pepperwind / docs /product.md
rasinmuhammed's picture
Pepperwind Trading Co. preview, version 1.0
27ec678 verified
|
Raw History Blame Contribute Delete
23.5 kB

product

Product analytics for the storefront website and the iOS and Android apps, shaped like a Segment warehouse export: sessions, the page, screen and track calls, identify calls, anonymous ids tied to customers, campaign touches and order attribution, the experimentation tool's tests and assignments, the feature flag list, and the analytics pipeline's daily rollup. It sees only what the tracking sent: visitors who block tracking or decline consent are missing, so it undercounts commerce. Every time is in UTC. Customers are commerce's ids (cus_...) and orders are commerce orders.

Table Rows What it holds
anonymous_alias 2,580 One row per anonymous id the shop has tied to a customer: when it was first seen and when it was linked (a sign-in, an account made at checkout, or a session that opened signed in). A customer usually has several ids: other devices, the app and the web, cleared cookies.
attribution 4,224 The analytics tool's credit for each order over the customer's campaign touches in the 30 days before it, under two models. last_click gives the last touch a credit of 1; linear gives each of n touches 1/n, rounded to four places with the remainder on the last. Each order's credits sum to exactly 1 per model; an order with no touch is direct with credit 1. A nightly job credits the previous day's orders, so orders of the last day in the data have none.
daily_metrics 12,648 The analytics pipeline's daily rollup: one row per market, local day and metric (sessions, users, orders, revenue). A day is the market's local day (India's, Amsterdam's, New York's). The job recomputes the last three days each morning, so events arriving later than that are not counted. Europe starts on 1 February 2024, when its store opened.
event 333,783 One row per page call (a web page view), screen call (an app screen) or track call (a named action such as Product Viewed or Order Completed), with Segment's fields, as the warehouse loader delivers them.
experiment 14 One row per A/B test, as the experimentation tool lists it. A visitor is assigned at their first page view while a test runs, or at Checkout Started for a checkout test.
experiment_assignment 5,889 One row per visitor assigned to a test, as the flag service logged it: the visitor's first exposure while the test ran. The client then sends an Experiment Viewed call to event.
feature_flag 48 One row per feature flag in the flag service: release flags, test flags (exp- and the test's key), operational kill switches (ops-) and market switches, with their rollout at the end of the data.
identify 1,183 One row per Segment identify call, which ties a visitor's anonymous id to a customer with their traits: when the account is made at checkout, and when the customer signs in on a device that does not yet know them.
session 58,337 One row per session on the storefront website or the app, as the warehouse's sessions model builds it from events: a web session ends after 30 minutes without activity, an app session after 5 minutes in the background. Only sessions the tracking saw are here.
utm_touch 3,360 One row per session of a known customer that arrived with campaign parameters, or from a social site without them, as the warehouse reads it. These are the touches attribution credits.

anonymous_alias

One row per anonymous id the shop has tied to a customer: when it was first seen and when it was linked (a sign-in, an account made at checkout, or a session that opened signed in). A customer usually has several ids: other devices, the app and the web, cleared cookies.

2,580 rows; key alias_id.

Column Type Empty Range or values Meaning
alias_id integer, key 0.0% 1 to 2580 The alias row's key.
anonymous_id text 0.0% 2eab287b-9038-4b39-9767-b9ed0446cac3, a28c2cef-8cf0-43c6-9185-76b2b99ab4ec, 71514270-2966-41d1-ae63-643bda188970 The device or browser's anonymous id.
user_id text 0.0% cus_Nv3DeFWiyRTLlA, cus_ENIcR9NSxcGpCo, cus_0AUFe33DuAzLdY The commerce customer the id is tied to.
first_seen_at timestamp, UTC instant 0.0% 2023-09-16 02:10:21 to 2026-09-30 20:16:55 When the anonymous id was first seen.
aliased_at timestamp, UTC instant 0.0% 2023-10-01 08:21:18 to 2026-09-30 20:16:55 When the id was tied to the customer.

attribution

The analytics tool's credit for each order over the customer's campaign touches in the 30 days before it, under two models. last_click gives the last touch a credit of 1; linear gives each of n touches 1/n, rounded to four places with the remainder on the last. Each order's credits sum to exactly 1 per model; an order with no touch is direct with credit 1. A nightly job credits the previous day's orders, so orders of the last day in the data have none.

4,224 rows; key attribution_id.

Column Type Empty Range or values Meaning
attribution_id integer, key 0.0% 1 to 4224 The credit row's key.
order_id integer, → commerce.order.order_id 0.0% 1 to 11247 The commerce order credited.
model text 0.0% linear, last_click The attribution model: last_click or linear.
channel text 0.0% paid_social, direct, paid_shopping The touch's marketing channel, as the UTM map reads its source and medium (for example paid_search, paid_social, email, messaging, organic_social); direct when the order had no touch.
campaign_code text 43.0% CMP-2023-011, CMP-2022-013, CMP-2022-012 The marketing campaign the touch's campaign parameter names, in its canonical form (CMP--). Empty when the touch carried no recognisable campaign code.
credit number, 4 places 0.0% 0.0588 to 1.0 The share of the order credited to this touch, 0 to 1.
computed_at timestamp, UTC instant 0.0% 2023-10-02 02:30:21 to 2026-09-30 02:32:59 When the nightly attribution job credited the order, about 02:30 the day after it was placed.

daily_metrics

The analytics pipeline's daily rollup: one row per market, local day and metric (sessions, users, orders, revenue). A day is the market's local day (India's, Amsterdam's, New York's). The job recomputes the last three days each morning, so events arriving later than that are not counted. Europe starts on 1 February 2024, when its store opened.

12,648 rows; key metric_row_id.

Column Type Empty Range or values Meaning
metric_row_id integer, key 0.0% 1 to 12648 The row's key.
date timestamp, UTC instant 0.0% 2023-10-01 00:00:00 to 2026-09-29 00:00:00 The market's local day the row counts.
market_code text 0.0% IN, US, EU The market: IN, EU or US.
metric text 0.0% sessions, users, orders sessions (sessions with an event that day), users (distinct visitors with an event that day), orders (Order Completed events) or revenue (the revenue of those events).
value number, 2 places 0.0% -0.0 to 15520.0 The metric's value. Revenue is in the market's currency (INR, EUR, USD) in major units and excludes shipping and tax.
computed_at timestamp, UTC instant 0.0% 2023-10-04 06:10:16 to 2026-09-30 06:24:46 When the job last computed the row: between 06:00 and 06:25 three days after the day, or on the last day of the data for the final days.

event

One row per page call (a web page view), screen call (an app screen) or track call (a named action such as Product Viewed or Order Completed), with Segment's fields, as the warehouse loader delivers them.

333,783 rows; key event_row_id.

Column Type Empty Range or values Meaning
event_row_id integer, key 0.0% 1 to 333783 The row's key, in timestamp order.
message_id text 0.0% ajs-next-ad9ac20cd7023970d79a4d93404e2314, ajs-next-af40bd5791e8ee6ec276c3b333f4ae53, ajs-next-629921d96694baae9339b689dbdb22cf Segment's message id for the call: ajs-next- and a hash on the web (ajs-next-, epoch milliseconds and a uuid from June 2024), an uppercase UUID on iOS, a lowercase uuid on Android.
session_id integer, → product.session.session_id 0.0% 1 to 58337 The session the call belongs to.
anonymous_id text 0.0% 9e70e38d-16f9-4340-ad74-9b1d569024bc, 5fdae6c0-a5dd-44fc-bc57-7d41d306e0db, 29FA9927-C965-4388-8BF6-03C76D6813EF The browser's or app install's anonymous id.
user_id text 61.8% cus_BbYHTEPSr1B6cP, cus_GCAJzH0Ybl1Y5M, cus_1XudikaNGJg7p0 The commerce customer id once the visitor is signed in; empty before.
type text 0.0% page, track, screen page (web), screen (app) or track (a named action).
event text 0.0% Product, Product Viewed, Club The track call's name (for example Product Viewed, Product Added, Checkout Started, Order Completed, Experiment Viewed, Application Opened), or the name of the page or screen viewed (Home, Collection, Product, Cart, Checkout).
timestamp timestamp, UTC instant 0.0% 2023-10-01 02:55:35 to 2026-09-30 21:14:59 When the call happened, corrected for the device clock's error.
properties text 0.0% {"path":"/products/munduchilli","referrer":"https://www.querent.example/","search":"","title":"Mundu Chilli | Pepperwind, {"url":"https://us.pepperwind.example/products/munduchilli","name":"Mundu Chilli","product_id":224,"sku":"SKU-SCHL-10019, {"path":"/pages/club","referrer":"https://us.pepperwind.example/products/munduchilli","search":"","title":"Pepperwind"," The call's properties as a JSON object: a page's path, url, title and referrer; a product's id, sku, category, price and currency; an order's id, subtotal, total, revenue, shipping, tax, discount and currency; a test's id and variation; a sign-in's username and whether it succeeded. A customer whose deletion the privacy desk completed has no username here.
sent_at timestamp, UTC instant 0.0% 2023-10-01 02:55:36 to 2026-09-30 21:33:39 When the device sent the call, by the device's own clock, which can be off.
received_at timestamp, UTC instant 0.0% 2023-10-01 02:55:35 to 2026-09-30 21:33:42 When Segment's servers received the call. App calls are sent in batches, so they can arrive well after they happened.
context_ip text 0.0% 2001:db8:56af:11b2:01cb:f548:330c:bf5d, 198.18.105.240, 2001:db8:064c:f0a9:dcd2:a075:79b5:b426 The IP address the call came from. Kept after a customer's deletion, as the security logs it feeds are.
context_user_agent text 37.1% Mozilla/5.0 (iPhone; CPU iPhone OS 17_-1 like Mac OS X) AppleWebKit/605.1.15 (KHTML, like Gecko) Version/17.-1 Mobile/15, Mozilla/5.0 (Linux; Android 10; K) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/118.0.0.0 Mobile Safari/537.36, Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/118.0.0.0 Safari/537.36 The browser's user agent string; empty on app calls.
context_locale text 2.0% en-US, en-IN, hi-IN The device's language and region setting, for example en-IN; empty where it was not sent.
context_timezone text 1.6% America/Los_Angeles, Asia/Kolkata, America/Chicago The device's time zone; empty where the browser withheld it.
app_version text 62.9% 3.19.0, 3.20.0, 3.20.1 The app's version on app calls; empty on web calls.
plan_tier text 97.0% reserve, spice, leaf The signed-in visitor's Pepperwind Club tier (leaf, spice or reserve), as the call sends it under the property name plan_tier. Empty when the call carries no tier.
tier text 99.0% leaf, reserve, spice The signed-in visitor's Pepperwind Club tier (leaf, spice or reserve), as the call sends it under the property name tier. Empty when the call carries no tier.

experiment

One row per A/B test, as the experimentation tool lists it. A visitor is assigned at their first page view while a test runs, or at Checkout Started for a checkout test.

14 rows; key experiment_id.

Column Type Empty Range or values Meaning
experiment_id integer, key 0.0% 1 to 14 The test's key.
key text 0.0% pdp-sticky-add-to-cart, home-hero-everyday-chai, eu-free-shipping-banner The test's key in the tool, kebab-case.
name text 0.0% Sticky add-to-cart bar on product pages, Everyday chai hero on the home page, Free-shipping threshold banner The test's name.
hypothesis text 7.1% If the add-to-cart button stays on screen while shoppers scroll a product page, more of them will add to cart, because o, Leading the home page with everyday chai blends instead of the seasonal gift range will raise revenue per visitor, since, Showing the free-shipping threshold in a banner on every page will lift average order value in Europe, as shoppers add a The hypothesis the test's owner wrote before it started. Empty where none was written.
primary_metric text 0.0% add_to_cart_rate, revenue_per_visitor, aov The metric the test is judged on: checkout_conversion, add_to_cart_rate, revenue_per_visitor, aov (average order value), signup_rate or app_d7_retention.
allocation text 0.0% {"control":50,"treatment":50}, {"control":33.34,"variant_a":33.33,"variant_b":33.33}, {"control":25,"variant_a":25,"variant_b":25,"variant_c":25} The planned split of visitors between the test's arms, in percent, as JSON.
started_at timestamp, UTC instant 0.0% 2023-10-09 06:09:02 to 2026-08-03 04:33:34 When the test started.
ended_at timestamp, UTC instant 0.0% 2023-11-06 05:19:00 to 2026-08-31 08:50:33 When the test ended.
owner_employee_no text 0.0% EMP-10025, EMP-10034, EMP-10277 The employee number of the product manager who owns the test.

experiment_assignment

One row per visitor assigned to a test, as the flag service logged it: the visitor's first exposure while the test ran. The client then sends an Experiment Viewed call to event.

5,889 rows; key assignment_id.

Column Type Empty Range or values Meaning
assignment_id integer, key 0.0% 1 to 5889 The assignment's key.
experiment_id integer, → product.experiment.experiment_id 0.0% 1 to 14 The test.
anonymous_id text 0.0% 0697f486-c2c7-401b-9c5c-ead9201a0e80, 95f6a425-61b8-4c31-8ce1-a02917bc4663, 777de0a9-c077-421f-a4d9-843e267aa7bd The visitor's anonymous id; assignment sticks to it across sessions.
user_id text 89.2% cus_HzaRNWoFUYFvkg, cus_KQ2qXl4TxNU4Wm, cus_84ZsMECvQpa8o2 The commerce customer id when the visitor was signed in at assignment; empty otherwise.
variant text 0.0% treatment, control, variant_b The arm assigned: control, treatment, or variant_a to variant_c.
assigned_at timestamp, UTC instant 0.0% 2023-10-09 06:21:46 to 2026-08-31 06:07:28 When the visitor was assigned.
browser text 7.1% Chrome, Mobile Safari, Edge The session's browser, or the social app whose in-app browser opened it; empty on the app.

feature_flag

One row per feature flag in the flag service: release flags, test flags (exp- and the test's key), operational kill switches (ops-) and market switches, with their rollout at the end of the data.

48 rows; key flag_id.

Column Type Empty Range or values Meaning
flag_id integer, key 0.0% 1 to 48 The flag's key.
flag_key text 0.0% checkout-guest-checkout, exp-pdp-sticky-add-to-cart, ops-disable-cod The flag's key in the service, kebab-case with a team prefix.
description text 0.0% Guest checkout without an account, Sticky add-to-cart bar on product pages, Turn off cash on delivery What the flag controls.
rollout_percent integer 0.0% 0 to 100 The share of traffic the flag is on for at the end of the data, in percent.
created_at timestamp, UTC instant 0.0% 2023-10-04 12:03:49 to 2026-09-15 09:48:05 When the flag was created.
owner_employee_no text 0.0% EMP-10047, EMP-10025, EMP-10051 The employee number of the engineer or product manager who owns the flag.
archived_at timestamp, UTC instant 45.8% 2024-01-26 15:56:40 to 2026-08-29 14:11:26 When the flag was archived. Empty for a flag still live; kill switches are never archived.

identify

One row per Segment identify call, which ties a visitor's anonymous id to a customer with their traits: when the account is made at checkout, and when the customer signs in on a device that does not yet know them.

1,183 rows; key identify_id.

Column Type Empty Range or values Meaning
identify_id integer, key 0.0% 1 to 1183 The identify call's key.
message_id text 0.0% ajs-next-5b54f20d3971af3928b7feb31191f9ed, ajs-next-87d85854e814317339d3f5fabac951df, ajs-next-971e5b4c238dc64aea3234a0536f0856 Segment's message id for the call, in the same shapes as event.message_id.
user_id text 0.0% cus_1XudikaNGJg7p0, cus_8qHESP9k0XUUub, cus_KQ2qXl4TxNU4Wm The commerce customer identified.
anonymous_id text 0.0% 30c9d093-ff86-48bc-bffa-47993467f299, 884ceb0b-2969-4c18-88c4-8d5a335c7abf, e18b82af-c0ce-43c7-b7e1-51e9252f3bbc The anonymous id the call ties to the customer.
timestamp timestamp, UTC instant 0.0% 2023-10-03 05:07:36 to 2026-09-30 04:59:03 When the call was made.
traits text 0.0% {"email":"hema.murugesan@mail.example","first_name":"Hema","last_name":"Murugesan","created_at":"2022-09-17T03:17:46","m, {"email":"patrick.carpenter@usmail.example","first_name":"Patrick","last_name":"Carpenter","created_at":"2023-04-05T03:4, {"email":"malavika.mohan@inmail.example","first_name":"Malavika","last_name":"Mohan","created_at":"2023-10-05T09:06:44", The customer's traits as JSON: email, first_name, last_name, created_at (when the account was made) and market; only created_at and market for a customer whose deletion the privacy desk completed.

session

One row per session on the storefront website or the app, as the warehouse's sessions model builds it from events: a web session ends after 30 minutes without activity, an app session after 5 minutes in the background. Only sessions the tracking saw are here.

58,337 rows; key session_id.

Column Type Empty Range or values Meaning
session_id integer, key 0.0% 1 to 58337 The session's key, in order of the sessions' start.
anonymous_id text 0.0% 9e70e38d-16f9-4340-ad74-9b1d569024bc, 5fdae6c0-a5dd-44fc-bc57-7d41d306e0db, 29FA9927-C965-4388-8BF6-03C76D6813EF The browser's or app install's anonymous id.
user_id text 70.6% cus_BbYHTEPSr1B6cP, cus_GCAJzH0Ybl1Y5M, cus_1XudikaNGJg7p0 The commerce customer id when any of the session's calls carried one (the visitor was signed in or made an account); empty for an anonymous session.
platform text 0.0% web, ios, android web, ios or android.
app_version text 82.0% 3.19.0, 3.20.0, 3.20.1 The app version on an app session; empty on the web.
browser text 18.0% Mobile Safari, Chrome, Samsung Internet The browser on a web session (for example Chrome, Mobile Safari, Samsung Internet), or the social app whose in-app browser opened it; empty on the app.
os text 0.0% iOS, Android, Windows The device's operating system: Android, iOS, Windows, macOS (an iPad reports macOS) or Linux.
device_id text 82.0% 5C91FE0A-3ED1-40B9-B664-69E4FD79E722, 955e23c400299680, 3eacf737a3ed58eb The app's device identifier on an app session; empty on the web.
started_at timestamp, UTC instant 0.0% 2023-10-01 02:55:35 to 2026-09-30 21:14:09 When the session's first event happened.
ended_at timestamp, UTC instant 0.0% 2023-10-01 02:59:21 to 2026-09-30 21:15:02 When the session ended.
landing_page text 18.0% https://us.pepperwind.example/products/munduchilli, https://in.pepperwind.example/products/malabarearlgrey?utm_source=sb&utm_medium=paid&utm_campaign=cmp_2023_011, https://us.pepperwind.example/pages/our-estates The full URL of the session's first page, with its query string; empty on the app.
referrer text 54.5% https://www.querent.example/, https://www.querent.example/?hl=en-IN, https://m.palaver.example/ The referring site, as the browser sends it (the origin only); empty when there was none and on the app.
utm_source text 51.2% sb, push, querent The utm_source parameter as it arrived, spelled as people and tools typed it; empty for an untagged visit.
utm_medium text 51.2% paid, push, shopping The utm_medium parameter as it arrived; empty for an untagged visit.
utm_campaign text 54.2% cmp_2023_011, newsletter-23, shopping_2023 The utm_campaign parameter as it arrived: a campaign code in one of several spellings or a hand-written name; empty when absent.
country text 0.0% US, IN, FR The shopper's country, as a two-letter code.
is_bounce boolean 0.0% False to True True for a bounce: one page or screen, under 10 seconds, no conversion.

utm_touch

One row per session of a known customer that arrived with campaign parameters, or from a social site without them, as the warehouse reads it. These are the touches attribution credits.

3,360 rows; key touch_id.

Column Type Empty Range or values Meaning
touch_id integer, key 0.0% 1 to 3360 The touch's key.
session_id integer, → product.session.session_id 0.0% 8 to 58328 The session that is the touch.
customer_id text, → commerce.customer.customer_id 0.0% cus_BbYHTEPSr1B6cP, cus_84ZsMECvQpa8o2, cus_1XudikaNGJg7p0 The commerce customer.
occurred_at timestamp, UTC instant 0.0% 2023-10-01 08:21:18 to 2026-09-30 19:51:14 When the session started.
utm_source text 0.0% m.palaver.example, l.shutterbloom.example, querent The utm_source as it arrived, or the social site's host for an untagged social visit.
utm_medium text 0.0% cpc, referral, shopping The utm_medium as it arrived, or referral for an untagged social visit.
utm_campaign text 8.8% CMP-2023-011, CMP-2022-013, cmp_2022_012 The utm_campaign as it arrived; empty when absent.
landing_url text 0.0% https://in.pepperwind.example/collections/groundcoffee?utm_source=m.palaver.example&utm_medium=cpc&utm_campaign=CMP-2023, https://in.pepperwind.example/, https://us.pepperwind.example/?utm_source=querent&utm_medium=shopping&utm_campaign=CMP-2022-013&qclid=Cj0K6c10e621fbbf5f The full URL the session landed on, with its parameters.