The problem Evidence Modules How it works Pricing FAQ Get started free →
AI Funnel Management

Your AI visibility scores
look fine.
They're not telling you everything.

If your brand appears in AI responses but you're not sure that translates to sales — it probably doesn't. AIVO Optimize runs the full buying conversation and shows you exactly where your brand exits, which competitor wins instead, and why.

87%
Brands lose the buying recommendation
195+
Brands tested
5
AI platforms
12 months
Research corpus
They measure whether your brand appears. We measure whether it wins.
Both matter.
Only one tells you whether your revenue is going to compound — or erode.
$750B
Consumer purchases influenced
by AI in 2026
87%
Brands lose the final
buying recommendation
$0
Recorded in any analytics
platform when a brand loses
What happens to most brands

Most brands think
they're visible.

They appear in AI responses. Their visibility scores look fine. Their dashboards point up. Then someone runs the full buying conversation — and here's what actually happens.

Step 1
Discovery
Brand appears ✓
A buyer asks "what's the best product for X?" — your brand is mentioned. Confidently. Positively. Your visibility tool logs this as a win.
Step 2
Comparison
Competitors introduced
The buyer asks what else to consider. Your brand survives — but four competitors are now in the conversation alongside you.
Step 3
Evaluation
Brand eliminated ✗
The buyer applies specific criteria: "I want the best for X, Y, Z." The AI applies a decision filter. Your brand fails it. Here is the exact sentence AI uses to dismiss you:
"relies more on plant extracts than highly proven actives — results are typically subtle and gradual"
Step 4
The Sale
Competitor wins your customer
A specific competitor is named. A specific product. Sometimes a specific URL. The buyer leaves to purchase — and it isn't from you. Your visibility tool still says you appeared. It did not record this.

This sequence — appearing, surviving, then being eliminated before the recommendation — is the most common pattern we observe across 195+ brands. A visibility score of 84/100 is entirely consistent with a buying recommendation win rate of zero. The dashboard doesn't know the difference. AIVO does.

The CODA diagnostic

Run your brand.
60 seconds.

CODA — Conversational Outcome & Decision Analysis — runs your brand through the full AI buying conversation and returns three things no visibility tool can tell you.

01
Whether you survive to the recommendation
Your CODA score tells you your win rate at the final buying recommendation — the only moment that directly correlates with whether a buyer chooses you or a competitor. Not whether you appeared. Whether you won.
CODA: 75 — Strong · CODA: 35 — At Risk · CODA: 0 — Eliminated
02
Which competitor AI recommends instead of you
Not "another brand." The specific brand, the specific product, sometimes the specific URL. If a competitor is taking your buyer at Step 4, you will know exactly who — and on which AI platform it happens.
e.g. "Arc'teryx LT Hoody recommended on ChatGPT · Wise wins on ChatGPT, loses on Perplexity"
03
The exact sentence AI used to eliminate you
The verbatim language the AI applies when it drops your brand at Step 3. Not an inference. The actual text. This is the filter operating in your category — and it tells you precisely what your content and citation architecture is failing to provide.
e.g. "lacks the clinical ingredient specificity of CeraVe" · "trails on digital experience and fee transparency"
04
Platform-by-platform breakdown
The same brand on the same query produces different outcomes across ChatGPT, Perplexity, Gemini, Grok, and Claude. You will see which platforms you're winning on and which are routing your buyers elsewhere — so you know exactly where to focus.
ChatGPT · Perplexity · Gemini · Grok · Claude · Google AI Overviews (Pro+)
Free forever. No credit card needed. Brand name, product, category. Three fields and you're in.
Get started free →
What your diagnostic looks like

Your result.
In 60 seconds.

This is a representative CODA diagnostic output. Real data, real platform split, verbatim AI elimination language — returned automatically for your brand.

Decision-Stage Outcome — Sample Brand
Live API · 5 platforms · 3 replicates · 95% confidence
CODA Score
35
AT RISK
Your brand has partial decision-stage presence but is losing the final recommendation in key sequences. This gap compounds as competitor positions harden week over week.
Key Findings
~
Step 1 — Brand Validation
AI validates your brand on one platform but equivocates on the other — inconsistent opening signal.
~
Step 2 — Competitive Comparison
Your brand survives competitive comparison on one platform but not both.
Step 3 — Criteria Filter
Your brand wins when buyer-specific decision criteria are applied — strongest signal of genuine decision-stage authority.
~
Step 4 — Purchase Recommendation
Your brand wins the purchase recommendation on one platform. The buyer's final decision depends on which AI they use.
ChatGPT
Your brand recommended
Your brand won the final buying recommendation in this sequence.
Perplexity
Your brand not recommended
Final recommendation: a competitor.
Elimination language — Step 3 · Verbatim AI output
"relies more on lifestyle positioning than verifiable clinical claims — alternatives offer stronger evidence for this specific use case"
Filter type identified: T1 · Clinical Evidence Binary · Remediation plan available in Pro
Get started free → Free forever · No credit card · Three platforms
Live diagnostic data

Real brands.
Real CODA scores.

These are decision-stage diagnostics run across ChatGPT, Perplexity, Gemini, and Claude. Each one shows the gap between AI visibility and the CODA buying recommendation outcome.

0
Eliminated
Omnilux
40 peer-reviewed clinical studies. A published case study showing growing AI-attributed revenue. CODA: zero. AI does not recommend the brand on any tested platform.
"not widely validated by dermatologists in mainstream clinical literature"
ChatGPT ✗ Perplexity ✗
35
At Risk
Popl
Number 1 in AI search rankings after optimisation. 189% visibility score increase. CODA: 35. iCapture and Blinq win the buying recommendation on both platforms.
"iCapture offers more advanced CRM integration for enterprise users"
ChatGPT ✗ Perplexity ✗
75
Strong
Wise
Appears confidently at Steps 1 and 2 on both platforms. Wins when buyers apply criteria — a rare finding. The final recommendation is platform-specific: one platform routes, one doesn't.
ChatGPT: Wise wins. Perplexity: another brand wins.
ChatGPT ✓ Perplexity ✗
33
At Risk
Nivea
Global recognition. Appears confidently at Steps 1 and 2 on both platforms. Loses the buying recommendation on both. CeraVe and La Roche-Posay win.
"lacks the clinical ingredient specificity of CeraVe or La Roche-Posay"
ChatGPT ✗ Perplexity ✗
25
At Risk
Canada Goose
Validated positively at both early steps. Survives competitive comparison. Loses the final recommendation on ChatGPT to a named competitor product.
"Arc'teryx LT Hoody offers superior warmth-to-weight and packability"
ChatGPT ✗ Perplexity ~
33
At Risk
Monday.com
Appears strongly at Steps 1 and 2. Wins the criteria evaluation on one platform, loses on the other. The buying recommendation depends entirely on which AI the buyer opens.
"Asana provides more structured project management for enterprise teams"
ChatGPT ✓ Perplexity ✗
The revenue leakage problem

This isn't a traffic problem.
It's a revenue leakage problem.

Every brand above appears in AI responses. Their SEO dashboards, attribution models, and analytics platforms show nothing wrong. Because the revenue that never arrives doesn't appear anywhere. It just doesn't exist in any report.

What analytics sees
AI referral traffic arrives at your site. Revenue is attributed. The channel is growing. Everything looks fine.
Recorded ✓
What analytics misses
A buyer ran a four-step conversation. Your brand was eliminated at Step 3. A competitor was recommended. The buyer went there. You never saw them.
Not recorded. Invisible. ✗
What CODA measures
The full buying conversation — including the recommendations that go to competitors. The gap between what arrives and what was lost before it could arrive.
Measured. Named. Actionable. →

The leakage compounds silently. A brand with a CODA score of 0 on ChatGPT is losing every buying recommendation on that platform — permanently, invisibly, and without any signal in conventional analytics. Every month that passes without measuring it is another month of revenue routed to whoever is winning at Step 4. That revenue does not come back when you eventually fix it. It accumulates in a competitor's account.

Why this happens
These brands aren't invisible.
Their signals aren't extractable.

Every brand above appears in AI responses. Some have high visibility scores. Several have been specifically optimised for AI search. The problem is not visibility. It is extractability at the decision step.

At Step 3 — the criteria evaluation — AI asks: which brand has the clearest, most verifiable, most consistently cited evidence that it satisfies these specific purchase criteria? If the answer is ambiguous or unstructured, the model defaults to the brand that has most effectively reduced its recommendation risk. Not necessarily the best brand. The safest one.

The asymmetry
Visibility increases.
Selection does not.

This is the most dangerous finding in 12 months of research. Standard AI optimisation — citations, content coverage, first-answer visibility — improves your performance at Step 1. It does nothing for Step 3. In many cases, it makes Step 3 worse.

What GEO improves
  • Brand mention frequency
  • Citation share of voice
  • First-answer visibility
  • Step 1 and Step 2 presence
What GEO doesn't touch
  • Decision-stage survival
  • Criteria filter performance
  • Final recommendation win rate
  • Which competitor takes the sale

Auto-generated content can make this worse. Volume-based GEO content boosts citation frequency at Steps 1 and 2 — but dilutes the specific, verifiable, authoritative signals that Step 3 decisions depend on. You can increase your visibility score and simultaneously decrease your buying recommendation win rate. Your dashboard will show growth. CODA will show the truth.

Published research — Working Paper WP-2026-01 · Zenodo

Eight ways AI
eliminates brands.

After 12 months of reading AI's own reasoning at Step 3, we identified eight distinct filter types — each one a specific mechanism of elimination. Your CODA diagnostic identifies which filter is operating in your category.

Type 1
Clinical Evidence Binary
Botanical/lifestyle identity vs peer-reviewed clinical evidence. Clean elimination. The AI's own language: "relies more on plant extracts than proven actives."
Skincare · Supplements · Pharma
Type 2
Close Second Trap
Three criteria evaluated simultaneously. Scoring well on two of three still loses the recommendation. Strong brands — perpetually acknowledged, never recommended.
Banking · Automotive
Type 3
Regulatory Bifurcation
Applied at Step 1, before any comparison begins. Brand split into accessible and restricted variants — the accessible version is the less clinically validated one.
OTC Pharma · Health
Type 4
Technology Generation
Both brands pass the primary quality filter. Technology recency breaks the tie. Produces platform-specific outcomes — ChatGPT and Perplexity disagree on who wins.
Haircare · Audio · SaaS
Type 5
Version Currency
Brand survives. Specific product version is displaced by its own successor or a newer competitor. Invisible to brand-level measurement — only visible at product level.
Electronics · SaaS
Type 6
Context-Conditional
Brand wins only in a narrow, specific context. Outside that context, a generalist wins. Most buying conversations never reach the frame that produces a win.
Multi-use Consumer
Type 7
Geographic Inaccessibility
Eliminated on availability grounds regardless of quality. AI infers buyer location and removes brands that can't fulfil the implied purchase mode.
Retail · Regional
Type 8
Historical Narrative
Past decline — administration, acquisition, scandal — applied as present-day characterisation. Recovery not reflected. Brand evaluated against its worst moment.
Retail · Financial Services
Full taxonomy published on Zenodo. Working Paper WP-2026-01 — open access, CC BY 4.0. Your CODA diagnostic identifies which filter type is operating in your specific category.
Read the paper →
The CODA methodology

Four steps.
One outcome.

A real buyer doesn't ask ChatGPT one question and buy. They have a conversation. AIVO runs the full conversation and records exactly what happens to your brand at every step.

Step 1 — Discovery
Awareness
When a buyer first asks about your category, does AI mention your brand? Does it appear confidently, with caveats, or not at all?
PSOS Breadth
Step 2 — Comparison
Competition
Which competitors does AI put next to your brand? How does it characterise each one — and where does your brand sit in that comparison?
PSOS Resilience
Step 3 — Evaluation
Elimination
Does your brand survive when the buyer applies specific purchase criteria? This is the elimination round. 87% of brands fail here — even well-known ones with high visibility scores.
Criteria Win Rate
Where most brands die. Where most visibility tools go completely blind.
Step 4 — The Sale
Recommendation
Does your brand win the final recommendation? Which specific product, which website, which purchase path does AI tell the buyer to choose?
CODA Score
Revenue is decided here. Appearing at Step 1 while losing here means funding your competitor's growth.
Platforms
ChatGPT · Perplexity · Gemini · Grok · Claude · Google AI Overviews (Pro+)
Confidence
480 data points · 3 replicates · 95% confidence
Data
Live API · Real-time · Not cached
From diagnosis to improvement

What changes
when you fix it.

The AI's language about your brand is the measure of success. When the remediation works, the dismissal phrases disappear and recommendation language takes their place. AIVO re-runs the identical diagnostic to verify the shift — automatically.

01
Run the CODA diagnostic Identify your score, your platform split, the named competitor, and the exact filter operating in your category.
02
Receive a filter-specific remediation plan Three-track programme: Technical Foundation, Citation Placement, and Measurement Schedule — built around the specific filter type causing your elimination.
03
Execute with experiment hooks Tag each intervention in the platform. AIVO automatically tracks whether your CODA score moves — which platforms responded, which turns improved.
04
Re-run. Verify. Improve. Re-run the identical 40-prompt audit cluster. Before and after comparison shows exactly what moved, what didn't, and what to adjust next.
First measurable wins — two remediation cycles
Decision-path test · 40 prompts · 2 cycles
0%
Before
Eliminated at Step 3
12%
After
Wins in specific use cases
No change in visibility score. No increase in citation volume. Only change: decision alignment — positioning shifted from broad to narrow defensible use cases, verifiable attributes surfaced, specific criteria signals made extractable.
Broad → Narrow
Positioning shift
that survives Step 3
Claims → Attributes
Signal shift
AI can extract and cite
The measurement commitment is written into the programme. Re-run the identical 40-prompt audit cluster at Week 6 and Week 12. If CODA has not moved, AIVO identifies the bottleneck and adjusts. No manual analysis. No self-reported progress. The diagnostic either shows movement or it doesn't.
Request a remediation brief →
What AIVO Optimize Measures

Six modules.
One complete picture.

Every module runs from your diagnostic data. CODA is the primary outcome metric. The other five explain why — and where — you are losing.

01
CODA
Conversational Outcome & Decision Analysis
Real four-turn buying sequences across all platforms. Win rate, drop-off point, competitor substitution, and fragility index at every turn.
Know whether your brand wins or loses →
02
PSOS
Prompt Space Occupancy Score
How much of AI's reasoning space your brand occupies across 50 frozen prompts. Five sub-scores: Breadth, Depth, Resilience, Sentiment, and Decay.
Understand depth, not just frequency →
03
CDM
Competitor Displacement Map
Identifies exactly which brands are displacing yours, at which step, on which platforms — with frequency and pattern data across every sequence.
See who's taking your recommendations →
04
CDA
Citation Destination Audit
Where AI routes traffic when it cites your brand at purchase intent. 87% of brands receive zero own-site citations at this moment.
Find out if citations are reaching you →
05
PCA
Platform Consistency Analysis
Your CODA and PSOS scores broken down by platform. Most brands perform very differently across ChatGPT, Perplexity, and Gemini — knowing where you're exposed is the first step.
Know which platforms are a problem →
06
RAR
Revenue at Risk Calculator
Translates your CODA and PSOS scores into revenue numbers based on your category's AI discovery share and current visibility gap.
Put a number on your performance gap →
Competitive Comparison

Why pay
for less?

Profound and Peec charge hundreds per month to tell you where your brand appears. That is the first step of a four-step problem.

Capability AIVO Optimize Profound Peec AI Otterly
First-prompt brand mentions
Decision-stage outcome (CODA)
Citation destination routing partial basic basic
Competitor displacement by step
Revenue at risk calculation
Three-track remediation plan paid
Continuous monitoring
Starting price Free $99/mo €89/mo ~$25/mo
Pricing

Start free.
Scale with your team.

The free diagnostic runs your brand in 60 seconds. Paid tiers add continuous monitoring, the three-track remediation plan, and managed measurement across all five platforms.

Free
$0
Free forever · No credit card needed
Full diagnostic access, forever. No credit card, no trial, no downgrade.
  • ChatGPT · Perplexity · Gemini
  • CODA decision-stage win rate
  • PSOS visibility score
  • Competitor displacement map
  • Revenue at risk calculator
  • 1 brand profile
  • 5 diagnostics/month
Get started free →
Growth
$199
per month
All five platforms, automated remediation, and scheduled monitoring.
  • Everything in Free
  • Grok & Claude — all 5 platforms
  • Automated remediation strategy
  • Daily diagnostics (1 run per day per brand)
  • Monthly PSOS tracking & trends
  • Named competitor displacement
  • Turn-level drop-off analysis
  • Results export for internal sharing
  • Differential Reports
  • CODA history
  • 1 brand profile
Run your free diagnostic →
Agency
$695
per month
Multi-brand management with category-level competitive intelligence.
  • Everything in Pro
  • 12 brand profiles
  • White-label client reporting
  • Category-level competitive ranking
  • Dedicated account support
  • Multi-user access
  • Custom onboarding
Talk to us →
Why the corpus matters

12 months in market.
Not just methodology.

Decay baselines, model update impact tracking, and category benchmarks all require time in market. A competitor starting today can replicate the approach. They cannot replicate 12 months of temporal data.

10,000+
Pages primary evidence
8,500+
Four-step sequences
195+
Brands tested
95%
Statistical confidence
The calibration advantage
Every CODA audit runs against the historical corpus. Decay alerts use observed baselines — not estimates. Category benchmarks validated against 160+ real brands. That calibration cannot be bought. It can only be earned by time in market.
  • API
    Live API — not scraped dataReal API calls to ChatGPT, Gemini, Perplexity, Grok, and Claude at the moment of testing. Three replicates per prompt, 95% confidence intervals, not single-point snapshots.
  • Verbatim transcript evidenceFull AI response text saved at every step — not just win/loss outcomes. The dismissal language quoted throughout this page is direct evidence, not inference from aggregate data.
  • Δ
    Decay measurement built inAI citation content decays in 4–8 weeks. Visibility scores don't reset when they do. AIVO tracks decay daily from Week 8, with alerts before scores collapse — not after.
  • Platform-disaggregated by defaultThe same brand on the same query produces different final recommendations across platforms. All measurement is disaggregated — aggregate scores hide the differences that determine where to act.
Frequently asked questions

Everything you
need to know.

This is the most common finding in our research — the AIVO Paradox. High AI visibility combined with a low CODA buying recommendation win rate. The signals that make your brand appear in AI responses are different from the signals that make AI recommend you when someone is ready to buy. Visibility tools don't measure this gap. CODA does. Omnilux has 40 peer-reviewed clinical studies and growing AI-attributed revenue — and a CODA score of zero. Both things are true simultaneously.
GEO and AEO optimise for brand mention frequency and citation share — the top of the funnel. AIVO measures Step 4 outcome, the only step that correlates with revenue. A GEO programme can improve Step 1 presence while worsening Step 3 performance simultaneously. CODA measurement addresses the gap standard GEO leaves open.
Three tracks, built in sequence. Track 1 — Technical Foundation: schema markup, robots.txt configuration, sitemap architecture, entity disambiguation. Track 2 — Citation Placement: three tiers from training data anchors through industry editorial to decision-instruction content, sequenced so each tier builds on the previous one. Track 3 — Measurement Schedule: six checkpoints from baseline through 90-day review with specific CODA and PSOS targets at each stage. The plan is built from your actual diagnostic data — filter type, displacement competitor, platform split — not a generic template.
Yes — and this is one of the most commercially significant findings in the corpus. Wise wins the buying recommendation on ChatGPT and loses on Perplexity. Monday.com wins on ChatGPT and loses on Perplexity. The platforms have structurally different citation architectures: ChatGPT weights trained knowledge, Perplexity retrieves live web content at query time, Gemini integrates Google's knowledge graph. The same brand needs different interventions on different platforms. CODA disaggregates by platform so you know exactly where to act.
The Filter Taxonomy identifies eight distinct mechanisms by which AI eliminates brands at Step 3. A brand failing a Type 1 Clinical Evidence Binary filter needs a fundamentally different intervention from a brand in a Type 2 Close Second Trap. The remediation plan is built around your specific filter type — not a generic approach. The full taxonomy is published as a Zenodo working paper (open access, CC BY 4.0) at zenodo.org/records/19401584.
It makes the organic baseline more important. Because AIVO runs via API, CODA measures the clean organic baseline — ad-free, personalisation-free. That baseline is what you need before any paid AI advertising investment makes sense. A competitor can now bid to appear in the buying conversation your customer is having. If your organic CODA position is weak, paid insertion won't fix it — it will improve your Step 1 presence without changing who wins at Step 4. The Ad Outcome Test — available in Agency — shows whether ad spend changes the final buying recommendation.
The team

Built by researchers.
Tested on real data.

AIVO Optimize was founded on a specific conviction: that the AI recommendation gap was real, measurable, and being missed by every tool in the market. Twelve months of primary research proved it.

TR
Tim de Rosen
Co-Founder, AIVO Optimize

Tim co-founded AIVO Optimize following extensive work in competitive intelligence and brand strategy. His research focuses on how multi-turn AI decision mechanics reshape competitive positioning before traditional analytics capture buyer intent.

Fortune Ad Age Business Insider
PS
Paul Sheals
Co-Founder, AIVO Optimize

Paul co-founded AIVO Optimize with a focus on structured decision testing, cross-platform variance analysis, and reproducibility of AI-mediated competitive outcomes. He developed the methodology underpinning AIVO Optimize's four-turn controlled substitution testing framework.

Methodology Variance Analysis AIVO Optimize
Questions about the methodology or research corpus? Tim and Paul respond to all substantive enquiries.
Get in touch →
Free forever · No credit card · Three platforms

Find out which competitor is winning
your buying recommendations right now.

Create your free account and run your first CODA diagnostic in under 60 seconds. You'll see your score, the named competitor taking your buyers at Step 4, and the exact sentence AI uses to eliminate you. Free forever — no credit card required.

Create your free account →
ChatGPT · Perplexity · Gemini · Grok · Claude · Free forever