Our Multi Touch Attribution eCommerce Test Across 3 Stores
We audited three scaling apparel stores spending a combined $420,000 a month on paid media. The multi-touch attribution software they used caused them to starve their cold prospecting pipelines by exactly 34 percent. The founders were looking at pretty dashboard models showing incredible returns on bottom-funnel channels. Meanwhile, their bank account cash flow was shrinking every single week.
Standard attribution dashboards create a false sense of certainty for media buyers. They assign credit to the last click or the easiest conversion. They ignore the heavy lifting done by top-of-funnel discovery campaigns on Meta and TikTok. The result is a misallocated budget that slowly chokes your customer acquisition.
I have seen this pattern across dozens of eCom accounts. Founders pay thousands for third-party tracking tools. They expect clarity and actionable insights. Instead, they get a distorted view of how their marketing drives profit. The software tells them they are making money, but they struggle to pay their suppliers.
The multi touch attribution ecommerce experiment setup
We set up a strict audit across three Australian apparel brands. These stores were scaling between $80,000 and $200,000 a month in paid media. They all had the exact same problem. Revenue growth had completely stalled despite increasing their ad spend quarter over quarter. We needed to find out why this was happening.
The tracking stack we evaluated was standard for brands at this level. We looked at platform in-app reporting from Meta Ads and Google Ads. We analysed data from third-party multi-touch attribution software like Triple Whale and Northbeam. Finally, we looked at the ultimate source of truth. We looked at the merchant bank balances and Shopify raw sales data.
The founders trusted their third-party multi-touch attribution tools completely. These platforms promised to map the entire customer journey from the first impression to the final checkout. They promised to assign exact dollar values to every touchpoint. This created a massive blind spot for the internal teams. The standard attribution dashboards created a false sense of certainty for the media buyers managing the accounts.
Media buyers love dashboards that show high returns. It makes their daily reporting look highly successful. When a multi-touch model claims a specific retargeting campaign is generating a ten-to-one return, buyers scale it. They stop looking at the broader business metrics. They ignore the blended marketing efficiency ratio entirely. They just chase the highest number on the screen.
We review these exact setups in our case studies regularly. The software itself is not inherently broken. The problem lies in how the data is interpreted and acted upon by human operators. The algorithms favour channels that capture existing intent. They heavily discount channels that generate new intent.
This experiment was designed to strip away the software guesswork. We wanted to see what happened when we stopped feeding the algorithm and started looking at raw cash flow. We needed to measure true incrementality. We set a baseline using 90 days of historical data from the multi-touch platforms. Then, we compared that data to the cash sitting in the bank. The discrepancy was alarming. The software claimed massive profitability. The bank statements showed stagnant growth. We had to break down exactly where the multi-touch models were failing these three stores.
Multi touch model flaws in brand versus prospecting splits
The multi-touch attribution algorithms were actively harming the growth trajectory of these three brands. The software credited Google Brand Search and Klaviyo retargeting flows with up to 34 percent of prospecting-driven revenue. It looked like bottom-funnel channels were performing absolute miracles. In reality, they were just taking credit for the heavy lifting done by Meta Ads. If you want to verify whether your retention channels are taking unearned credit, our free Klaviyo audit checks how your email setup attributes true revenue alongside paid traffic.
There is a massive downstream risk when you throttle top-of-funnel Meta Ads based on fractional attribution reporting. Media buyers see low returns on cold prospecting campaigns in their dashboards. They react by cutting the budget immediately. They shift that spend into retargeting and brand search to keep their daily numbers looking good. This creates a death spiral for customer acquisition. You stop filling the funnel with new people.
We compared the reported platform return on ad spend against the multi-touch calculated customer acquisition cost. Then we looked at the blended profit on ad spend. The platform metrics looked fantastic. The multi-touch software validated those platform metrics. The blended profit on ad spend told the real story. It was dropping week over week across all three stores.
The brand search credit trap
Algorithmic multi-touch models have a built-in bias. They lean heavily toward capturing intent over generating intent. A user sees a Meta ad on Tuesday. They ignore it. On Thursday, they remember the brand name and search for it on Google. They click a brand search ad and buy. The multi-touch software gives a massive chunk of that revenue to Google.
We paused brand search keywords across all three accounts for 14 days to measure true incrementality. We wanted to see how many of those sales would happen anyway through organic search. The results were incredibly clear. Total store revenue dropped by less than two percent. The multi-touch software had claimed Google Brand Search was driving 18 percent of total revenue. It was a complete illusion. The intent was already there. Google was just collecting the toll.
Meta prospecting cannibalization
Algorithmic post-click attribution severely depresses cold social prospecting spend. The software struggles to track users across multiple devices and long time delays. Meta prospecting campaigns often look like losers in third-party dashboards. The media buyers in our experiment had cut Meta prospecting budgets by an average of 40 percent over six months.
We found a direct correlation between these Meta prospecting cuts and returning customer drops 60 days later. You cannot have returning customers if you do not acquire new ones first. The multi-touch software encouraged the media buyers to cannibalise their own prospecting efforts. They were starving the exact campaigns responsible for long-term business growth. If you want effective Google Ads management, you must understand this dynamic. You have to separate intent generation from intent capture. You cannot let the dashboard dictate your top-of-funnel strategy.
Operational steps to remove bloated MTA ecommerce tracking
Fixing this issue required a complete overhaul of the tracking architecture. We started with a full audit and removal of redundant client-side tracking scripts. The three stores were running multiple overlapping tracking tags. They had native Meta pixels, Google tags, Triple Whale scripts, TikTok pixels, and custom event codes firing simultaneously on every single page load.
This bloated setup degrades site speed significantly. A slow site kills conversion rates before attribution even matters. We saw load times drop by over a full second just by cleaning up the tag manager container. More importantly, overlapping scripts confuse the Conversions API signals sent back to the ad platforms. When Meta receives conflicting data from a browser pixel and a third-party tracking script, the machine learning models struggle to optimise delivery.
We stripped the tracking stack down to the absolute minimum. We established a single source of truth using Shopify raw sales data. We paired this with server-side first-party event tracking via Google Tag Manager. This approach is much cleaner. It relies on reliable server data rather than fragile browser cookies that get blocked by iOS updates. You can read the technical specifications for setting up server events in the Meta Business Help Centre.
Server-side tracking gives the ad platforms the accurate signals they need to find buyers. It removes the conflicting data points caused by third-party attribution software. We then rebuilt the account structures from the ground up.
The goal was to evaluate net-new customer acquisition without software guesswork. We separated cold prospecting campaigns from retargeting campaigns with strict exclusion audiences. We excluded past purchasers, Klaviyo subscribers, and 180-day website visitors from our top-of-funnel Meta campaigns. This forced the ad platforms to go after completely new people.
We stopped looking at fractional attribution models entirely. Instead, we focused on marketing efficiency ratio and true customer acquisition cost based on total store revenue. This operational shift changed everything. It removed the false certainty of the dashboard. It forced the media buyers to look at the business metrics that matter to the founder.
Clean tracking leads to better Meta Ads management. The algorithms need accurate data to perform. When you remove the bloated tracking scripts, you give the machine learning models a clear target. You stop confusing the system. You stop double-counting conversions. You get a much clearer picture of how your marketing dollars turn into profit.
Contribution margin gains 90 days post-MTA transition
The operational changes produced immediate financial outcomes. We tracked the data for 90 days after ditching the multi-touch attribution software. We shifted entirely to contribution-based budgeting. The numbers proved that the previous attribution models were actively suppressing profit.
We saw a net margin improvement averaging 4.8 percentage points across the three stores. This was not a slight bump in return on ad spend on a dashboard. This was real, bankable profit added to the bottom line. When you are spending up to $200,000 a month on ads, a 4.8 percent increase in net margin transforms the business. For one of the stores, this equated to an extra $32,000 in pure profit every single month.
This growth came directly from reallocating the 34 percent misallocated prospecting budget. We took the money that was previously wasted on brand search and heavy retargeting. We pushed it back into high-intent cold audiences on Meta. We let the algorithms find new customers based on broad targeting and strong creative.
At first, the in-platform metrics looked worse. Meta reported a lower return on ad spend than the third-party software used to show. But we were no longer managing the accounts based on those specific platform metrics. We were managing them based on total store profitability.
We compared the raw cash flow and bank statement growth to the previous multi-touch projections. The multi-touch software had projected steady growth based on scaling bottom-funnel channels. That growth never materialised in the bank account. Once we shifted budget back to cold prospecting, the actual cash flow surged.
The daily sales volume increased significantly. The cost to acquire a new customer dropped when measured against total store spend. We stopped paying Google for sales we would have captured organically. We stopped paying Meta to show ads to people who were already going to buy from a Klaviyo email campaign.
This is the exact strategy we outline in our breakdown of How One Store Doubled Profit Using MER, Not ROAS. You have to look at the blended metrics. Multi-touch attribution tries to slice the pie into tiny pieces. Contribution margin looks at the size of the whole pie.
When you focus on contribution margin, you make better media buying decisions. You stop chasing cheap, low-value conversions. You start investing in campaigns that grow the customer base. The 90-day data across these three stores proved that simple, macro-level metrics beat complex attribution models every single time.
Next steps for multi touch attribution ecommerce audits
Attribution is not a software subscription issue. It is a media architecture and incrementality discipline. You cannot buy a tool to solve a fundamental flaw in your marketing strategy. The team at Elite Brands sees this misunderstanding constantly. Brands spend thousands on software, hoping it will tell them exactly what to do.
There are three questions founders must ask before renewing five-figure attribution software contracts. First, does this tool change our media buying decisions, or does it just validate what we already want to believe? Second, are we measuring true incrementality, or just taking credit for existing intent? Third, does the dashboard profit match the cash in our bank account?
If the answer to that last question is no, you have a major problem. You need to conduct an incrementality holdout test on your Google Brand and Meta campaigns immediately. You have to verify the numbers your software is feeding you.
Holdout tests are simple to run. You pause a specific campaign for 14 to 30 days. You monitor the impact on total store revenue and blended customer acquisition cost. If you pause your Google Brand Search campaigns and total revenue stays flat, those ads were not incremental. You were just cannibalising your organic traffic. If you pause a Meta retargeting campaign and total sales do not drop, that campaign was useless.
These tests reveal the truth behind the dashboard numbers. They expose the bias in algorithmic attribution models. You do not need expensive software to run a holdout test. You just need discipline and a willingness to accept that your dashboard might be lying to you.
We integrate these incrementality tests into our process for every new client. We do not trust third-party tracking tools blindly. We rely on raw Shopify data, server-side tracking, and strict holdout tests. This is the only way to build a media architecture that scales. We test everything against the cash in the bank.
If your brand is spending over $50,000 a month on paid media, you cannot afford to rely on flawed attribution models. The misallocation of budget is costing you thousands of dollars in lost profit every week. You need a clear, unbiased look at your media efficiency. You need to strip away the bloated tracking scripts and focus on contribution margin.
Not sure if your Klaviyo setup is leaving money on the table?
We’re Klaviyo Master Gold partners. Our free Klaviyo Audit checks the 24 things that most often kill email revenue on Shopify stores. Takes 5 minutes to request, delivered within 48 hours.
Request the free Klaviyo Audit →
If you want a hand with this, we can run a full diagnostic on your tracking and media architecture.
