Skip to main content

GMV Max Model Explainer

This article explains how Fospha measures TikTok Shop GMV Max performance.

Arina Sugako avatar
Written by Arina Sugako
Updated over a week ago

Fospha gives you a clear, finance-ready view of GMV Max performance that’s directly comparable to your other channels.


Dive into how the model works

👀See GMV Max Measurement in action! Watch our 4-minute model explainer, or dive into the details below.

1. Data Ingestion

TikTok Shop Orders & Revenue

  • TikTok Shop Orders: Ingested directly from Shopify, TikTok Shop API, or manual upload.

  • Why it matters: TikTok’s 1P reporting inflates conversions and revenue:

    • Orders: TikTok counts SKUs (each item in a basket) as separate conversions. Fospha aligns to Shopify’s definition: 1 basket = 1 order.

      • Example: toothbrush + toothpaste = TikTok reports 2 conversions, Fospha reports 1.

    • Revenue: TikTok uses gross revenue (items paid – taxes + discounts). Fospha aligns to net revenue (gross – discounts + taxes + shipping).

      • Result: Fospha numbers are finance-ready and consistent with other channels.

GMV Max Ads Cost, Impressions and Clicks

  • Source: TikTok’s invite-only MMM API.

  • Granularity: Campaign-level (no ad/ad set breakdown).

  • Signals ingested: Cost, Clicks, Impressions.

Why it matters:

  • TikTok Ads Manager provides incomplete data for GMV Max.

  • MMM API is the only way to access a complete signal, combining upper-funnel (impressions) and lower-funnel (clicks) signals of GMV Max performance.

  • Fospha uses these raw signals to model impact — we do not rely on TikTok’s first-party attribution.

2. Modelling

  • Fospha runs a daily MMM to calculate the relative importance of GMV Max ads on TikTok Shop sales.

  • Both clicks and impressions are used as inputs — essential because GMV Max combines upper-funnel (impression-driven) and lower-funnel (click-driven) activity.

  • An XGBoost model learns the relationships between GMV Max impressions and clicks and TikTok Shop sales, capturing its full funnel impact.

  • Once the model learns these relationships, Shapley values are calculated. Shapley uses game theory to fairly split credit for conversions between all channels, showing how much each impression and click truly contributed.⁠⁠⁠⁠

  • Shapley values are then normalised so total contributions align with observed TikTok Shop sales.

Why it matters:

  • TikTok’s rule-based reporting gives 100% credit to GMV Max whenever a promoted SKU sells. Fospha instead captures the incremental impact of GMV Max — showing the true share of sales driven by ads.

  • This approach provides an independent read on GMV Max performance.

You can learn more about TikTok’s attribution of GMV Max here.

Granular Redistribution

  • Once the share of sales influenced by GMV Max is established, Fospha redistributes conversions down to campaign level.

Why it matters:

  • Marketers don’t just see whether GMV Max works overall — they can optimise within GMV Max, reallocating spend to the campaigns that are genuinely driving sales.


3. Crediting Organic

  • Any TikTok Shop sales not explained by GMV Max ads are attributed to TikTok Shop Organic.

  • Organic covers unpaid activity such as boosted posts, creator/affiliate sales, and direct shop purchases.

  • Today, TikTok does not expose a breakdown of these sub-sources, so Fospha reports them as a single organic category.

Why it matters:

  • You get a clear, independent split between paid GMV Max and organic TikTok Shop activity.

  • This prevents over-crediting paid campaigns for sales that were actually generated by organic demand, giving you a more balanced measurement output.


4. New vs Repeat in GMV Max Measurement

Fospha uses your order data (from Shopify, TikTok Shop API, or manual upload) to identify which TikTok Shop conversions came from new customers vs repeat customers.

  1. Establishing the overall split

    • If Shopify is connected, we use Shopify’s customer IDs to calculate the true overall % of new vs repeat customers across TikTok Shop.

    • If only the TikTok Shop API is connected, we calculate new vs repeat within TikTok Shop only. In this case, the same person could appear “new” in TikTok Shop and “new” on your web store, because GDPR prevents us from stitching identities across platforms.

  2. Applying the split to GMV Max

    • All TikTok Shop orders (GMV Max + organic activity) are pooled together at the start of the pipeline.

    • Our daily MMM then assigns credit for these conversions between GMV Max (paid) and TikTok Shop Organic, based on the relative importance of ads on sales.

    • Once this split is done, we apply the overall new/repeat percentage proportionally to both GMV Max and Organic conversions.

      Example: If Shopify shows that 35% of TikTok Shop orders were new, then 35% of the GMV Max-attributed conversions are marked as new, and 35% of the Organic conversions are marked as new.

  3. Why we do it this way

    • TikTok’s reporting doesn’t provide paid vs organic new/repeat detail.

    • Applying the overall percentage consistently ensures alignment with your source of truth while avoiding over-precision we can’t defend.

    • This approach will evolve as more reliable signals become available.


Why Fospha Models GMV Max This Way

GMV Max is unique: impressions spark awareness, while clicks validate the path to conversion. To capture this full impact, our model ingests both impressions and clicks rather than relying on a single signal.
How it works

  • Separate signals in the model: Impressions and clicks are modelled independently, each with its own Relative Importance (RI) score.

  • Integrated view after modelling: Once trained, we combine RI values, giving the GMV campaign fair credit for its role in driving sales.

  • Validation through training: When we tested impressions-only vs. impressions + clicks, the latter consistently delivered stronger model performance (lower NRMSE). This proved clicks add predictive power beyond impressions alone.

Why it matters

  • Captures full funnel impact: Impressions reflect top-of-funnel demand creation, while clicks confirm lower-funnel engagement. Using both avoids undervaluing GMV Max.

Why Fospha’s Numbers Differ from TikTok’s

Orders: SKUs vs. True Orders

  • GMV Max Orders (SKU-based)

    • GMV Max counts every SKU in a basket as a separate order.

    • Example: if a customer buys a toothbrush + toothpaste in one checkout, GMV Max records 2 SKUs = 2 orders.

    • For brands with larger basket sizes, this inflates conversions.

  • Fospha Orders (Shopify-based)

    • Fospha reports orders in the same way as Shopify and all your other channels: 1 basket = 1 order, regardless of how many SKUs are inside.

    • This keeps GMV Max measurement consistent with the rest of your reporting.

Revenue: Gross vs. Total (Net) Revenue

  • TikTok’s Gross Revenue (GMV Max reporting)

    • Includes the amount people pay, minus sales taxes, plus product discounts.

    • This means revenue can look higher, but doesn’t reflect the true net value you recognize.

  • Fospha’s Total Revenue (Shopify-based)

    • Gross Revenue – Discounts + Taxes + Shipping Charges.

    • Reflects the actual revenue your finance team sees.

    • Typically:

      • TikTok Shop discounts reduce revenue by ~25%.

      • Across other channels, discounts average closer to 10%.

    • This difference can significantly skew TikTok’s reported ROAS if you don’t adjust.

Organic: TikTok vs. Fospha

  • TikTok’s definition of Organic

    • Very broad: includes everything that is not explicitly reported as paid — boosted posts, creator affiliates, regular content, etc.

    • Essentially: “everything else that exists.”

  • Fospha’s definition of Organic

    • Fospha ingests all impressions and clicks from the GMV Max MMM API — including paid ads, affiliate videos, and organic activity.

    • Our model then measures the relationship between impressions/clicks and Shopify orders & revenue.

    • Example output:

      • 50% of TikTok Shop conversions are explained by GMV Max impressions/clicks → attributed to GMV Max.

      • The other 50% are not explained by GMV Max activity → attributed to Organic (within TikTok Shop).

    • Important: We cannot split TikTok’s impressions into “paid” vs “organic” buckets. What Fospha can show is the incremental effect of GMV Max, with the remainder assigned to Organic.

  • How to think about it:

    • Fospha’s Organic = TikTok Shop conversions not explained by GMV Max impressions/clicks.

    • This makes Organic in Fospha narrower and more realistic than TikTok’s all-inclusive Organic.

Based on Fospha’s beta cohort data:

  • Average basket = ~1.3 SKUs/order.

  • This leads to ~30% difference in reported order and revenue numbers between TikTok’s GMV Max and Fospha’s Shopify-based measurement.

  • Organic definitions differ:

    • TikTok = everything outside paid.

    • Fospha = only the conversions not explained by GMV Max activity.

Metric

GMV Max Reporting (TikTok 1P)

Fospha Reporting (Shopify-aligned)

Impact

Orders

SKU-based: each item in a basket counted as an order

Basket-based: 1 basket = 1 order

TikTok inflates conversions when baskets have multiple items

Revenue

Gross Revenue = Amount paid – Sales taxes + Discounts

Total Revenue = Gross Revenue – Discounts + Taxes + Shipping charges

Fospha aligns with finance-ready revenue; TikTok numbers often look inflated

Discounts

~25% average discount rate on TikTok Shop

~10% average discount across other channels

TikTok’s higher discounting lowers net value of sales

Organic

All non-paid activity lumped together

Conversions not explained by GMV Max impressions/clicks

Fospha shows true incremental impact of GMV Max

Consistency

Different definitions from other channels

Same definitions as Shopify, Meta, Google, Amazon

Apples-to-apples view across all channels

Takeaway: Fospha makes GMV Max results comparable and finance-ready by aligning definitions of orders, revenue, and Organic with Shopify and your other channels. This ensures you can measure GMV Max Ads apples-to-apples with your other marketing channels.


Did this answer your question?