Everything HonestTag does, and the math behind it
HonestTag publishes how every metric is computed, in plain English, with thresholds in numbers and order proof on every attribution. This page is the full tour. Every term links to a guide.
Short on time? Start with the glossary or the learn hub. Ready to try it? Join the early-access waitlist.
The measurement core
A consent-aware first-party pixel captures ad clicks; every Shopify order is matched to the first click that brought the customer in and classified new or returning against your store's order history. Every purchase HonestTag verifies from your web orders goes to your configured destinations among Google Ads, Meta, Klaviyo, Microsoft Advertising, TikTok, Google Analytics 4, Pinterest, Snapchat, Reddit and OpenAI Ads, within your plan's simultaneous-delivery limit, new and returning customers both, each marked new or returning. Refunded Google Ads and Microsoft Advertising orders trigger the correction those platforms make available; Klaviyo and Google Analytics 4 receive a refund event; Meta, TikTok, Pinterest, Snapchat, Reddit and OpenAI Ads refunds are recorded in order proof.
The free Mirror shows you the result of that measurement read-only: what Google Ads, Meta, Microsoft Advertising, TikTok and Criteo claim next to what your store recorded, and the gap. Delivery to Google Ads, Meta, Klaviyo, Microsoft Advertising, TikTok, Google Analytics 4, Pinterest, Snapchat, Reddit and OpenAI Ads, the per-order proof records, and full reporting are what a paid plan and its free trial turn on. Details on the pricing page.
-
Order proof for every number
Each attribution stores a proof record: click id, timestamps, match method, new-or-returning classification, and any refunds and adjustments. Open any order and see exactly why it was counted.
-
The Gap
What Google Ads, Meta, Microsoft Advertising, TikTok and Criteo claimed, next to what your store recorded, channel by channel. The disagreement is not a bug in HonestTag; it is the point of it.
-
The honest scorecard
Spend, new customers, NCAC, NMER, MER, and AOV from your store's back end, with confidence badges when volume is too thin to trust a ratio.
Verdicts: Scale, Hold, or Cut, with receipts
Every Monday, each ad channel gets a verdict (Scale, Hold, or Cut) computed by a fixed, published rule from your own numbers and your own NCAC ceiling. Every verdict prints the exact figures that produced it and the date to re-judge it.
And when there is not enough evidence, HonestTag says "No verdict" and names the floor that failed and what clears it, instead of inventing a verdict on thin data. The evidence floors are published in numbers, the rule order is public, and the verdict history is stored so the track record is auditable. Verdicts are reads of your own data, not advice: the decision stays yours.
Campaign-level truth
Google Ads, Meta, Microsoft Advertising and TikTok claims vs your store's first-click record, campaign by campaign: claimed conversions and value next to new-customer orders and revenue, campaign first-click NCAC, and a weekly verdict chip per campaign.
Unmatched spend and unmatched orders are shown, never hidden, and an account-level adjustments row makes the campaign table add up to your channel topline by construction. A match-health line tells you exactly what percent of spend matched and which tracking template to paste to fix the rest. How the campaign join works, and why honest tools render its holes, is documented in plain English. Campaign figures update nightly.
Attribution models: the comparison, never the switch
See what last-click would have claimed, side by side with the first-click number. The model comparison shows first-click new-customer, last-click, and any-click participation per platform, with a delta column that quantifies exactly what choosing a model does to a number.
There is no model toggle anywhere in HonestTag. Every reported metric uses one stated model, first-click new-customer, and the comparison exists so you can see the theater other tools sell, quantified on your own orders. Delivery is separate: every verified purchase is sent, new and returning customers both.
Cohort payback and LTV by channel
Cohorts track what customers acquired each month go on to spend, and when (if ever) they repay what you paid to acquire them. The new channel cut splits every cohort by the acquiring first-click channel, so you can see which channel's customers actually come back.
Every channel cell honors a 30-customer honesty floor: below it you get unlock copy, not a number, because a seven-customer LTV curve is noise wearing a chart. Members acquired before channel tracking began sit in an explicit "(before channel tracking)" bucket. History is never guessed.
LTV projection: from your history, not a fitted curve
Projections extend a young cohort's curve by chaining the median month-over-month path of your own older cohorts. No curve fitting. The formula is published, and every projected figure traces to cohorts your store actually produced.
Projections are clearly marked, carry confidence badges, never extend beyond what your deepest cohort has actually exhibited, and are never blended into actuals: observed and projected always render as two separate figures.
Tracking links on your own domain
Branded short links that live on your store's own domain and redirect to any page on your store with your UTMs applied. Built for the placements ads cannot tag: influencers, podcasts, email signatures, link-in-bio.
Every click is counted, and because the link lands on your own first-party rails, a link click can become the recorded first click. Influencer traffic shows up in the same attribution and order proof as your ad clicks. Redirects use a 302 on purpose, so clicks keep counting and edited targets keep working. Definitions in the glossary.
Click-to-order lag: judge spend on the right day
How long do your buyers actually take from first click to order, per channel? HonestTag records it in honest buckets, same day through 31+ days, so you stop judging Tuesday's spend on Friday.
That lag makes verdict re-check dates measured instead of asserted: once a channel has enough measured orders, its verdicts are re-judged on a schedule your own conversion behavior justifies. A per-channel lag chart is not in the app yet.
New vs returning purchase signals
HonestTag sends Google Ads the standard purchase conversion plus the applicable classified actions: new customer, returning customer, and first-click new customer. Meta receives classified custom events; its standard Purchase is sent only when the merchant confirms the other Meta purchase source is off or compatible. TikTok receives classified custom events through TikTok Events API. TikTok standard CompletePayment is off by default, so HonestTag sends no second purchase event beside another TikTok pixel; it turns on only when the merchant chooses HonestTag as the purchase source.
If you set a campaign's optimization goal to the new-customer event, the feed Google Ads optimizes against contains only purchases classified as new customers, not repeat orders from customers you already own. Which event you optimize on is your decision. TikTok's HonestTag custom events are reporting and audience signals, not a claim that they drive Smart Performance bidding. The mechanics are in new vs returning signals.
Email's real role: coming next
Email and SMS clicks belong in order proof, not in a blended "email revenue" number. The email/SMS touch ingestion is built; we're holding it until the privacy wording is signed off, because a feature that touches shopper identity ships when the policy does, not before.
The honesty rails are structural: an email click is labeled an assist and never steals first-click credit from the ad that acquired the customer, and HonestTag reads click-engagement events only, never your lists, segments, or profile properties. Touch records keep message and campaign ids plus timestamps, expire in 90 days, and never contain an email address. Details in the privacy policy.
When the data cannot support a figure, we say so.
Every threshold above (the evidence floors, the 30-customer cohort floor, the projection floors) is published in numbers and enforced in code. See what each tier includes or get early access.