Your Google Ads dashboard says your campaigns returned 6x last month. Meta says its campaigns returned 4x. Add up everything each platform claims credit for and you’ve apparently generated more revenue than your Shopify store actually recorded. Every marketer who has managed more than one channel has run into this, and in 2026 the problem is getting worse, not better.

This post covers why platform-reported numbers have drifted further from reality, the framework we use to evaluate third-party attribution platforms, and what we found when we recently ran that evaluation for a client selling across five channels, including one that no attribution platform on the market can natively track.

Why attribution and impact have diverged

Platform attribution was always self-graded homework. Google reports on Google’s contribution using Google’s model, and unsurprisingly, Google’s contribution always looks good. The same is true of every ad platform. But for a long time, the numbers were at least directionally useful because the customer journey mostly happened in places a pixel could see.

That’s no longer a safe assumption. A growing share of product research now happens inside AI Overviews, AI Mode, ChatGPT, and Gemini. These are surfaces where a shopper can compare options, narrow a shortlist, and form a brand preference without a single trackable session on your site. When that shopper finally acts, they often search your brand name and click the first thing they see. Your branded Search campaign books the conversion at a spectacular CPA, and whatever actually created the demand gets nothing.

Branded search has always been the clearest example of attribution confused with impact. It reaches people who already know what they want. It will always look efficient. The question that matters, for branded campaigns and every other line item in your account, isn’t “which channel got credit for this conversion?” It’s “what would have happened if this spend didn’t exist?”

The three questions a measurement stack needs to answer

Those are different questions, and they need different tools.

When a client asks us about attribution tooling, we translate the request into three practical questions before we look at a single vendor:

Where should the next dollar go? Cross-channel budget allocation is the real job most brands are using an attribution platform for. If a tool can’t help you decide whether to move budget from Meta to Google Shopping (or tell you that both are saturated and the next dollar belongs in email), it’s a reporting layer, not a decision-making layer.

What’s actually converting? Deduplicated, cross-channel conversion tracking. Every platform claiming full credit for the same order is the core issue; a third-party view that reconciles claimed conversions against actual store orders is what we want to see.

Are we profitable, or just converting? Revenue ROAS hides a lot of sins. A tool that incorporates COGS, margin by product, cohort behaviour, and customer lifetime value tells you whether your “winning” campaigns are actually making you money, or just efficiently acquiring your least profitable customers. There is a lot of talk these days about profit-based bidding and conversions with cart data; the same logic applies at the measurement layer.

If a platform doesn’t move the needle on at least two of these, the monthly fee needs to be reassessed.

Our evaluation framework

We recently ran this process end to end for a client with a genuinely awkward channel mix: Google Ads, Microsoft Ads, Meta, Amazon Ads, and Etsy Ads. Five channels, one Shopify store, and a marketplace channel with a completely closed ads ecosystem. We shortlisted five platforms and scored them against the three questions above, plus a few practical criteria:

Channel coverage. Native integrations for every paid channel, or as close as possible. Manual data stitching is where attribution projects go to die.

Attribution methodology. Multi-touch attribution, marketing mix modelling, first-party pixel tracking, or some combination. MTA is more granular but degrades as tracking degrades; MMM is more robust to signal loss but needs spend history and works at a higher altitude. In 2026, we weight toward platforms that offer both, because MTA alone is measuring a shrinking share of the journey.

Profitability depth. Does it stop at revenue, or does it get to margin? COGS support, cohort analysis, and CLV reporting separate the analytics platforms from the pretty dashboards.

Platform fit. A tool built specifically for Shopify eCommerce brands will be running in days. A general-purpose enterprise platform will be running in quarters.

Cost and setup burden. The best attribution platform is one the client will actually keep using after month three.

What we found

Our recommendation for this channel mix was Klar. It was the only platform we reviewed that addressed all three core questions in a single tool: multi-touch attribution for conversion tracking, profitability and margin reporting including COGS, cohorts and CLV, and a Marketing Mix Model for budget allocation guidance. It has native integrations for Google, Microsoft, Meta, and Amazon, and it’s built specifically for eCommerce brands on Shopify, which kept the projected setup burden low.

The runner-up was Lebesgue, and for a lot of brands it would be the right answer. It covers Google, Microsoft, Meta, and Amazon, offers first-party attribution via its own pixel, and includes a genuinely useful competitor intelligence layer that tracks competitor ad creatives and spend trends. Core analytics pricing starts at $79/month, and setup is fast. The tradeoff is shallower profitability reporting and less robust MMM capability. That’s fine if allocation isn’t your primary question, limiting if it is.

The important caveat: this recommendation was specific to this client’s channel mix, spend level, and Shopify stack. Run the same framework against a different brand and you may land somewhere else entirely.

The Etsy problem (and the closed-ecosystem lesson)

Here’s a finding that surprised the client, and that you won’t find in any vendor’s comparison chart: no third-party attribution platform currently offers a native Etsy Ads integration. Not one. And that’s unlikely to change, because Etsy’s ecosystem is closed by design. The marketplace owns the customer relationship and has no incentive to expose click-level data to outside tools.

The practical workaround is manual ad cost uploads, which most serious platforms (including Klar) support. You won’t get click-path attribution for Etsy traffic, but you can at least account for the spend in blended profitability reporting, which keeps your total marketing efficiency numbers honest.

The broader lesson applies beyond Etsy: any closed marketplace channel (and there are more of them every year) will be a blind spot in your attribution stack.

Where each measurement layer fits

The mistake we see most often is treating an attribution platform as the finish line. It isn’t. It’s the middle layer of a three-layer stack, and each layer answers a different question:

Platform data is fast, free, and biased. Use it for in-platform optimization: bid strategies need conversion signal, and platform data is what feeds them. Just don’t use it to compare channels against each other.

Third-party attribution and MMM tools give you a deduplicated, cross-channel view and better allocation logic. But they’re still models. A model with better assumptions is still an opinion about the truth, not the truth.

Incrementality testing (geo holdouts, spend pauses, matched-market tests) is the only ground truth available. It directly answers “what would have happened without this spend?” The cost is time, money, and the discipline to actually turn campaigns off. Most brands should run at least one meaningful incrementality test a year on their biggest line items, starting with branded search.

The layers check each other. When your MMM says a channel is incremental and a holdout test agrees, you can allocate with confidence. When they disagree, you’ve found the most valuable question in your account.

Who actually needs a tool

Not everyone. Some honest guidance by stage:

If you’re running one or two channels at modest spend, platform data plus your Shopify order reports plus common sense is usually enough. Save the subscription fee.

If you’re running three or more paid channels, spending enough that a 10% misallocation costs real money, or making decisions on margin rather than revenue, a third-party platform starts paying for itself quickly.

If you’re spending six figures a month or making structural budget decisions, layer in incrementality testing. At that spend level, “the model said so” isn’t a good enough answer.

Whatever you choose, validate before you commit. Most platforms in this space offer free trials (Klar’s is 14 days with free onboarding), and two weeks of your real data will tell you more than any comparison post, including this one.

If you want help evaluating attribution options against your channel mix, or you suspect your platform-reported numbers are telling you a flattering story, this is part of the measurement and audit work we do at Take Some Risk. Get in touch.