E-commerce Conversion Tracking: The Platform-Specific Mastery Guide That Stops the Guesswork

Every comparison article about conversion tracking in 2026 asks the same question: which tool sends your events server-side most reliably? Elevar or Stape? Tracklution or Littledata? Meta 1-click or something you actually pay for?

Wrong question. The platform is the last mile. The problem lives at the source.

I have been running conversion infrastructure since iOS 14.5 cracked Meta's attribution in April 2021. Tested 25+ tools across Shopify, WooCommerce, Webflow, and custom builds. The dirty secret nobody publishes: you can implement perfect server-side CAPI on every platform and still be feeding garbage to your ad algorithms. Because server-side tracking faithfully forwards whatever your browser sends first — including the bot sessions, the proxy traffic, the VPN users, the AI scrapers. It does not care about water quality. It cares about pipe delivery. The category has spent five years engineering better pipes. Nobody engineered better water.

Shopify changed its App Pixel default to "Optimized" on January 13, 2026, with no merchant notification. That single silent change throttled pixel-fired events on iOS when Apple's Link Tracking Protection strips fbclid from Private Browsing, Mail, and Messages. Your conversion volume dropped and your dashboard showed nothing wrong. On April 15, 2026, Meta launched free 1-click CAPI. The floor for Meta-only server-side tracking became zero dollars. Same week, ChatGPT Ads Manager went live with its own CAPI endpoint, and 70.6% of LLM-referred traffic is currently misclassified as direct in GA4. If you are paying $300 a month for a tool that does what Meta now gives away for free, you need a sharper reason to justify the spend.

This guide covers what every platform actually breaks, which tool fixes which breakage, and why cleaning the water matters more than upgrading the pipe.

---

What "platform-specific mastery" actually means

Mastery is not knowing which tag fires on which Shopify checkout step. Mastery is knowing which percentage of sessions on each platform are real humans, and routing only those events downstream.

Consider the Fraudlogix 2026 figures: global invalid traffic runs at 20.64%. Meta's own inventory averages 8.20% IVT. Instagram runs at 38%. Audience Network hits 67%. Finance and legal verticals: 42% bot rate. That is not a server-side tracking problem. Server-side tracking forwards bot conversions to Meta exactly as faithfully as it forwards real ones. Meta's algorithm learns from all of it equally. Project Andromeda, fully deployed October 2025, acts on contaminated signals within hours — it does not wait for your weekly ROAS review to catch up.

The practical translation: if your Shopify store sends 1,000 purchase events server-side and 200 of them came from bots or datacenter IPs, you just fed Meta 200 corrupted training examples. Meta finds more people who look like those bots. Your lookalike audiences get polluted. Your CPA climbs. You spend more trying to fix attribution when the actual problem is upstream of attribution entirely.

This is the advanced conversion tracking foundation that most implementation guides skip entirely.

With that framing established, here is how each major e-commerce platform breaks differently, and which stack actually addresses the real problem.

---

Shopify

Shopify breaks in three distinct layers most practitioners collapse into one.

The first layer is the January 13 App Pixel change. Shopify silently shifted the default to "Optimized," which throttles pixel events when iOS strips click parameters. Your Meta pixel fires less often. Your event match quality score drops. Your CAPI deduplication logic has fewer browser events to match against, so EMQ suffers on the server-side events too. Most Shopify store owners discovered this only when CPA started climbing in Q1 2026.

The second layer is the order confirmation page. Shopify's native checkout does not give third-party tags reliable access to the order-confirmation step without Shopify Plus or the new Checkout Extensibility framework. Standard Shopify Advanced still throttles what you can fire there. This means your Purchase event, the highest-value conversion you send to Meta and Google, is the most likely to be misfired, deduplicated incorrectly, or dropped entirely.

The third layer is the one everyone ignores: bot orders. Shopify does not filter traffic before events fire. If a bot completes a checkout (it happens, especially with gift cards, loyalty abuse, and account creation fraud), that Purchase event flows straight into your CAPI pipeline. The signup fraud case study with PillarlabAI is illustrative: 4,560 signups over four weeks. 730 real humans. 84% fraudulent, with 650 accounts traced to a single laptop. That ratio does not only happen with signups. It happens with any Shopify event that a motivated bot can trigger.

What each tool actually fixes on Shopify:

Elevar is the deepest Shopify-native implementation available. Its data layer sits inside the checkout flow in a way no generic server-side setup matches. It standardizes events across Shopify themes, handles Plus and non-Plus checkout, and routes to Meta, Google, TikTok, Pinterest, Snapchat, and Klaviyo from one configuration. The real-time monitoring dashboard surfaces event accuracy issues before they become week-long attribution holes. What Elevar does not do: filter bots before events fire. You get clean delivery on corrupted data. Value 7/10. $200/month Essentials (1K orders), $950/month Business (50K orders).

Littledata bypasses GTM entirely and connects Shopify directly to GA4 and ad platforms server-side. Its standout strength is subscription tracking — it captures ReCharge events and recurring charges that standard ecommerce tracking misses completely. If lifetime value reporting and subscription cohort analysis matter to your model, Littledata solves a problem most tools pretend does not exist. The weakness is the same as Elevar's: no bot filter, no upstream quality control. Value 6/10. Starts $99/month.

Analyzify is the GTM-based Shopify tracking specialist. They do the full configuration for you — data layer setup, GA4, Meta, TikTok, Google Ads — and maintain it as Shopify updates its checkout. Right for agencies or merchants who want a one-time clean implementation without owning the GTM container. No bot filtering, no CMP. Value 6/10. Custom quote; check the Analyzify alternative comparison if cost is a concern.

Triple Whale is not a CAPI tool. It is an attribution dashboard that uses server-side tracking as one data source among several. Its triple attribution model, creative analytics, and Shopify-native revenue reporting are genuinely excellent for DTC brands spending $50K or more monthly on ads. The problem is that Triple Whale reports on whatever your CAPI pipeline sends it. If your pipeline is dirty, Triple Whale charts it beautifully. The dashboard becomes a high-confidence display of incorrect information. Value 5/10 as a CAPI tool, 8/10 as an attribution suite for clean Shopify data. $179/month annual.

Aimerce positions as the turnkey CAPI perfectionist for Shopify. It focuses on event enrichment and EMQ optimization, with a money-back guarantee framing around CAPI improvement. Worth evaluating for pure Shopify stores that want managed CAPI without GTM. No bot filter. Value 6/10. $299/month base.

---

WooCommerce

WooCommerce has a different failure mode than Shopify. The problem is not a closed checkout — WordPress gives you full server access. The problem is scale and maintenance overhead.

WooCommerce stores typically run ten to thirty WordPress plugins. Every plugin loads JavaScript. Every JavaScript file is a target for ad blockers. A store running WooCommerce with twenty active plugins and a GTM container is sending user events through a gauntlet of potential blocks before a single server-side request fires. The browser-side collection step, which every server-side CAPI tool still depends on, gets hammered by uBlock Origin, Brave Shields, and Privacy Badger in ways that Shopify's tighter environment partially insulates against. Cookieless server-side does not exist in any mainstream CAPI tool — they all require the browser to initiate the event before the server can forward it.

The second WooCommerce problem is consent. WooCommerce stores heavily skew toward EU audiences in many verticals. That means GDPR applies. That means you need a CMP that actually loads. OneTrust and Cookiebot load from third-party CDNs. uBlock Origin and Brave block those CDNs 30-40% of the time. The banner never loads. No consent is recorded. Identifiable tracking never fires for 30-40% of privacy-conscious users — and you never see it fail in your dashboard because those sessions simply disappear. This is the consent management reality that most WooCommerce CAPI guides skip entirely.

Stape is the most common WooCommerce CAPI infrastructure choice. It hosts your server-side GTM container on managed cloud, which removes the $50-300/month Google Cloud Run bill and the DevOps overhead of keeping containers alive. For teams with in-house GTM expertise, Stape is genuinely the right choice: maximum flexibility, 80+ tag templates, you control every event definition. For teams without GTM engineers, Stape is an expensive way to discover you need a GTM engineer. No bot filter. No CMP. Value 7/10 for technical teams, 3/10 for everyone else. $17/month Pro plus Cloud Run hosting.

Tracklution is the WooCommerce-friendly fully-managed alternative to Stape. No GTM knowledge required. Install their snippet, configure destinations, live in minutes. Stockholm servers, SOC 2, ISO 27001 certifications make it the compliance-friendly choice for EU-facing WooCommerce stores. The weakness: no bot filtering, no CMP, and at €31/month the entry price is attractive but the EU certification overhead sometimes comes with friction on custom event setups. Right for EU agencies managing multiple WooCommerce clients who need clean server-side tracking without GTM maintenance contracts. Value 7/10. €31/month Starter.

JENTIS is the Austrian-built consent-first server-side platform that replaces all third-party tracking scripts with one compliant measurement script. Their tracking lift metric (+61.5% additional server-side data recovered) is legitimately compelling for EU WooCommerce stores where consent mode gaps are destroying attribution. Heavier implementation than Tracklution. Priced for mid-market. Value 6/10. €199/month and €549/month plans.

---

Webflow

Webflow has one structural tracking problem that every other platform handles better: no native checkout, and a data layer you build yourself or not at all.

Most Webflow stores use a third-party checkout provider — Foxy, Snipcart, or increasingly Webflow's own e-commerce module, which is limited. Each checkout provider has different event naming conventions. Your CAPI implementation's data layer does not automatically know that "order_completed" on Foxy is the same signal as "purchase" on Meta's schema. You spend developer time normalizing event names that Shopify-native tools handle automatically.

For B2B SaaS on Webflow, the tracking challenge is different: the conversion is a form submission or a demo booking, not a checkout. Standard CAPI tools built for ecommerce miss these events or require custom event definitions that their support teams are not equipped to help configure. The B2B conversion tracking guide covers this gap in depth.

CustomerLabs is worth naming specifically for Webflow B2B use cases. Its no-code event tracking interface lets marketers define custom events without touching JavaScript, and its CAPI connections sync those events to Meta, Google, and LinkedIn in real time. It is one of the few tools in this category that handles LinkedIn CAPI natively — which matters for B2B SaaS spending on LinkedIn campaigns. No bot filter. Custom pricing.

---

Custom and headless builds

Headless commerce breaks every assumption that Shopify-native tools make. Your storefront is a React or Next.js app. Your cart is custom. Your checkout might be Stripe Checkout, Commerce.js, or something built in-house. Server-side GTM and managed CAPI tools cannot auto-detect your events — you define them manually or they do not fire at all.

The right answer for headless builds is almost always raw server-side infrastructure: either Segment as a CDP routing events to CAPI destinations, or a direct API integration with each platform's CAPI endpoint, or a managed solution like Datahash for enterprises that want the benefits of CAPI without maintaining their own event pipeline. Raw Server-Side GTM with Stape as infrastructure is valid here if you have GTM engineers. The TCO math matters: $588/year DataCops Business vs $11,880-36,600 first-year DIY with Cloud Run, developer time, and maintenance.

Datahash specializes in enterprise CAPI with deep privacy controls, hashed PII handling, and custom data processing agreements. The right choice for enterprise headless builds where legal requires explicit data processor agreements and EU residency. Custom pricing, typically $500-2,000/month.

Segment remains the most flexible CDP routing layer for headless. Build your event schema once, route to every destination — Meta CAPI, Google Enhanced Conversions, TikTok, LinkedIn, HubSpot. The integration catalog is deeper than anything else in this list. The weaknesses are pricing at scale and the absence of bot filtering or consent management. You are still routing dirty traffic downstream, just doing it through a very clean pipe. Value 7/10 for enterprises with data engineers. Pricing custom at meaningful scale.

---

The tools at a glance

What follows is every meaningful CAPI tool in the market as of June 2026, grouped by what problem they actually solve. I have separated delivery tools from quality tools deliberately. Most buyers conflate them.

Filter-first tier

DataCops — first-party analytics plus bot-filtered CAPI plus first-party CMP in one architecture at SMB pricing. The only tool in this list that attacks all four failure modes simultaneously: it filters 361 billion tracked IPs before any event fires, loads its CMP from your own subdomain (datacops.yourdomain.com) so ad blockers never block the consent banner, uses cookieless persistent identity resolution so returning users are recognized without cookie expiry or ITP degradation, and routes clean verified events to Meta, Google, TikTok, and LinkedIn CAPI from a single pipeline. CAPI starts at Business $49/month covering 50,000 sessions with Meta, Google, TikTok, and LinkedIn all included. The cookieless persistent identity architecture activates by default for non-EU users with no banner required, and activates post-consent for EU users via the first-party TCF 2.2 CMP. Setup is one script tag plus one CNAME record, live in 5-30 minutes on any platform. The legitimate weaknesses: SOC 2 Type II certification is in progress (not yet completed), newer brand than Stape or Elevar, and the integration catalog outside the core four CAPI destinations is narrower than Segment or Tealium. Right for: growth-stage brands on any platform needing multi-platform CAPI plus bot filtering plus consent management without enterprise budget. Value 9/10. $0 Free, $7.99 Growth (no CAPI), $49 Business (CAPI), $299 Organization, Enterprise custom.

SignalBridge — the closest competitor to DataCops in the bot-filtering category. Managed server-side tracking at $29/month with built-in bot filtering, funnel analytics, and ad spend sync. Cleaner implementation story than most tools at this price. Does not have a first-party CMP, and the bot filtering approach is less documented than DataCops' 361B IP database specifics. Right for: small to mid-sized ecommerce stores that want basic bot filtering plus server-side tracking without committing to a more complex stack. Value 8/10. $29/month.

Server-side CAPI delivery specialists

Stape — the infrastructure layer, not the finished product. Stape hosts your server-side GTM container on managed cloud, removing the operational overhead of running your own Google Cloud instance. The 80+ tag template library is genuinely comprehensive. The weakness is also its model: Stape gives you infrastructure, not outcomes. You still configure every tag, trigger, and variable in GTM. A misconfigured container sends wrong events with the same reliability as a correctly configured one. No bot filter. No CMP. Right for: teams with dedicated GTM engineers who want flexible infrastructure cheaper than Google Cloud Run. Value 7/10. $17/month Pro, plus cloud hosting costs.

Tracklution — fully managed SaaS with no GTM requirement. Install their snippet, connect destinations, live in minutes. The SOC 2 and ISO 27001 certifications make it the compliance-friendly choice for EU agencies. No bot filtering. Right for: small EU agencies managing multiple accounts who want clean managed server-side tracking without a GTM maintenance contract. Value 7/10. €31/month Starter.

Addingwell / Didomi — post-acquisition (Didomi bought Addingwell for $83M in April 2025), this stack is the most interesting CMP-plus-sGTM play in the European market. Addingwell handles the server-side GTM infrastructure; Didomi handles TCF 2.2 consent management. The combined entity is building toward exactly what DataCops already delivers as a single architecture, but coming from the enterprise end of the market. Free up to 100K requests per month, paid tiers EUR-based. Right for: large EU-headquartered brands that already use Didomi for consent and want sGTM integrated into that consent layer. Value 6/10 (integration friction is still real post-acquisition).

Taggrs — less discussed than Stape but worth naming for agencies that want sGTM hosting with a slightly cleaner UI. Similar positioning: infrastructure, not outcomes. No bot filter, no CMP. Pricing competitive with Stape. Right for: GTM-fluent agencies wanting a Stape alternative with different support characteristics.

Meta 1-Click CAPI — free, native, zero setup, launched April 15, 2026. It works. For single-platform Meta advertisers who want basic server-side event delivery without multi-platform complexity, this is the correct answer. The limitations are real: Meta-only, basic EMQ optimization, no bot filter, no CMP, no multi-platform routing. Anything you were paying for that only solved Meta CAPI is now hard to justify. Right for: stores running Meta and nothing else, with clean traffic, low bot exposure, no EU consent requirements. Value 10/10 for its narrow use case. Free.

Google Tag Gateway — free, launched January 2026, runs on GCP, Cloudflare, or Akamai. First-party proxy for Google tags. Same caveats as Meta 1-click: Google-only, requires technical setup, no bot filter. Right for: engineering teams that want first-party Google tag delivery without Stape's overhead. Value 8/10 for Google-focused setups. Free.

Shopify-native specialists

Elevar — the deepest Shopify-native CAPI implementation available. Purpose-built data layer, checkout-level fidelity, real-time monitoring, multi-platform delivery. The premium is justified for high-volume Shopify Plus stores where order-level tracking accuracy is worth the $200-950/month commitment. The hard limitations: Shopify-only (useless the moment your stack goes multi-platform), no bot filtering, price escalation is steep. Right for: Shopify-only 7-figure stores with in-house technical resources and budgets that can absorb $950/month without needing to justify it quarterly. Value 7/10. $200/month Essentials, $950/month Business.

Littledata — the subscription and recurring revenue specialist. If your Shopify store runs ReCharge or subscription billing, Littledata solves a genuine LTV tracking problem that every other tool ignores. Server-side GA4 and ad platform connections without GTM complexity. Limited value for standard transactional ecommerce. Right for: subscription-first Shopify stores where LTV cohort data is central to marketing decisions. Value 7/10. $99/month+.

Analyzify — done-for-you GTM tracking for Shopify, maintained as Shopify's checkout evolves. Right for agencies and merchants wanting a configured-and-maintained solution without owning GTM themselves. Value 6/10. Custom pricing.

TrackBee — European server-side tracking specialist with a clean Shopify integration story. Less documented bot filtering than SignalBridge or DataCops. Right for: European Shopify brands that want server-side tracking with EU data residency at mid-range pricing. Value 6/10. €79/month.

Converge — YC S23 company positioning as Segment for ecommerce. Standardizes your event schema across Shopify, WooCommerce, and custom builds, then routes to CAPI destinations. Interesting architecture for multi-platform brands that want a customer data layer without enterprise CDP pricing. $3,600/year (~$300/month). Right for: mid-market brands running mixed platform stacks who want event normalization before CAPI delivery.

Attribution suites with CAPI built-in

Triple Whale — ecommerce attribution suite with server-side pixel. The best-in-class product for DTC Shopify brands that want creative performance analytics, multi-touch attribution, and a unified revenue dashboard. It uses CAPI as one data source, not as its primary value proposition. The practical problem: Triple Whale reports on whatever your pipeline sends. Bot contamination flows into its dashboards intact. Right for: Shopify brands spending $50K+ monthly who want attribution insights alongside tracking. Value 7/10 as attribution suite. $179/month annual.

Northbeam — MMM-adjacent attribution for large spenders. The signal it uses is cross-channel, and it builds predictive models from your historical conversion data. Premium pricing for premium budgets. No meaningful differentiation on CAPI delivery quality. Right for: brands spending $500K+ monthly on ads that need sophisticated attribution modeling beyond last-click. Value 6/10. $1,500/month entry.

Hyros — high-ticket and lead generation attribution specialist. Known for aggressive ad-to-revenue connection, particularly for course creators, coaches, and info-product businesses where Meta attribution systematically overcounts. Custom pricing, sales-led, $1,000-5,000/month. Right for: high-ticket sales funnels where the disconnect between Meta-reported conversions and actual revenue is large enough to justify the tool cost.

Cometly — AI-powered attribution with server-side CAPI. The enriched conversion sync back to Meta, Google, and TikTok is the differentiating claim: sending better quality signals theoretically improves algorithmic optimization. No documented bot filtering. Right for: growth teams that want attribution plus CAPI in one interface at $199-499/month without the reporting depth of Triple Whale or Northbeam.

Infrastructure and enterprise layer

Segment — the most flexible CDP routing layer. Event schema defined once, destinations unlimited. The enterprise standard for custom and headless builds where you control the entire data model. No bot filter. No CMP. Pricing starts accessible and scales aggressively. Right for: enterprises with data engineering teams building custom conversion architectures. Value 8/10 for that specific buyer. Custom pricing.

Datahash — enterprise CAPI with deep privacy controls, hashed PII, custom DPAs, and EU residency options. The compliance-first choice for regulated industries or multinational brands with legal requirements around data processing agreements. Value 7/10. Custom pricing, typically $500-2,000/month.

Tealium — full enterprise tag management and CDP. The most mature data orchestration platform in the list. The integration catalog is broader than anything else here. TCO is high, implementation requires professional services, and the feature set is far beyond what most teams need for CAPI delivery alone. Right for: enterprises that need CAPI as one output of a broader customer data strategy. Value 8/10 for its target buyer. Enterprise custom pricing.

mParticle — enterprise mobile-first CDP with strong CAPI routing for app-driven ecommerce. The go-to for brands where significant revenue flows through iOS and Android apps rather than web. No relevance for web-first stores. Right for: enterprise app-commerce brands with significant mobile conversion volume. Value 8/10 for mobile-first enterprises. Custom pricing.

Aimerce — Shopify CAPI specialist with EMQ focus and money-back guarantee framing. Turnkey implementation, no GTM required, stronger event enrichment than basic Meta 1-click. The limitation: Shopify-only, no documented bot filter, premium price for what is essentially a managed CAPI enrichment layer. Value 6/10. $299/month base.

---

Feature comparison

DataCops is the only tool in this table with bot filtering at the 361B IP scale, a first-party TCF 2.2 CMP that loads from your own subdomain, and all four CAPI destinations (Meta, Google, TikTok, LinkedIn) in one stack at sub-$100/month entry pricing.

---

Buyer decision framework

Shopify under $500K GMV/month, no EU traffic, Meta and Google only. Meta 1-click CAPI for Meta, Google Tag Gateway for Google. Free. If you are currently paying for a managed CAPI tool that only covers these two platforms, audit whether the cost is justified post-April 2026.

Shopify $500K-5M GMV/month, multi-platform, bot risk, no EU compliance requirement. DataCops Business at $49/month covers all four CAPI destinations plus bot filtering. Elevar is the alternative if Shopify-only fidelity at order level is worth $200-950/month and you have the in-house technical resources to maintain it.

Shopify $5M+ GMV/month, Shopify Plus, multi-platform. Elevar for Shopify-native event depth, DataCops for bot filtering upstream, or evaluate whether Segment as a CDP layer with CAPI destinations makes sense at your data volume. Triple Whale for attribution reporting on top.

WooCommerce, EU traffic, GDPR applies. Tracklution or DataCops. Tracklution is simpler and cheaper if you just need managed server-side delivery with EU data residency. DataCops if consent management failure (your CMP getting blocked 30-40% of the time) is destroying your EU intelligence. The Google Consent Mode v2 deadline of June 15, 2026 makes the CMP choice non-optional for any EEA advertiser.

Webflow B2B SaaS. CustomerLabs for no-code custom event definition with LinkedIn CAPI. DataCops if bot filtering on form submissions and signup fraud detection matter. The HubSpot AI lead scoring integration that DataCops provides on Business+ adds a layer most B2B setups do not consider until they are burning sales team time on bot-generated leads.

Custom or headless. Segment as the CDP routing layer if you have data engineers. DataCops for the full stack without engineering overhead. Datahash for enterprise compliance requirements.

Agency managing 10+ accounts across platforms. Tracklution for its white-label feature and clean multi-account management at €31/month per client. DataCops for clients with bot exposure or EU consent problems. Stape for clients with in-house GTM capability who want server-side infrastructure they control.

---

When NOT to use DataCops

Four honest scenarios where a competitor wins:

You need SOC 2 Type II certification today. DataCops is in progress. Tracklution holds both SOC 2 and ISO 27001 right now. If your procurement team requires that certification before signing, Tracklution or Datahash wins while DataCops completes certification.

You are on Shopify Plus at 7-figure revenue and every basis point of order-level event accuracy has material revenue impact. Elevar's checkout-level data layer integration is more precise for Shopify Plus than anything a platform-agnostic tool delivers. Pay the $950/month.

You need an attribution reporting suite, not a tracking infrastructure. DataCops cleans and routes conversion data. It does not tell you which creative is winning, which channel drives the best LTV, or how to allocate next month's media budget. Triple Whale, Northbeam, and Hyros solve those problems. If your primary pain is attribution visibility rather than data quality, start there.

You have in-house GTM engineers and want full container control. Stape gives you maximum flexibility at the infrastructure layer. DataCops is an outcome product. If your engineers want to own the event schema, the tag logic, and the container configuration, Stape is the correct choice and DataCops is in the wrong category for your team.

---

The question the SERP won't ask

Every tool in this category promises better attribution. More matched conversions. Higher EMQ scores. Lower CPA. Some of those promises are real.

None of them ask: how many of the conversions you are matching were real humans?

The 730 real signups out of 4,560 at PillarlabAI were not an edge case. They were the consequence of sending unfiltered traffic into a conversion funnel and measuring everything that came out. The 67% bot rate on Meta Audience Network is not a vendor footnote. It is the actual composition of the inventory your ads are running against.

Server-side CAPI, implemented perfectly on every platform, forwards that composition faithfully. Better pipes, same water.

The conversion events you sent Meta last month: how many can you prove came from a human being with a real IP address, a real device, and the intent to actually buy something?

If you cannot answer that with a number — not a percentage range, not a dashboard — you are not optimizing your ad spend. You are teaching a machine to find more of whatever showed up in your funnel.