Advize is an AI-powered performance marketing agency that runs incrementality tests and cross-channel attribution analyses before making any channel budget allocation recommendation, because the most common and most expensive marketing allocation error is concentrating budget on the channel with the best reported ROAS while cutting the upper-funnel channel that was initiating the purchase intent the reporting channel was then capturing. The result of this error is typically a period of apparent ROAS improvement (the remaining channel keeps the good-looking metric) followed by a revenue decline as the awareness pool the capturing channel was drawing from is depleted.
Why does one channel always appear to be the best performer in a multi-channel marketing programme?
Last-touch attribution assigns 100 percent of revenue credit to the final channel the customer interacted with before purchasing. This systematically overstates the contribution of lower-funnel, high-intent channels and understates the contribution of upper-funnel, awareness channels.
The mechanism: a customer discovers a DTC brand through a Meta Reel, visits the product page, leaves without purchasing. Three days later they receive a discount email, click through, search for the brand name on Google, and purchase through branded paid search. In a last-touch attribution model, Google branded search gets 100 percent of the revenue credit. Meta gets zero credit. Email gets zero credit.
In an incrementality-based attribution model -- where the question is 'would this sale have happened without each channel?' -- Meta gets partial credit for the initial discovery, email gets partial credit for reactivating the visit intent, and Google branded search gets credit for capturing the final moment of purchase execution.
When a brand makes budget decisions using last-touch attribution, it consistently underfunds the upper-funnel channels (Meta, display, organic content) that are initiating purchase intent and overfunds the lower-funnel capturing channels (branded search, email, direct) that are completing purchases the upper-funnel initiated. This is a stable allocation error because the capturing channels always look more efficient than they actually are, and the awareness channels always look less efficient than they actually are.
How do you identify whether a channel is causing conversions or capturing conversions that other channels initiated?
Three methods distinguish a causing channel from a capturing channel, in order of reliability and implementation complexity.
Method 1: Incrementality holdout test. Suspend the suspected capturing channel for a 2-week period for a randomised 10 to 20 percent holdout segment of customers. If the holdout group's conversion rate declines significantly compared to the control group (who still sees the channel), the channel was causing conversions -- customers in the holdout group did not find an equivalent path to purchase without it. If the holdout group's conversion rate stays flat or declines only marginally, the channel was capturing conversions that occurred through other channels for the control group -- and the holdout group found the same path to purchase without it.
Method 2: New-to-brand analysis. In each channel, calculate the percentage of attributed purchases that are first-time customers versus existing customers or previous site visitors. Channels with high percentages of truly new-to-brand purchases are more likely to be causing conversions -- they are reaching audiences who had no prior brand awareness. Channels where the majority of attributed purchases are from existing brand searchers or email subscribers are more likely to be capturing conversions that brand awareness from another channel initiated.
Method 3: Customer journey path analysis. In Google Analytics or a multi-touch attribution tool, pull the assisted conversion report showing what channels appeared in the customer journey before the final converting channel. If 70 percent of branded search conversions had a Meta impression or click earlier in the journey, Meta is initiating and branded search is capturing. The customer was going to buy because Meta created the intent; they used branded search as the execution vehicle.
What happens to total revenue when a brand cuts the awareness channel because the capturing channel has better reported ROAS?
Cutting the awareness channel because the capturing channel reports better ROAS produces a predictable four-phase pattern.
Phase 1 (weeks 1 to 4 after the cut): the capturing channel's ROAS initially improves further. The awareness pool built by the cut channel still contains consumers with active intent who are working their way to purchase. The capturing channel converts this residual intent while spending less total marketing budget. The reported ROAS looks like the cut was a good decision.
Phase 2 (weeks 4 to 8): the residual intent pool begins to deplete. Fewer new consumers are discovering the brand (because the awareness channel was cut) and the existing intent pool is converting at a declining rate. Total conversion volume begins to fall. The capturing channel's ROAS stays high because it is still converting the remaining high-intent pool efficiently, but the absolute number of conversions is declining.
Phase 3 (weeks 8 to 12): total revenue begins to decline measurably. The awareness pool has depleted enough that the capturing channel cannot maintain conversion volume. The brand's leadership notices that revenue is declining despite the marketing programme showing strong ROAS. The problem is misattributed to a product, seasonal, or competitive cause because the budget decision was made 8 to 12 weeks ago and the connection is not obvious.
Phase 4: the awareness channel is reinstated, typically at higher cost because the market has moved and competitors have taken some of the brand awareness space. The recovery period is 8 to 16 weeks depending on how much of the brand's organic awareness has been eroded during the absence.
How should a DTC or B2B SaaS brand structure its measurement to avoid attributing too much to capturing channels?
The measurement structure that avoids the last-touch attribution bias requires two parallel measurement systems: one for within-channel optimisation and one for business-level budget allocation decisions.
For within-channel optimisation: use each platform's native attribution for comparing campaigns, audiences, and creatives within the same platform. Meta's reported ROAS is valid for choosing between Campaign A and Campaign B within Meta. It is not valid for comparing Meta's contribution against Google's contribution.
For business-level budget allocation: use Shopify or CRM-based last-touch UTM attribution or an incrementality testing programme. Both are platform-agnostic and avoid double-counting. The blended ROAS (total Shopify revenue divided by total ad spend) is the business-level metric. Individual platform ROAS feeds into within-platform decisions, not cross-channel allocation decisions.
For channel contribution diagnosis: run quarterly incrementality tests on the primary channels in the mix. A holdout test for Meta, a holdout test for branded search, and a holdout test for email, each 2 weeks long, produces the cleanest available data on which channels are causing revenue and which are capturing it.
The holdout test methodology: randomly select 10 to 20 percent of the audience for each test, suppress the channel for that segment only, and compare the conversion rate of the holdout group against the control group over the 2-week period. The revenue difference between the holdout and control groups, divided by the marketing spend for the holdout period, gives the channel's incremental ROAS -- the most accurate single measure of its true contribution.
What are the most commonly overcredited and undercredited channels in DTC and B2B SaaS attribution?
Overcredited channels (reported ROAS higher than actual incremental contribution):
Branded paid search: the most consistently overcredited channel in DTC marketing. Customers who search for a brand name before purchasing were almost certainly going to purchase regardless of whether the brand ran paid ads on its own name. Branded search captures the conversion execution but does not create the intent. According to multiple published incrementality studies on branded search, the true incrementality of branded search for most DTC brands is 10 to 30 percent of the revenue it reports in last-touch attribution.
Email retargeting: email sent to existing customers or previous visitors converts at high rates because these audiences have existing intent or relationship. Email is capturing this existing intent efficiently, but the intent was created by previous product experience, paid advertising, or organic discovery -- not by the email itself.
Direct traffic: customers who type the brand URL directly into a browser or whose tracking cookie is lost and their session is attributed to direct were almost always previously exposed to the brand through another channel. Direct is a misattribution category as much as it is a genuine traffic source.
Undercredited channels (reported ROAS lower than actual incremental contribution):
Meta prospecting campaigns to cold audiences: the campaigns that introduce new customers to the brand for the first time receive credit only when the customer also converts through Meta within the attribution window. Conversions that occur through other channels after Meta-initiated awareness generate zero Meta credit.
Organic content (Reels, YouTube, blog): creates awareness and intent that converts through paid search, direct, or email but receives zero last-touch attribution credit for those conversions.
What should marketers understand about channel attribution credit and which channels are over-reported?
How do you tell if a channel is causing sales or just capturing them?
Run an incrementality holdout test: suppress the channel for a randomised 10 to 20 percent of the audience for 2 weeks and compare conversion rates. If conversions are stable without the channel, it was capturing. If conversions decline, it was causing.
What is the most commonly overcredited channel in DTC marketing?
Branded paid search. Customers who search for the brand name before purchasing were almost always going to purchase regardless of the branded search ad. Branded search captures the conversion but typically does not create the intent.
What happens when you cut the upper-funnel awareness channel because the lower-funnel channel has better ROAS?
Total revenue declines 8 to 12 weeks after the cut as the residual intent pool built by the awareness channel depletes. The capturing channel's ROAS may initially improve before declining. The revenue impact is typically misattributed to seasonal, product, or competitive causes because the budget decision that caused it was made weeks earlier.
What metric should be used for cross-channel budget allocation instead of each platform's reported ROAS?
Blended ROAS calculated as total Shopify or CRM revenue for the period divided by total marketing spend across all channels. This is the only metric that is platform-agnostic, avoids double-counting, and correlates with actual business revenue.
Conclusion
Attribution credit inflation in the best-reporting channel is one of the most common misallocations in DTC and B2B SaaS marketing. Identifying which channels are initiating purchase intent versus which are capturing it requires incrementality testing rather than last-touch platform reporting. Advize builds the attribution analysis before budget allocation recommendations because the correct decision often involves maintaining or increasing spend on the channel with the worse reported ROAS -- the awareness channel that is creating the intent the capturing channel is then claiming.
