Triple Whale Review (2026): Where the Attribution Earns Its Keep, and Where the Free Tier Misleads

This is a review of the product, not the price list. For the full GMV ladder, the annual formula, the band cliffs and the derived cost per order, see Triple Whale pricing — the short version is that Triple Whale prices on your annual gross merchandise value, and as of 2026-09-10 it is not displaying those prices on its own pricing page.

Everything below was verified on 2026-09-10 against triplewhale.com and Triple Whale's public help centre, with feature-level detail taken from the plan comparison matrix on the pricing page.

What Triple Whale is actually for

Triple Whale exists because every ad platform grades its own homework. Meta claims the order, Google claims the same order, your affiliate platform claims it too, and the sum of channel-reported revenue exceeds what Shopify actually banked — often by a wide margin. You cannot allocate budget from numbers that do not add up.

Triple Whale's answer has three parts, and they are genuinely well-chosen:

1. A first-party pixel — the Triple Pixel, with a mobile SDK — that observes the journey directly instead of trusting each platform's account of it.
2. One attribution model applied to everything, rather than seven models each applied to itself. Foundation and above unlock Linear All, Linear Paid, First Click, Last Click, Triple Attribution, Total Impact and Clicks & Deterministic Views, with 1-, 7-, 14-, 28-day and lifetime lookback windows.
3. A second, independent signal — a native post-purchase survey collecting zero-party data on channel influence, folded back into the model Triple Whale calls Total Impact Attribution.

That third part is the one worth crediting most. Asking the customer is not a substitute for tracking, but it is genuinely independent of it, and a model that reconciles observed behaviour against stated influence is better evidence than either alone. It is also the piece that survives browser privacy changes best, which is not an accident.

What it does well

The reconciliation job, done properly. This is the core and it works. One consistent model across Meta, Google, TikTok, Snapchat, Pinterest, Microsoft, X, Taboola, Outbrain, Criteo, AdRoll, MNTN and more, plus Amazon and — on paid tiers — TikTok Shop and Walmart. For a brand buying across four or more channels, the reconciliation alone justifies a serious evaluation.

The BI layer is real BI, not a dashboard skin. Foundation includes a no-code custom dashboard builder, a custom metric builder, a segment builder, and a SQL editor against your raw data, plus 50+ dashboard templates covering creative, LTV, site performance, keywords and inventory. Cohort analysis, product analysis down to SKU with product-level ROAS and LTV, multi-geo and multi-store reporting are all in the same tier. A lot of competitors in this category ship a fixed set of charts; Triple Whale ships an analytics environment. If you have an analyst, they will not hit the ceiling quickly.

Unlimited users on every paid plan. This deserves more credit than it gets, because the category norm is per-seat billing that quietly triples the bill for a team. Triple Whale's own FAQ commits to it: every paid plan includes unlimited users and as many stores as needed. Finance, ops, the founder and three media buyers can all have logins at no incremental cost. Free is capped at 10 users, which is still generous.

Data retention that does not punish you for staying. Free gets 12 months. Every paid tier is unlimited. Historical data is the raw material of every cohort and LTV question, and tools that expire it are quietly degrading the thing you bought them for.

It is not a closed box. Both a Data-In API and a Data-Out API are available on every tier including Free, so you can push arbitrary sources in and sync the enriched output back out. Triple Whale MCP — a read-only connection letting external AI tools like Claude and ChatGPT query your live Triple Whale data — starts at Foundation. For a vendor whose commercial interest is being the place you look, shipping a read-only MCP endpoint and a Data-Out API is a real openness decision.

Moby has moved past chat. On Automate, Moby connects to your ad accounts and can scale budgets, pause underperformers and launch campaigns, with every action logged and either individually approved or executed automatically inside guardrails you define. Creative Studio edits images, text and backgrounds and produces ad variations; Landing Pages generates Shopify-native pages. Triple Whale publishes its own counters for this — over 1M conversations, 100K+ automations run, 45K+ creatives generated, 15K+ actions taken. Those are vendor-reported and unaudited, but the ratio is the interesting part: actions taken is by far the smallest of the four numbers. Read it as a capability that is real and in early adoption rather than as the centre of how customers use the product today.

The AI meter fails safe. Moby usage is metered against the LLM cost Triple Whale incurs on your behalf. You get a warning at 80 percent and at 100 percent Moby work pauses cleanly — Triple Whale's stated position is no surprise bill and no silent overage, with an optional top-up sized to the rest of the cycle. Given how many AI products chose the opposite default, choosing a hard stop over an uncapped invoice is the right call and should be credited.

Where it falls short

The Free tier cannot demonstrate the product

This is the sharpest criticism on the list, because it shapes how most people form an opinion.

The Free tier is genuinely generous — 10 users, 12 months retention, the Triple Pixel and mobile SDK, Moby chat and analytics, Moby Artifacts and Audiences, web search, image and video creative generation, a light post-purchase survey, and the whole integration catalogue.

But on Free, attribution is first and last click only, with a lifetime lookback window only. Multi-touch attribution: not included. Total Impact Attribution: not included. Every other attribution model and every other lookback window: not included. Also absent are ad-platform controls, CSV export, the dashboard template library, automated reports, the custom metric builder, the SQL editor, custom dashboards, the segment builder, cohort analysis, product analysis, creative analysis and MCP access.

So the tier that most evaluators judge Triple Whale on is missing both the attribution it is famous for and the BI layer that is arguably its strongest feature. The correct use of Free is to install the pixel early and start accumulating history — not to decide whether the product works.

The meter is your revenue, not your usage

Triple Whale prices by annual GMV band. That has a defensible logic and it is progressive in the buyer's favour at scale, but it produces one clear inequity: two brands doing identical GMV with identical channel mixes pay identically even if one processes four times as many orders as the other. A high-AOV brand is paying several times more per tracked order for the same tracking work. It also means a low-margin business subsidises a high-margin one at the same top line. The arithmetic is worked through on the pricing page; as a product criticism, the point is that the meter is disconnected from the work performed.

Everything that acts is behind the top self-serve tier

The split between Foundation and Automate is not a handful of extras. Automations, Actions, Creative Studio, Landing Pages, Industry and Peer Benchmarks and the full AI model library are all Automate-only, and Segment Syncing, Custom Events, Path and Funnel Analysis, Site Search Analytics and Page Speed Analytics are add-ons on Foundation that come included on Automate.

Foundation is therefore a very good analytics product, and Automate is a different product that happens to contain it. If you are evaluating Triple Whale because you saw a demo of Moby executing a budget change, be clear with yourself that you are evaluating Automate.

The unpublished AI allowance sits inside a 12-month commitment

Triple Whale describes Moby usage qualitatively — Explore Moby, Use Moby daily, Run Moby at scale, Enterprise grade Moby usage — with no published unit, no published quantity per band and no published top-up rate. It also selects the underlying model on your behalf, choosing for what it calls quality, consistency and usage efficiency, with direct model selection reserved for Automate and above.

Individually each of those is reasonable. Together they mean the practical capacity of the AI you are buying is undisclosed, controlled by the vendor, and locked in for twelve months, since all three paid tiers are 12-month subscriptions regardless of billing cadence. That is the single term most worth pinning down in writing before signing.

Meaningful support starts at $10M GMV

Live chat is included from Foundation. A dedicated Customer Success Manager and a dedicated Implementation Specialist are marked Qualifying accounts on Foundation and Automate, and Triple Whale's own tooltip defines that as brands with GMV of $10M+. A $3M brand paying $1,799 a month for Automate is a self-serve customer with a chat window. Nothing wrong with that — but price the onboarding effort as yours.

The defensibility ladder concedes the attribution critique

Triple Whale sells Compass — MMM, incrementality testing via GeoLift and Conversion Lift studies, incremental revenue reporting, and what it calls auditable incrementality proofs — as the layer that makes measurement statistically defensible, and it is included only on Enterprise, with an add-on route elsewhere. Triple Whale's own documentation recommends Compass for brands at roughly $10M+ annual GMV so the models have enough data.

Read that honestly and it is an admission that multi-touch attribution alone is directional. That is true of every MTA product and Triple Whale is more candid about it than most. But it means the answer to "how do I know these numbers are right" is a separate, quote-only product with a de facto revenue floor. Below $10M GMV, you are buying the model and trusting it.

Who it fits

Buy it if you are:

• A DTC or ecommerce brand doing roughly $1M+ in GMV, buying across three or more paid channels, where channel-reported revenue visibly exceeds banked revenue.
• A team with an analyst or a technical marketer who will actually use the SQL editor, segment builder and custom dashboards — that is where the Foundation price converts into leverage.
• A multi-store or multi-geo operator. Consolidated reporting across stores and regions is included from Foundation and is expensive to build yourself.
• A brand with a low-to-mid AOV and high order count, which is the side of the GMV meter that comes out ahead.
• Somewhere with an existing AI workflow you want fed with real commerce data — the read-only MCP endpoint is a genuinely useful integration point.

Do not buy it if you are:

• Lead-gen or B2B. The data model is orders, the attribution is orders, the pricing is GMV. Very little of it maps.
• Single-channel. If 90 percent of spend is one platform, most of the value is reconciliation you do not need, and the platform's own reporting plus a holdout will get you close for nothing.
• Bottlenecked on creative, not on measurement. If you already know which ad is fatiguing and the delay is producing the replacement, a better dashboard measures the delay rather than removing it. Automate's Creative Studio addresses this and is the most expensive rung of the ladder.
• Under roughly $500K GMV, where the automation tier is a large share of gross revenue and the Free tier plus platform-native reporting is a reasonable stage-appropriate answer.

The reconciliation conversation to have before you sign

The most common source of buyer's remorse in this category is not the product — it is the week three discovers Triple Whale's revenue number does not match Shopify's, and neither matches Meta's. All three are correct within their own definitions of a conversion, a window and an attributable touch.

Agree in advance, in writing, which number governs which decision: which one finance closes the month on, which one the media buyers optimise against, which one goes in the board deck. Triple Whale will not settle that for you, and a tool that disagrees with your system of record is not defective — it is doing the job you bought it for.

The verdict

Triple Whale is a strong, mature ecommerce measurement platform with a real technical asset in the Triple Pixel, an unusually good BI layer for the tier it sits in, an admirably buyer-friendly stance on seats and data retention, and an AI meter that fails safe instead of failing expensive. For a mid-market DTC brand buying across several channels, it is a credible answer to a question that genuinely costs money to get wrong.

Its weaknesses are structural rather than sloppy. The single biggest one: the Free tier — the version most people form their opinion on — has multi-touch attribution switched off, so the product's headline capability cannot be evaluated without a 12-month commitment. Around that sit a meter tied to revenue rather than usage, an execution layer gated behind the top self-serve tier, an AI allowance whose size is not published, and a defensibility story that starts at $10M GMV.

And the honest framing for anyone comparing it against an ads-execution tool: since Moby Actions and Creative Studio shipped, Triple Whale is no longer purely measurement, and any comparison saying otherwise is out of date. What has not changed is the shape of the purchase. Triple Whale's execution arrives attached to a GMV-priced, ecommerce-anchored annual measurement contract. Soku comes at it from the other end — a chat-driven AI marketing agent that generates the ad creative and launches and optimises campaigns across Meta, Google and TikTok toward a ROAS target, free to start, and applicable to lead-gen as well as ecommerce. It does not do multi-touch attribution and does not claim to. Running both is a coherent choice, and the trade-offs are laid out on the Triple Whale alternative comparison.

Product capabilities, tier gating, plan terms and support thresholds verified 2026-09-10 against triplewhale.com and Triple Whale's public help centre. Vendor-published usage counters are unaudited and reproduced as claims.