AdPix features: what each tool measures and which plan includes it

AdPix features cover two jobs. The first is the reporting you know from Google Analytics 4: realtime visitors, where traffic comes from, what people do, the conversion funnel and e-commerce revenue. The second is a marketing measurement layer that GA4 does not ship: marketing mix modelling, campaign lift tests, invalid-traffic detection, built-in consent management and an AI assistant that writes verified briefs. The Free plan covers most of the first layer; the measurement layer starts on Starter. This page walks through each feature, what question it answers, how it works and which plan has it.

AdPix is a web analytics platform that replaces Google Analytics 4: realtime, acquisition, engagement, conversion and e-commerce reports, plus attribution, lift testing and marketing mix modelling, with no data sampling and a free plan.

What AdPix features are there, and what question does each answer?

Start from the decision you need to make, then pick the feature. The table maps each AdPix feature to the question it answers and the first plan that includes it, using the plan placement on the AdPix pricing page.

Two design choices sit under every report. AdPix does not sample data in reports. And channel is computed when you open a report rather than frozen onto stored events, so when you correct a channel rule your whole history is reclassified at once, with no reprocessing job (AdPix docs). If you are still deciding whether to move off GA4 at all, the Google Analytics alternative guide compares AdPix, GA4 and self-hosted Matomo criterion by criterion.

Which AdPix features are included in each plan?

Each AdPix plan covers one production website, and each plan includes everything in the plan before it. The plan belongs to the website, not to your account, so one person can see a feature on one site and an upgrade screen on another (AdPix docs: plans and entitlements).

AdPix 360 is about scale, governance and support. It is not where the measurement features live; a Starter website already has MMM and lift. A website’s first paid plan starts with a 14-day free trial, and an upgrade opens features immediately on the events you have already collected.

How does AdPix show where your visitors come from?

AdPix reads source, medium and campaign from three places in a fixed order: an ad-network click ID beats utm_* parameters, and utm_* beats the referrer. If none is present, the visit is direct. Channel is then derived from those values by an ordered rule set where the first matching rule wins (AdPix docs: acquisition overview). This is why consistent tags matter; the UTM parameters guide gives a naming convention that holds up when five people tag links.

Two acquisition reports answer two different questions, and mixing them up is the most common reason numbers seem to disagree:

• Acquisition overview credits each user to their first touch, permanently. A customer who first arrived from a Google ad six months ago and came back today from Instagram is still a Google user here. Use it for “which channels put people into my funnel?”
• Traffic acquisition is session-scoped. Use it for “what drove traffic this month?”

For Iranian media, AdPix ships a registry of the local market. Yektanet, Tapsell, Sabavision, Clickyab and Adivery are recognised as ad networks; Torob, Digikala and Divar as marketplaces; Telegram, Eitaa, Bale and Rubika as messengers; and every Shaparak bank gateway as a payment return. Paid traffic from those networks lands in a paid channel under the network’s own name, and a source AdPix does not know can be classified in one click (AdPix docs: traffic networks).

Realtime is the confirmation screen. It shows active users in a rolling 5, 30 or 60-minute window and refreshes itself every five seconds. Use it to check that a new tag or event fires and that a campaign’s traffic is landing; it deliberately has no bounce rate, session duration or period comparison (AdPix docs: realtime).

How do AdPix funnels and e-commerce reports follow the path to a sale?

The AdPix conversion funnel takes 2 to 8 steps that you pick from standard events (page_view, sign_up, add_to_cart, begin_checkout, purchase and others) and your own key events. It shows what share of the people who entered step one reached each later step, and where they dropped (AdPix docs: conversion funnel). Three counting rules decide how to read it:

1. Order is enforced. Someone reaches step three only after steps one and two, in that order.
2. The completion window equals the date range you pick, from one day up to one year. A 90-day range allows 90-day journeys, which matters for products people think about for weeks.
3. People are grouped by identity, not by browser. If a visitor browses anonymously and later signs in, the earlier anonymous events join that person.

Funnels are on every plan, including Free. For method, the sales funnel analysis guide shows how to choose the steps and which drop-off to fix first.

The E-commerce report reads the line items of the orders you send. Revenue is counted once per transaction ID, so a three-item order is not counted three times. Refunds are shown in their own card and are not subtracted from revenue, and money stays in the currency it arrived in, never silently converted. For the purchase event itself, the docs recommend sending it from your server, which an ad blocker or a payment-gateway redirect cannot lose (AdPix docs: e-commerce). Order analysis is on Free; online sales and revenue analysis start on Starter.

A cohort retention grid exists in the console but is marked Coming soon. The docs say its cohorts are derived from the selected range and its days close in UTC, so conclusions from it are not yet safe (AdPix docs: cohort retention). Until it ships, read repeat behaviour through funnels and explorations, and estimate long-run value with the method in the customer lifetime value guide.

How does AdPix prove whether your ads actually worked?

AdPix has two studies for this, and they answer different questions. Lift tells you whether one specific change, such as a budget increase on a date, caused extra conversions. Marketing mix modelling tells you what each channel returns at today’s budget across the whole portfolio and where it saturates (AdPix docs: MMM or Lift). Both are included from the Starter plan. The general case for measuring incrementality instead of trusting attribution is in the incrementality testing guide.

How does a lift test work in AdPix?

You create a test with the action type (budget up or down, campaign start or stop, creative change and others), the date it took effect, the channel and a target metric, and pick a measurement window of 7, 14 or 28 days. AdPix fits a model on the days before the action date, forecasts what the metric would have done with no change, and sums the gap between that counterfactual and what happened (AdPix docs: run a lift test). This is the same idea as the Bayesian structural time-series approach Brodersen and colleagues published in 2015 (arXiv). A setting that is on by default feeds organic and direct sessions in as a control, so a market-wide swing or a campaign that only moved organic buyers onto a paid channel is not credited to your action.

A hypothetical example shows the arithmetic. You raise a channel’s daily budget from $500 to $750 on 1 March and run a 14-day test on orders.

• Counterfactual orders over 14 days: 1,960. Observed orders: 2,240.
• Incremental orders: 2,240 − 1,960 = 280
• Extra spend: $250 × 14 = $3,500
• Incremental cost per order: $3,500 ÷ 280 = $12.50
• At a $40 average order value, incremental revenue is 280 × $40 = $11,200, so incremental ROAS = $11,200 ÷ $3,500 = 3.2

That 3.2 is only meaningful with its interval. If the 95% interval on incremental orders ran from −40 to 600, the verdict would be Neutral and the honest conclusion would be “not proven yet”. The docs suggest a useful sanity check: build a test on a date when nothing changed, and the verdict should come back Neutral. To compare an incremental ROAS with the ordinary platform figure, run both through the calculator below; the ROAS formula guide explains the break-even line.

How does marketing mix modelling work in AdPix?

The AdPix model splits each week’s conversions into a non-paid baseline (organic and direct demand, trend, seasonality) and the contribution of each paid channel. Its inputs are the weekly spend per channel that you import as a CSV on the Cost & ROAS screen, your purchase events, and organic and direct sessions as a control. A seasonality calendar adds Nowruz, Ramadan, Yalda and other events, so a holiday wave is not credited to your ads (AdPix docs: run an MMM). Calibrating a model with experiment results is also what Google recommends for its own open-source MMM, Meridian (Google).

The model is strict about when it speaks. Each run gets a trust tier (Trusted, Directional or Insufficient), and each channel is badged Identified, Uncertain or Not identified. A channel whose weekly spend barely moves cannot be separated from the baseline: the docs set the bar at a coefficient of variation of 0.30. Say a channel spends $1,000 a week on average with a standard deviation of $200; its coefficient of variation is 200 ÷ 1,000 = 0.20, so it gets no ROI until you vary its budget more. With less than two years of data, a channel’s ROI is shown only after a lift test or a measured spend change has calibrated it; otherwise it reads “Needs a lift test”. Pro adds each channel’s share of sales.

For budget changes you never set up as tests, an Incrementality page detects them in your cost data: a change of at least 20% that holds for at least five of the next seven days becomes an activity that is measured like a lift test, and the results build a response curve per channel (AdPix docs: incrementality).

How does AdPix keep your data clean and compliant?

Invalid-traffic detection (Starter and up) separates the traffic you pay for that is not a real, interested visitor. It follows the industry’s two layers. General invalid traffic is what lists catch: known data-center, proxy and Tor ranges, and declared bots. Sophisticated invalid traffic only shows in behaviour, such as hundreds of sessions with near-identical dwell times or one device fingerprint across unrelated businesses (AdPix docs: invalid traffic). The split matches the Media Rating Council’s definitions of GIVT and SIVT (MRC).

AdPix leans towards precision over recall, because flagging a channel that really sells is the more expensive mistake. Per the docs, a source that converts at 2% or more is never auto-flagged, zero conversions alone never cause a flag, a VPN or a shared IP is never enough on its own, and traffic from your own home market is never treated as suspicious origin. For a store in Iran selling to Iranians, Iranian traffic is simply normal traffic.

Cookie consent management (Starter and up) is built into the tracker rather than bolted on. You set four categories (Necessary, Preferences, Statistics, Marketing), write the banner and choose opt-in or opt-out. Under opt-in, the tracker fails closed: until a visitor chooses, nothing non-necessary runs and no page view is sent, so expect a real drop in counted traffic (AdPix docs: consent banner). The docs also note that region-based switching is not implemented yet; if you have European visitors, set opt-in for everyone.

What do the AdPix AI assistant and AI reports do?

An AI report is a short brief for a 7, 30 or 90-day window: a headline, four core metrics against the previous period, top findings and a numbered “what to do next” list. After the language model writes it, AdPix checks every number against the website’s data and removes any finding whose number is not there. The model receives only aggregate numbers; no visitor, email, phone or IP address is sent (AdPix docs: AI insights). Briefs are written in Persian, English or Arabic.

Allowances are 1, 4, 40 and unlimited reports a month on Free, Starter, Pro and AdPix 360. On paid plans you can ask follow-up questions about a brief and export it as a PDF or an editable PowerPoint deck. The docs are frank about the limit: the text is still written by a language model, so open the linked report before an important decision.

How do audiences and custom reports work in AdPix?

An audience is a named group of people defined by rules, such as “added to cart and did not purchase within 3 days”. AdPix keeps its membership up to date and lets you use it in three places: as a comparison on any report, in the overlap view, and as a list pushed to your CRM or any webhook with identified members only (AdPix docs: audiences). The overlap view compares two or three audiences and labels every region with its own count. Audiences, segment overlap and custom reports are included from Starter.

One honest caveat: the separate Segments page is still marked Coming soon in the docs. It saves conditions combined with AND only and does not apply to standard reports yet, so build groups as audiences.

What is AdPix not built for?

Knowing the limits saves you a wasted trial. Per the AdPix docs:

How do you start using AdPix features?

Install the tag, check it, then add features as your questions grow.

1. Sign up for the Free plan and add your website. Copy the snippet from Admin, Data streams, and paste it once, as high in the <head> as you can, on every template. You can also import a GTM container. Setup takes under five minutes.
2. Open Realtime, visit your own site in another tab, and confirm your visit appears within a few refresh cycles.
3. Tag every paid link with UTMs and build your first funnel from landing page to purchase.
4. Keep GA4 running alongside for four to six weeks and compare the same metrics over the same range. AdPix runs next to Google Analytics without interfering.
5. When you are ready to measure ad impact, move the website to Starter, import your weekly ad spend, and run a lift test on your next deliberate budget change. That result is what later unlocks channel ROI in the marketing mix model.

If you track ads against targets, the marketing KPIs guide helps you pick the few numbers worth watching, and the attribution models guide explains why last-click credit and lift results rarely agree.

ROAS calculator

Return on ad spend and your break-even point. The free plan covers 100,000 events a month, no credit card. Setup takes under five minutes.

Start freeTry it free