Skip to content
English
Contact us
  • There are no suggestions because the search field is empty.

GMV Max Halo Model Explainer

Fospha’s GMV Max Halo Model explains how we isolate the true incremental impact of TikTok GMV Max by separating paid vs organic revenue and capturing cross-platform “halo” sales on DTC and Amazon to calculate Total Commerce ROAS.

Overview

This article explains the four-step process Fospha uses to measure GMV Max performance across TikTok Shop, DTC, and Amazon.

It covers how data is ingested, how the model separates paid from organic activity, how conversions are attributed to GMV Max campaigns, and how cross-platform halo revenue is captured and rolled into your Total ROAS figure.


How the GMV Max Halo Model works

Step 1 — Data Ingestion: Two clean streams, zero overcounting

Fospha draws on two separate data sources:

  1. TikTok Shop order data — ingested via Shopify integration, TikTok Shop API, or manual upload. This provides a ground-truth record of every order placed through TikTok Shop, including order value, product, and timestamp.
  2. TikTok MMM API — provides impression-level ad performance data for GMV Max campaigns, including spend, impressions, and reach by campaign and ad group.

Why two streams? Because TikTok Ads Manager blends paid and organic GMV Max activity into a single reported figure. By pulling directly from both the order data and the MMM API independently, Fospha isolates the paid contribution - preventing the overcounting that inflates platform-reported ROAS.


Step 2 — Baseline Decomposition: What would have happened anyway?

Before attributing any revenue to GMV Max, Fospha's daily Marketing Mix Model establishes what TikTok Shop revenue would have been in the absence of paid ads - the organic baseline.

The model decomposes total TikTok Shop revenue into:

  • Organic / baseline revenue — driven by brand equity, seasonality, and non-paid factors
  • Paid ad contribution — the incremental revenue attributable to GMV Max ads and TikTok Shop ads specifically

This step is critical because without an organic baseline, any revenue that would have happened anyway gets credited to the ad - inflating ROAS. 

Step 3 — Impression-Based Attribution: Credit only where it's earned

TikTok Shop conversions are attributed to GMV Max campaigns using an impression-based methodology — not last-click. This matters because GMV Max is a discovery format; the purchase often happens on a different platform, at a different time.

The model attributes a conversion to a GMV Max ad only where it identifies a statistically meaningful relationship between GMV Max ad exposure (impressions by campaign) and purchasing behaviour. This is tested using NRMSE (Normalised Root Mean Squared Error) on held-out data the model has not been trained on - confirming the model's predictive accuracy rather than in-sample fit.

Incorporating GMV Max data has been shown to improve overall model accuracy. Validation is strongest for TikTok Shop and DTC; Amazon attribution is accurate at platform level with lower confidence at campaign level.

Step 4 — Halo Capture: Where else did it drive revenue?

This is the step that changes the number.

Customers who were exposed to a GMV Max impression and subsequently converted on DTC (web) or Amazon, without purchasing on TikTok Shop, are captured as halo contribution. The model identifies these customers using a combination of:

  • Cross-platform order matching - linking TikTok impression exposure windows to purchase events on DTC and Amazon within the same customer journey
  • AdStock decay modelling - quantifying how long GMV Max ad exposure continues to influence purchasing behaviour after the impression, across each platform separately

This produces a Total Commerce ROAS: total net revenue credited to GMV Max across all platforms (TikTok Shop + Amazon + DTC) divided by total GMV Max spend. Calculated on an order-based, net revenue methodology to enable like-for-like comparison with Meta, Google, and all other paid channels.


FAQ’s

Q: How does GMV Max compare to other TikTok ad formats like Smart+ or conversion campaigns?

A: Fospha measures GMV Max and Smart+ as separate channels, each with their own Total ROAS calculated on the same net revenue, order-based methodology. This means you can compare them directly in the Channel Performance view — something that is not possible in TikTok Ads Manager, where the two products report on different bases.

On cross-platform impact specifically: Fospha's AdStock modelling shows GMV Max has a shorter decay curve than most other channels, which is consistent with its role as a discovery format driving awareness and promotional spikes rather than nurturing consideration over a longer window. Whether GMV Max or another TikTok product drives more halo revenue will vary by brand, category, and channel mix.

Q: Why does Fospha's Total ROAS differ from what TikTok Ads Manager reports?

A: There are four main reasons. First, Fospha includes halo revenue - purchases completed on DTC or Amazon by customers who were exposed to a GMV Max impression - which TikTok Ads Manager cannot see. This is typically the largest driver of the difference. Second, Fospha separates paid and organic activity, whereas TikTok Ads Manager blends them into a single figure. Third, Fospha reports on net revenue at order level; TikTok Ads Manager reports gross revenue at SKU level. Fourth, Fospha deducts an organic baseline before crediting any revenue to GMV Max, so revenue that would have occurred without paid activity is not attributed to the campaign.

Q: Can I see Total ROAS and revenue broken down by channel - TikTok Shop, Amazon, and DTC separately?

A: Sales Channel Segmentation is currently on our roadmap and will be released shortly.

Q: If I reduce or pause GMV Max spend, how quickly would TikTok Shop sales be affected?

A: This depends on the AdStock decay curve for GMV Max in your model. GMV Max typically shows a shorter decay curve than channels like Meta - meaning its influence on purchasing behaviour fades more quickly after ad exposure. In practice, this means a reduction in GMV Max spend is likely to show up in TikTok Shop performance relatively quickly, rather than with a significant lag. Your specific decay parameters are visible in the Optimization Dashboard. Note that a portion of TikTok Shop revenue will also be attributed to organic factors and other ad products, so the impact of pausing GMV Max will depend on what share of sales the model is crediting to paid GMV Max activity.

Q: Can you explain the AdStock decay curve for GMV Max?

A: AdStock modelling quantifies how long an ad impression continues to influence purchasing behaviour after it was served. A shorter decay curve means the effect fades quickly; a longer one means residual influence persists over days or weeks. Fospha's modelling consistently shows GMV Max has a relatively short decay curve compared to channels like Meta — reflecting its role as a high-frequency discovery format rather than a consideration nurturer. This has practical implications for how you schedule campaigns around promotional windows, as the lift from GMV Max tends to be concentrated rather than sustained.

Q: Can the halo analysis be broken out by product category or customer cohort?

A: The current model operates at campaign and channel level rather than by product category or customer cohort. If you need performance broken down by SKU category or customer segment, this is not available in the current release. If this is a priority for your team, raise it with your Customer Success contact.

Q: How granular is GMV Max attribution - campaign, ad group, or creative level?

A: Attribution is allocated at campaign level for GMV Max and at ad level for TikTok Shop Ads. Creative-level attribution for GMV Max is not currently available in Fospha. For Amazon, attribution is accurate at platform level with lower confidence at campaign level, which is a function of the data available via the Amazon integration rather than a modelling limitation.

Q: How does Fospha isolate the impact of GMV Max on Amazon from Amazon's own advertising?

A: Fospha's model draws on DTC and Amazon order data alongside GMV Max impression data from the TikTok MMM API. The model attributes an Amazon conversion to GMV Max only where it identifies a statistically meaningful relationship between GMV Max ad exposure and the subsequent purchase - not simply where the two events coincided. Amazon's own ad activity (Sponsored Products, DSP, etc.) is treated as a separate channel in the model and is not conflated with GMV Max attribution. As noted above, Amazon attribution is reliable at platform level; campaign-level confidence is lower.