Where you stand, the three things worth doing about it, then what moved and what your published work earned.
What changed.Was: a Dashboard whose "to-do today" repeated the first three cards of Actions verbatim, plus a separate Action Center doing a third version of the same job. Now Today is the only screen that answers "what happened since I last looked," and it is the only place the three surfaces overlap by design.
Answer share
12%3 of 25 tracked questions name FactionFab
+2 questions since Aug 1Category average 19%
30 days. Each question re-checked every 6 hours.
What changed.Was: "46/100 Visibility score" with no definition, sitting beside "0 of 6 places buyers search where you appear," which contradicted it. One number, defined in the sentence next to it, with the denominator visible. A score you cannot recompute by hand is a score merchants will not believe.
Attributed revenue
$3,430from 9 orders
Up from $840 in July2 assets producing
Orders traced from an AI answer to checkout.
—
The checkout pixel is not reporting, so there is no honest way to tell you what your answers were worth.
Why this sits beside answer share. Winning an answer is the input. Revenue is the only number that renews a subscription, so it belongs in the same glance, not three clicks away. When the pixel is quiet this cell says so rather than showing a smaller, safer-looking fake.
By engine
ChatGPT Live2 of 25
Perplexity Live1 of 25
ClaudeOctoberNot checked
GeminiOctoberNot checked
What changed.Was: nine channels listed (ChatGPT, Gemini, Perplexity, Claude, Google AI, Grok, YouTube, Reddit, TikTok), five of them labeled "Estimated," which the app itself defined as "simulation." Four engines, matching the brand and the site. Two are real on October 1. The other two say "October" and show nothing. Simulated data never renders as a result.
Worth doing today
Ranked by demand, gap size, and whether you hold data nobody else has.
Answer the 2024 WRX fitment question on your coilover product pages
Asked in 3 of 4 engines and rising. No competitor answers it with chassis-specific fitment, and you have fitment data in your product metafields that no crawler can infer.
Product pages6 creditsUses your fitment dataOpen slot, nobody holds it
84High
Rewrite the coilovers collection page
34 products, 41 words of copy. Four tracked questions route buyers to this page, and it gives engines almost nothing to quote.
Category page12 creditsAffects 4 questions
78Medium
Take back the Whiteline sway bar comparison
You lost this slot on August 24. Their page compares diameters. You have spring rates, bar rates, and 41 customer reviews mentioning body roll, and none of it is on a page an engine can read.
Blog post18 creditsRecently lost
What changed.Was: 47 unranked cards, several literal duplicates (two "What material are quilted jackets made from?", two identical comparison pages), each with the same boilerplate reason: "Competitors currently lead the answers for this question." Three items, each with a reason specific to this store, priced in credits, and marked when it draws on data only this merchant has. The other 34 are still in the queue and are not pretending to be urgent.
Assets typically take 3 to 6 weeks to appear in answers. A row with nothing in it yet is normal, not a failure.
Why this is new. There is nowhere in the current app to see what a published asset did. "Recently published (9)" is a collapsed line at the bottom of a 47-item list. This table is the proof Darren keeps asking for, and it is the reason a merchant renews in month three.
Answers
The 25 questions buyers ask about your category, what the engines say back, and why you are or are not in the answer.
What changed.Was: Research, a list of 30 checkbox rows with a "Ran today" stamp and a results table that mixed Live and Estimated rows. Same data, but the unit is now the answer, not the query. Every row can be opened to read the engine's actual words and the pages it cited. That is the evidence the Write queue rests on.
Question
Demand
ChatGPT
Perplexity
Claude
Gemini
Who wins it
Checked
do FactionFab coilovers fit a 2024 WRX
Fitment · rising
40/mo
Absent
Absent
—
—
Nobody. Both engines hedge.
2h ago
best budget coilovers for a daily driven WRX
Category discovery
210/mo
3
Absent
—
—
Bilstein, Fortune Auto
2h ago
FactionFab vs Whiteline sway bars
Comparison · lost Aug 24
90/mo
2
Absent
—
—
Whiteline
2h ago
are FactionFab coilovers any good
Branded
140/mo
1
1
—
—
You do
2h ago
what size sway bar for a 2015 STI
Fitment
70/mo
Absent
Absent
—
—
Subimods, RalliTEK
2h ago
cheapest way to lower a WRX without ruining ride quality
Jobs to be done
180/mo
Absent
5
—
—
Cobb Tuning, Subimods
2h ago
alternatives to FactionFab
Alternatives
30/mo
4
Absent
—
—
Perrin, Grimmspeed
2h ago
1 your position in the answerAbsent checked, not named— not checked yetClaude and Gemini connect in October. Until they return real answers they show a dash, never an estimate.
do FactionFab coilovers fit a 2024 WRX
ChatGPT · checked 2 hours ago
What ChatGPT said
FactionFab does list coilovers for the Subaru WRX, but I would confirm fitment directly with the seller before ordering. The 2024 WRX uses the VB chassis, which changed the front strut mounting compared with the previous VA generation, and not every listing makes clear which chassis it covers. Retailers such as Subimods and RalliTEK publish fitment charts you can check.
2 VA to VB chassis differences explained rallitek.com
3 2024 WRX coilover thread reddit.com/r/WRX
Why you are not in it
All three cited pages state chassis codes and mounting differences in plain text. Your coilover pages list part numbers and a "2015+ WRX" range, which does not answer the question a 2024 owner is asking.
What you have that they do not
MFitment metafields1,240 SKUs mapped to chassis codes
OOrder history318 coilover orders since the VB launch
RReviews41 mention fitment on a 2024 car
This panel is the whole argument. The current app tells a merchant they are absent and then hands them a generic blog post. This says what the engine actually said, which pages beat them, exactly what is missing, and which of their own non-crawlable data closes the gap. It is also the thing a competitor scraping public answers cannot replicate.
Write & publish
Seventeen items, grouped by where they live on your store. Ranked, priced, and drawn from your own data.
What changed.Was: Actions and Action Center, 47 unranked cards with duplicates, filtered by content type only. One surface. Grouped by the three places content lives (product pages, category pages, blog and FAQ), which is how Darren describes the work and how a merchant thinks about their store. Ranked, deduped, and priced against the credits they are paying for.
312 of 400 content credits leftGrowth plan · resets Sep 1
Drafting spends credits. Editing, previewing, and publishing do not.
Everything in this queue
184 credits
Why credits are on this screen. Growth includes 400 credits a month. A queue that never mentions them lets a merchant burn a month of entitlement on low-value drafts, then hit a wall. Showing the price next to the reason is what makes the ranking mean something.
Merged
Eleven near-duplicate suggestions became four. Three separate items all wanted a coilover buyer's guide, and two proposed the same sway bar comparison. One page can only win a question once, so Selix now merges them before you see them.
What changed.Was: the live queue contains two identical "What material are quilted jackets made from?" cards and two identical "quilted blankets, jackets, and hobo bags" comparison pages, presented as separate work. Deduplication happens before the merchant sees the list, and the app says it happened rather than hiding it.
Product pages
3 items · 22 creditsWhere the buying decision happens
91High
2024 WRX fitment block for 34 coilover pages
Answers a question rising in 3 engines that nobody currently answers. Built from your fitment metafields, so no competitor can copy it from a crawl.
6 credits
Published title
2024 WRX (VB chassis) fitment
Yes, with the correct part number. The 2024 WRX uses the VB chassis, which changed the front strut top mount from the 2015 to 2021 VA cars. FactionFab lists a separate front assembly for VB.
Which part number you need
Model year
Chassis
Coilover part
2022 to 2024 WRX
VB
1.10225.1
2015 to 2021 WRX and STI
VA
1.10205.2
What owners report
Across 318 coilover orders placed since the VB launch, 41 reviews mention fitment on a 2024 car. The recurring note is that the front endlink needs to be disconnected before the strut drops out, which adds about 20 minutes per side on a first install.
Ride height and spring rates
The VB kit ships at 8 kg/mm front and 6 kg/mm rear, with 25 to 60 mm of drop. Owners running mostly street miles typically set 30 to 35 mm front.
Publishing is free. Regenerating costs 6 credits.
What changed.Was: an editor containing "Wondering about the best online shopping for everyday use? You're in good company," with no indication of where any of it came from. The right panel is the fix for generic content. It shows which of the merchant's own data went into the draft, flags anything unsourced, and it is the reason this content can win an answer that a competitor's generic page cannot.
72Med
Spring rate comparison block for 12 suspension pages
Three tracked questions ask about ride quality. Your spec sheets have the rates, the pages do not.
8 credits
54Low
Install difficulty note for 22 sway bar pages
Comes up in reviews often. Low search demand, but it reduces returns.
34 products, 41 words of copy. Four tracked questions route here.
12 credits
69Med
Split the suspension collection by chassis
VA and VB buyers ask different questions. One page cannot rank for both.
16 credits
61Med
Buying guidance for the sway bars collection
Diameter, rate, and what each is for. Whiteline has this. You do not.
10 credits
43Low
Copy for the wheels collection
No tracked question routes here yet. Worth doing when credits allow.
8 credits
Blog posts and FAQs
10 items · 116 creditsWhere a question gets a long answer
78Med
FactionFab vs Whiteline sway bars, compared honestly
You held this slot until August 24. Their page has diameters. You have rates, weights, and 41 reviews about body roll.
Recently lost18 credits
66Med
How much drop is too much on a daily driven WRX
180 searches a month. Currently answered by a Reddit thread from 2019.
14 credits
58Med
What spring rate should I run on a street car
Extends the FAQ you published August 8, which is already winning one answer.
10 credits
29Low
7 more low-priority items
Below the line Selix recommends acting on this month.
The line matters. Everything scoring under 40 sits below a visible cutoff instead of padding the count. A queue is only useful if the app is willing to say which parts of it are not worth your credits.
Revenue
What the answers you won were worth. Nothing on this screen is modeled or estimated.
What changed.Was: Insights, which drew a sessions chart labeled "Mock data shown," claimed a "strong relationship (r = 0.64)" against a visibility column reading 0% every day, and printed a weekly digest about ADAS calibration on a quilt store. If the pixel is not reporting, this screen shows setup instructions and no chart at all. A fabricated correlation is worse than an empty state, because a merchant who catches it once stops believing the rest of the product.
!
We cannot show attributed revenue yet
The Selix pixel is not reporting from your checkout. Until it does, there is no honest way to tell you which orders came from an AI answer, so this screen stays empty rather than guessing.
Open Settings, then Customer events in your Shopify admin
Custom pixels live here, not in the app section.
Add a custom pixel named Selix and paste the loader
About 12 lines. We keep it thin so we can update what it does without asking you to paste again.
Set customer privacy to Data sale opt-out, then Save
Selix records the referring answer, not the shopper.
Turn on "Pixel installed" in the left rail to see this screen with data.
Attributed in the last 30 days
$3,430across 9 orders
Up from $840 in July468 sessions from AI answers1.9% conversion
Attributed means the visit arrived from a link an AI assistant showed, and the same browser completed checkout. Orders where an AI answer was one touch among several are counted separately below and never added to this figure.
Pixel
Checkout events Healthy2 min ago
Referrer capture Healthy468 this month
Consent suppressionNormal31 events
Order reconciliation Matching9 of 9
Shopify sends no uninstall notice, so Selix watches the pixel itself. If it goes quiet for 24 hours you get an email.
By published asset
Only assets that produced at least one session appear here.
Asset
Surface
Published
Answers won
Sessions
Orders
Attributed
Assisted
Coilover buyer's guide for daily driven WRXs
Blog post
Aug 12
2
310
6
$2,140
$1,190
What spring rate should I run on a street WRX
FAQ
Aug 8
1
146
3
$1,290
$420
STI sway bar end link FAQ
FAQ
Aug 26
0
12
0
—
—
Six other assets published this month have not produced a session yet. Assets typically take 3 to 6 weeks.
What we cannot tell you
A shopper who reads an AI answer and later types your URL directly leaves no referrer. Those visits are real and we cannot see them, so the figure above is a floor, not a total. We would rather undercount than sell you a number you cannot defend to your own board.
Why this box exists. Attribution is the proof point the whole product rests on, which makes overclaiming here the single most expensive mistake Selix could make. Stating the limit out loud is what makes the rest of the number credible.
Connection
Selix is not a Shopify app, so setup is a real 20 minutes. Here is exactly where you are in it.
What changed.Was: a Settings page offering "Log into Shopify to sync" and a field for an shpat_ Admin API token. Shopify stopped allowing admin-created custom apps on January 1, 2026, so that paste flow no longer exists. The real path is the merchant creating an app in Shopify's Dev Dashboard and handing over a Client ID and Client Secret. It is the top risk in the October 1 plan, so it gets a screen, not a form field.
Why health is a permanent panel. Shopify sends no uninstall webhook under this model, and access tokens expire every 24 hours. Nothing tells Selix when a merchant revokes access or a scope drifts, so the app has to watch its own connection and say so in plain language.
Setup
3 of 5 stages done. About 4 minutes left.
✓
Create an app in Shopify's Dev Dashboard
Six steps, done August 20. This is a different area from Apps and sales channels, which is where most people look first.
✓
Grant the seven scopes Selix needs
Each one does a specific job. Nothing is requested that we do not use.
Selix exchanges these for a fresh access token every 24 hours. The secret is encrypted at rest and you can rotate it here at any time, which cuts our access immediately.
Client ID a4f2…9c11Secret •••••••••••• Next refresh 3:12 PM tomorrow
4
Install the checkout pixel
Three steps, about four minutes. Without it Selix can show you which answers you won, but not what they were worth, which is most of why you are here.
4 min
5
Confirm your first full sync
Unlocks once the pixel reports its first event.
Why this is a screen and not a wizard you exit. Onboarding takes 15 to 30 minutes across two Shopify admin areas, and it is the thing most likely to decide whether there are 10 paying customers by November 30. Merchants abandon partway through. A permanent, resumable checklist that says how many minutes are left, with a "do it with us" escape hatch on the hardest stage, is worth more than any polish elsewhere in the app.
Why this is not one click
Selix is not listed on the Shopify App Store. That means no app review, no revenue share on your subscription, and no third party sitting between us and your data. It also means these twenty minutes. We think it is the right trade, and we will do it with you on a call if you would rather not do it alone.
Voice & catalog
What Selix is allowed to say on your behalf, and which parts of your catalog it draws on.
Why this screen is new. Nothing in the current app lets a merchant control the voice of content published under their own brand, which is the single biggest reason a merchant would stop trusting the product after their first draft. This is also where the generic-copy problem gets fixed at the root rather than per draft.
How you sound
Learned from 34 of your existing pages. Edit anything that is wrong.
DirectTechnical, assumes knowledgeNumbers over adjectivesNo hypeSecond person
Never say
best in classunparalleledgame changingyou're in good company
What Selix may draw on
✓Product specs and metafields1,240 products, 86% with fitment data
✓Customer reviews2,180 reviews, used as quoted patterns only
✓Order and quantity dataUsed for "what people actually buy", never published raw
✕Customer names and emailsOff. Requires Shopify protected data approval.
Order and quantity data is the part no crawler and no competitor can see. It is why a Selix page can answer a question that a well-written blog post cannot.