tag assistant and test the full purchase flow to see if the conversion fires once, multiple times, or not at all. then compare analytics data with backend orders. if analytics matches the backend, it is an ads import issue, if not, the tracking is broken. it is better not to scale without one reliable source of truth, because running ads on incorrect data increases risk..

bouncekiller

Oh, honey. Before you even think about scaling, you need to burn that account down and rebuild the tracking from scratch. Here's what I'd do (and have done, more times than I care to admit):

do a real test purchase yourself, end to end. Don't just watch - get your phone out, record the screen if you have to.
Fire up GTM Preview / Tag Assistant and watch the damn thing like a hawk.
Confirm the purchase event fires exactly once, with the right value, currency, and a unique transaction/order ID.
then hop into GA4 DebugView to make sure it lands there without getting lost.
Check whether Google Ads is using a native conversion tag, imported GA4 conversions, or - god forbid - both at the same time.
compare backend orders, GA4, and Google Ads for the same short time window. if your backend shows sales but GA4 doesn't, tracking is toast. If GA4 sees it but Ads doesn't, it's usually an import/mapping/attribution mess. And if that event fires more than once? smart Bidding has been learning from garbage data for weeks, and your account is basically running on hope.

Personally, I wouldn't touch PMax until you've got one clean, proven conversion path. Scale on broken data and you'll just be burning cash faster. trust me, I've been that idiot

maya-b2b

That kind of disconnect between Google Ads and GA4 usually points to something breaking in the tracking flow. I'd start by running a few test purchases manually across different devices and browsers, then compare what each platform actually records.

In my experience, it often comes down to attribution windows, cookie issues, or cross-domain tracking headaches. Rather than bouncing between reports, try pulling raw conversion data side by side. I've found Windsor AI handy for that - it merges Google Ads and GA4 into one view so you can spot missing transaction IDs, timestamp mismatches, or events only firing in one platform. Saves a lot of guesswork.

jamesaso

i’d bring in a specialist for 1-2 days if budget allows. debugging tracking blindly as a beginner can take weeks. someone experienced will spot patterns fast

bouncekiller

A ROAS of 0.27 for years? That's not a red flag, that's a five-alarm fire. Either your tracking's been fucked since day one, or someone's been happily torching cash. Either way, that's not your mess to clean up - but you're the one holding the mop.

The "fixed briefly then broke again" pattern screams developer slapped a plaster on a bullet wound. Classic symptom of a conflict - Google Ads tag and GA4 both firing purchase events, double counting, then randomly skipping. Same shit, different account.

Validation steps you can run:

Fire up Google Tag Assistant or GTM Preview. Watch the purchase event fire live during a test transaction. If it's shy, you've found your culprit.
Check whether your conversion action in Google Ads is pulling from GA4 or a direct tag. Mixing sources causes exactly this inconsistency - it's like having two cooks in the kitchen, one adding salt, the other sugar.
Compare "all conversions" vs "conversions" columns in Google Ads. Big gap means something's being excluded or filtered - probably the wrong thing.

Don't pause Search over this during peak season. Trim the budget, use manual monitoring - order counts from your backend, revenue numbers - as your sanity check. Fix tracking in parallel. Going dark when demand's high is just as dumb as scaling blind.

PMAX being paused? Honestly, that's probably a blessing. You've got enough on your plate without that nonsense.

trendspotter

That’s a pretty common situation when taking over an account-tracking is often half-broken or inconsistent.
I wouldn’t try to optimize anything yet. First step is to validate tracking from scratch
- Check what conversion actions exist in Google Ads
- See if they’re coming from GA4 import or direct tags
- Run real test conversions and confirm they show up
- Make sure events fire only once (no duplicates) and on the correct step (thank-you page)
A lot of accounts end up with duplicate tags, outdated GA4 imports, or conversions firing in the wrong place, which makes the data unreliable.
Until that’s clean, the campaigns are basically optimizing on bad signals.
Once tracking is verified, then it’s worth touching campaigns-otherwise you’re just guessing.

emailwiz

I remember taking over an account a few years back where the conversion rate was stuck at 0.3% for months. turned out the tracking had been broken since a site update nobody flagged. That feeling of staring at wasted spend? Absolutely gutting.

You're right to question the tracking first. A test purchase or Tag Assistant is the quickest way to confirm if it's firing properly. if you're on WooCommerce or Shopify, sorting it out is usually straightforward-grab a freelancer for an one-off fix and you're golden. Custom ecom platforms? That's where it gets painful. Platform updates or plugin changes often break things silently, and unless someone's actively monitoring, it slips through.

A 0.27% conversion rate for years screams that nobody was checking the data quality. That's a tough pill to swallow, but you can't scale or optimise if Google Ads is flying blind. what's your current setup-are you using Google Ads tag, GA4, or something else? Sharing that would help narrow down where the leak is. honestly, fixing the tracking is priority one before you touch bids or audiences.

davidsearch

Tracking inconsistencies are a nightmare when inheriting an account. Had the exact same problem last year: data looked fine for a few days, then went completely wonky. Turned out it was a duplicate purchase event firing - once via GTM, once hardcoded on the thank-you page. Tag Assistant showed it clean, but it double-counted inconsistently depending on browser cache state.

What cracked it for me: cross-referencing GA4 → Google Ads import separately from native Ads conversion tags, then checking if spikes or drops lined up with any site deploys. Also brought in Prooflytics for end-to-end attribution - not a magic fix, but having everything in one place made the audit far less painful.

On your pause vs. run question: keep Search live with a capped budget. Use it as a signal check while debugging. Pausing everything leaves you blind and makes it harder to confirm if any fix actually worked.

miaaffiliate

That "fixed then breaks again" pattern is a margin killer - every time you think you've plugged the leak, it springs another. Usually it's either the purchase event firing before the dataLayer's fully loaded (race condition) or you've got duplicate tracking that just happens to align with GA4 by accident. Either way, you're burning budget on bad data.

Grab Tag Assistant Companion in Chrome, place a dummy order, and watch the sequence. See if the Google Ads conversion tag fires twice per transaction or if it triggers before the order confirmation page has finished rendering. That's where the waste starts.

The 0.27 ROAS lasting years is the real nightmare. Smart Bidding has been optimising on broken signals for that long - the model's essentially trained on noise. Even once you fix the tag, you're looking at a reset period. Switch to Maximise Clicks or a very loose tCPA to let it retrain on clean data. Don't jump back into PMax yet. Run a controlled Search campaign with Enhanced CPC or Manual CPC while you stabilise tracking. After you've banked 30+ clean conversions over a couple of weeks, you can rebuild PMax with a proper baseline.

There's a solid breakdown here that covers exactly this diagnosis: https://adpredictor.ai/en/blog/conversion-tracking

emmareach

fixing the signal path first is the only move that actually matters. everything else is just guessing with more steps.

i've been using a tool that auto-alerts for tracking drift on a few accounts, and it's saved me from this exact rabbit hole a couple of times now

harper-local

Same situation here with the developer. Going to try this - call tracking is the only way to sort out attribution properly.