The AI Consistency Paradox via @sejournal, @DuaneForrester

Doc Brown’s DeLorean didn’t just travel through time; it created different timelines. Same car, different realities. In “Back to the Future,” when Marty’s actions in the past threatened his existence, his photograph began to flicker between realities depending on choices made across timelines.

This exact phenomenon is happening to your brand right now in AI systems.

ChatGPT on Monday isn’t the same as ChatGPT on Wednesday. Each conversation creates a new timeline with different context, different memory states, different probability distributions. Your brand’s presence in AI answers can fade or strengthen like Marty’s photograph, depending on context ripples you can’t see or control. This fragmentation happens thousands of times daily as users interact with AI assistants that reset, forget, or remember selectively.

The challenge: How do you maintain brand consistency when the channel itself has temporal discontinuities?

The Three Sources Of Inconsistency

The variance isn’t random. It stems from three technical factors:

Probabilistic Generation

Large language models don’t retrieve information; they predict it token by token using probability distributions. Think of it like autocomplete on your phone, but vastly more sophisticated. AI systems use a “temperature” setting that controls how adventurous they are when picking the next word. At temperature 0, the AI always picks the most probable choice, producing consistent but sometimes rigid answers. At higher temperatures (most consumer AI uses 0.7 to 1.0 as defaults), the AI samples from a broader range of possibilities, introducing natural variation in responses.

The same question asked twice can yield measurably different answers. Research shows that even with supposedly deterministic settings, LLMs display output variance across identical inputs, and studies reveal distinct effects of temperature on model performance, with outputs becoming increasingly varied at moderate-to-high settings. This isn’t a bug; it’s fundamental to how these systems work.

Context Dependence

Traditional search isn’t conversational. You perform sequential queries, but each one is evaluated independently. Even with personalization, you’re not having a dialogue with an algorithm.

AI conversations are fundamentally different. The entire conversation thread becomes direct input to each response. Ask about “family hotels in Italy” after discussing “budget travel” versus “luxury experiences,” and the AI generates completely different answers because previous messages literally shape what gets generated. But this creates a compounding problem: the deeper the conversation, the more context accumulates, and the more prone responses become to drift. Research on the “lost in the middle” problem shows LLMs struggle to reliably use information from long contexts, meaning key details from earlier in a conversation may be overlooked or mis-weighted as the thread grows.

For brands, this means your visibility can degrade not just across separate conversations, but within a single long research session as user context accumulates and the AI’s ability to maintain consistent citation patterns weakens.

Temporal Discontinuity

Each new conversation instance starts from a different baseline. Memory systems help, but remain imperfect. AI memory works through two mechanisms: explicit saved memories (facts the AI stores) and chat history reference (searching past conversations). Neither provides complete continuity. Even when both are enabled, chat history reference retrieves what seems relevant, not everything that is relevant. And if you’ve ever tried to rely on any system’s memory based on uploaded documents, you know how flaky this can be – whether you give the platform a grounding document or tell it explicitly to remember something, it often overlooks the fact when needed most.

Result: Your brand visibility resets partially or completely with each new conversation timeline.

The Context Carrier Problem

Meet Sarah. She’s planning her family’s summer vacation using ChatGPT Plus with memory enabled.

Monday morning, she asks, “What are the best family destinations in Europe?” ChatGPT recommends Italy, France, Greece, Spain. By evening, she’s deep into Italy specifics. ChatGPT remembers the comparison context, emphasizing Italy’s advantages over the alternatives.

Wednesday: Fresh conversation, and she asks, “Tell me about Italy for families.” ChatGPT’s saved memories include “has children” and “interested in European travel.” Chat history reference might retrieve fragments from Monday: country comparisons, limited vacation days. But this retrieval is selective. Wednesday’s response is informed by Monday but isn’t a continuation. It’s a new timeline with lossy memory – like a JPEG copy of a photograph, details are lost in the compression.

Friday: She switches to Perplexity. “Which is better for families, Italy or Spain?” Zero memory of her previous research. From Perplexity’s perspective, this is her first question about European travel.

Sarah is the “context carrier,” but she’s carrying context across platforms and instances that can’t fully sync. Even within ChatGPT, she’s navigating multiple conversation timelines: Monday’s thread with full context, Wednesday’s with partial memory, and of course Friday’s Perplexity query with no context for ChatGPT at all.

For your hotel brand: You appeared in Monday’s ChatGPT answer with full context. Wednesday’s ChatGPT has lossy memory; maybe you’re mentioned, maybe not. Friday on Perplexity, you never existed. Your brand flickered across three separate realities, each with different context depths, different probability distributions.

Your brand presence is probabilistic across infinite conversation timelines, each one a separate reality where you can strengthen, fade, or disappear entirely.

Why Traditional SEO Thinking Fails

The old model was somewhat predictable. Google’s algorithm was stable enough to optimize once and largely maintain rankings. You could A/B test changes, build toward predictable positions, defend them over time.

That model breaks completely in AI systems:

No Persistent Ranking

Your visibility resets with each conversation. Unlike Google, where position 3 carries across millions of users, in AI, each conversation is a new probability calculation. You’re fighting for consistent citation across discontinuous timelines.

Context Advantage

Visibility depends on what questions came before. Your competitor mentioned in the previous question has context advantage in the current one. The AI might frame comparisons favoring established context, even if your offering is objectively superior.

Probabilistic Outcomes

Traditional SEO aimed for “position 1 for keyword X.” AI optimization aims for “high probability of citation across infinite conversation paths.” You’re not targeting a ranking, you’re targeting a probability distribution.

The business impact becomes very real. Sales training becomes outdated when AI gives different product information depending on question order. Customer service knowledge bases must work across disconnected conversations where agents can’t reference previous context. Partnership co-marketing collapses when AI cites one partner consistently but the other sporadically. Brand guidelines optimized for static channels often fail when messaging appears verbatim in one conversation and never surfaces in another.

The measurement challenge is equally profound. You can’t just ask, “Did we get cited?” You must ask, “How consistently do we get cited across different conversation timelines?” This is why consistent, ongoing testing is critical. Even if you have to manually ask queries and record answers.

The Three Pillars Of Cross-Temporal Consistency

1. Authoritative Grounding: Content That Anchors Across Timelines

Authoritative grounding acts like Marty’s photograph. It’s an anchor point that exists across timelines. The photograph didn’t create his existence, but it proved it. Similarly, authoritative content doesn’t guarantee AI citation, but it grounds your brand’s existence across conversation instances.

This means content that AI systems can reliably retrieve regardless of context timing. Structured data that machines can parse unambiguously: Schema.org markup for products, services, locations. First-party authoritative sources that exist independent of third-party interpretation. Semantic clarity that survives context shifts: Write descriptions that work whether the user asked about you first or fifth, whether they mentioned competitors or ignored them. Semantic density helps: keep the facts, cut the fluff.

A hotel with detailed, structured accessibility features gets cited consistently, whether the user asked about accessibility at conversation start or after exploring ten other properties. The content’s authority transcends context timing.

2. Multi-Instance Optimization: Content For Query Sequences

Stop optimizing for just single queries. Start optimizing for query sequences: chains of questions across multiple conversation instances.

You’re not targeting keywords; you’re targeting context resilience. Content that works whether it’s the first answer or the fifteenth, whether competitors were mentioned or ignored, whether the user is starting fresh or deep in research.

Test systematically: Cold start queries (generic questions, no prior context). Competitor context established (user discussed competitors, then asks about your category). Temporal gap queries (days later in fresh conversation with lossy memory). The goal is minimizing your “fade rate” across temporal instances.

If you’re cited 70% of the time in cold starts but only 25% after competitor context is established, you have a context resilience problem, not a content quality problem.

3. Answer Stability Measurement: Tracking Citation Consistency

Stop measuring just citation frequency. Start measuring citation consistency: how reliably you appear across conversation variations.

Traditional analytics told you how many people found you. AI analytics must tell you how reliably people find you across infinite possible conversation paths. It’s the difference between measuring traffic and measuring probability fields.

Key metrics: Search Visibility Ratio (percentage of test queries where you’re cited). Context Stability Score (variance in citation rate across different question sequences). Temporal Consistency Rate (citation rate when the same query is asked days apart). Repeat Citation Count (how often you appear in follow-up questions once established).

Test the same core question across different conversation contexts. Measure citation variance. Accept the variance as fundamental and optimize for consistency within that variance.

What This Means For Your Business

For CMOs: Brand consistency is now probabilistic, not absolute. You can only work to increase the probability of consistent appearance across conversation timelines. This requires ongoing optimization budgets, not one-time fixes. Your KPIs need to evolve from “share of voice” to “consistency of citation.”

For content teams: The mandate shifts from comprehensive content to context-resilient content. Documentation must stand alone AND connect to broader context. You’re not building keyword coverage, you’re building semantic depth that survives context permutation.

For product teams: Documentation must work across conversation timelines where users can’t reference previous discussions. Rich structured data becomes critical. Every product description must function independently while connecting to your broader brand narrative.

Navigating The Timelines

The brands that succeed in AI systems won’t be those with the “best” content in traditional terms. They’ll be those whose content achieves high-probability citation across infinite conversation instances. Content that works whether the user starts with your brand or discovers you after competitor context is established. Content that survives memory gaps and temporal discontinuities.

The question isn’t whether your brand appears in AI answers. It’s whether it appears consistently across the timelines that matter: the Monday morning conversation and the Wednesday evening one. The user who mentions competitors first and the one who doesn’t. The research journey that starts with price and the one that starts with quality.

In “Back to the Future,” Marty had to ensure his parents fell in love to prevent himself from fading from existence. In AI search, businesses must ensure their content maintains authoritative presence across context variations to prevent their brands from fading from answers.

The photograph is starting to flicker. Your brand visibility is resetting across thousands of conversation timelines daily, hourly. The technical factors causing this (probabilistic generation, context dependence, temporal discontinuity) are fundamental to how AI systems work.

The question is whether you can see that flicker happening and whether you’re prepared to optimize for consistency across discontinuous realities.

More Resources:

This post was originally published on Duane Forrester Decodes.

Featured Image: Inkoly/Shutterstock

Google Isn’t Going Anywhere: Ahrefs Ambassador On LLM Inclusion & Why Relationships Still Win via @sejournal, @theshelleywalsh

There’s a divided line in the industry between those who think optimizing for AI is separate from SEO and those who think LLM discovery is just SEO. But, this is an unproductive argument, because whatever you think, LLM inclusion is now part of SEO discovery.

So, let’s just focus on how the search journey works now and where you can find real business value.

To discuss inclusion in LLMs, I invited Patrick Stox to the latest edition of IMHO to find out what he thinks. As product advisor, technical SEO, and brand ambassador at Ahrefs, Patrick has plenty of data to work with and insights into what’s actually working for LLM inclusion right now.

In the face of the AI takeover, Patrick’s take is that Google isn’t going anywhere, and he still thinks human relationships are critical.

You can watch the full interview with Patrick on IMHO below.

Google Isn’t Going Anywhere

With the industry obsessing over ChatGPT, AI Overviews, and AI Mode, it’s easy to assume that traditional search really is dead. However, Patrick was quick to say, “I’m not betting against Google.”

“Google is still everything for most people … Most of the people that are using [LLMs] are tech forward, but the majority of folks are still just Googling things”

Recent Ahrefs data estimated that Google owns an estimated 40% of all traffic to websites, with LLM referrals still a fraction by comparison. Although Google’s share of traffic may be down a couple of percent this year, it still dominates.

After experimenting with ChatGPT and Claude when they first launched, Patrick found himself returning to Google’s AI Mode and Gemini, and thinks others will do the same. “Even I just went back to Google,” he admitted. “I think we’re going to see more of that as they improve their systems.”

Google continues releasing competitive AI innovations, and Patrick predicts these will pull many users back into Google’s ecosystem.

“I’m not betting against Google,” he says. “They’ve got more data than anyone, and they’re still on the bleeding edge.”

The Attribution Problem: LLMs Might Drive Conversions, But We Can’t Prove It

Even though sites are seeing growing referrals from LLMs, establishing attribution to any real value from LLM traffic is a challenge right now. We can talk about brand awareness, but C-Suite is only interested in business value.

Patrick agreed that while you can count mentions and citations in AI answers, that doesn’t easily translate into board-level reporting.

“You can measure how often you’re mentioned versus competitors … but going back to a business, I can’t report on that stuff. It’s all secondary, tertiary metrics.”

For Patrick, revenue and revenue-adjacent metrics still matter. That said, Ahrefs has had some signals from AI search traffic.

“We did track the signups. When I first looked at this data back in July, all the traffic from AI search was half a percent of our traffic total. But at the time, it was 12.1% of our total conversions.” He explained.

This has now dropped below 10%, while the traffic share has grown slightly.

Two Strategies That Are Working For LLM Inclusion

I asked if Ahrefs is actively investing in LLM inclusion, and Patrick said they are trying a number of different things, and the two fundamental approaches that determine LLM visibility are repetition and differentiation.

“Whatever the internet says, that’s kind of what’s being returned in these systems,”

Repetition means ensuring consistent messaging across multiple websites. LLMs synthesize what “the internet says,” so if you want to be recognized for something, that narrative needs to exist broadly. For Ahrefs, this has meant actively spreading the message that they have evolved beyond just SEO tools into a comprehensive digital marketing platform.

Differentiation through original data works alongside the repetition to stand out. Ahrefs has invested heavily in unique data studies throughout the year, including non-English language research. “This data is being heavily cited, heavily returned in these systems because there’s nothing else out there like it,” Patrick explained.

The more surprising tactic that is also currently working is listicles.

“I hate to say it, but listicles … they work right now. I don’t think it’s future-proof at all, but at the same time, I don’t want to just not be there.”

Agentic AI And The Threat Of Closed Systems

I then asked about agentic AI and systems, and does Patrick have concerns about systems becoming closed.

As LLM agents begin booking travel, making purchases, or accessing APIs directly, most likely they would rely on a small set of partners from big brands.

“ChatGPT isn’t going to make deals with unknown companies,” Stox says. “If they book flights, they’ll use major providers. If they use a dictionary, they’ll pick one dictionary.”

This would be the real threat to smaller businesses. “If an agent decides ‘we only check out through Amazon,’ a lot of stores lose sales overnight,” Patrick warns. There is no guaranteed defense. The only strategy we can follow right now is to grow your brand and footprint.

“What was the thing they used to say for Google? Make them embarrassed to not have you included.”

Beyond LLM Optimization: Channels That Still Matter

Patrick emphasized a point that’s possibly been forgotten in the AI hype: “It’s not ChatGPT that’s the second largest search engine, it’s still YouTube by far.”

YouTube has been a hugely successful referral platform for Ahrefs, and the company invested heavily in video. Patrick recommends both long and short-form, for brand discovery.

Community participation on platforms such as Reddit, Slack, and Discord also offers substantial value, but only when companies genuinely participate rather than spam.

While many brands have tried to brute-force Reddit with spam, Patrick says there can be huge value in genuine participation, especially when employees are allowed to represent the company authentically.

“You have literally a paid workforce of advocates who work for your company. Let them go out and talk to people … answer questions, basically advertise for you. They want to do it already. So let them.”

If You Started A Product Today, Where Would You Bet?

As a final question, I asked Patrick where he’d invest if launching a startup today; he did not hesitate to say relationships.

“If I launched a startup, the first thing I’d invest in is relationships. That’s still the most powerful channel … I think if I did do something like that, I’d probably grow it pretty fast. More from my connections than anything else,” he said.

After relationships, he’d focus on YouTube, website content creation, and telling friends about the product. In other words, “just normal marketing.”

“We’ve gone through this tech revolution, and now we’re realizing everything still comes back to direct connections with people.”

And that may be the most important insight of all. In an era of AI-driven discovery, the brands that win are the ones that remain unmistakably human.

Watch the full video interview with Patrick Stox here:

Thank you to Patrick Stox for offering his insights and being my guest on IMHO.

More Resources:

Featured Image: Shelley Walsh/Search Engine Journal

Selling AI Search Strategies To Leadership Is About Risk via @sejournal, @Kevin_Indig

Boost your skills with Growth Memo’s weekly expert insights. Subscribe for free!

AI search visibility isn’t “too risky” to invest in for executives to buy-in. Selling AI search strategies to leadership is about risk.

Image Credit: Kevin Indig

A Deloitte survey of +2,700 leaders reveals that getting buy-in for an AI search strategy isn’t about innovation, but risk.

SEO teams keep failing to sell AI search strategies for one reason: They’re pitching deterministic ROI in a probabilistic environment.

The old way: Rankings → traffic → revenue. But that event chain doesn’t exist in AI systems.

LLMs don’t rank. They synthesize. And Google’s AI Overviews and AI Mode don’t “send traffic.” They answer.

Yet most teams still walk into a leadership meeting with a deck built on a decaying model. Then, executives say no – not because AI search “doesn’t work,” but because the pitch asks them to fund an outcome nobody can guarantee.

In AI search, you cannot sell certainty. You can only sell controlled learning.

1. You Can’t Sell AI Search With A Deterministic ROI Model

Everyone keeps asking the wrong question: “How do I prove my AI search strategy will work so leadership will fund it?” You can’t; there’s no traffic chain you can model. Randomness is baked directly into the outputs.

You’re forcing leadership to evaluate your AI search strategy with a framework that’s already decaying. Confusion about AI search vs. traditional SEO metrics and forecasting is blocking you from buy-in. When SEO teams try to sell an AI search strategy to leadership, they often encounter several structural problems:

1. Lack of clear attribution and ROI: Where you see opportunity, leadership sees vague outcomes and deprioritizes investment. Traffic and conversions from AI Overviews, ChatGPT, or Perplexity are hard to track.
2. Misalignment with core business metrics: It’s harder to tie results to revenue, CAC, or pipeline – especially in B2B.
3. AI search feels too experimental: Early investments feel like bets, not strategy. Leadership may see this as a distraction from “real” SEO or growth work.
4. No owned surfaces to leverage: Many brands aren’t mentioned in AI answers at all. SEO teams are selling a strategy that has no current baseline.
5. Confusion between SEO and AI search strategy: Leadership doesn’t understand the distinction between optimizing for classic Google Search vs. LLMs vs. AI Overviews. Clear differentiation is needed to secure a new budget and attention.
6. Lack of content or technical readiness: The site lacks the structured content, brand authority, or documentation to appear in AI-generated results.

2. Pitch AI Search Strategy As Risk Mitigation, Not Opportunity

Executives don’t buy performance in ambiguous environments. They buy decision quality. And the decision they need you to make is simple: Should your brand invest in AI-driven discovery before competitors lock in the advantage – or not?

Image Credit: Kevin Indig

AI search is still an ambiguous environment. That’s why your winning strategy pitch should be structured for fast, disciplined learning with pre-set kill criteria instead of forecasting traffic → revenue. Traditionally, SEO teams pitch outcomes (traffic, conversions), but leadership needs to buy learning infrastructure (testing systems, measurement frameworks, kill criteria) for AI search.

Leadership thinks you’re asking for “more SEO budget” when you’re actually asking them to buy an option on a new distribution channel.

Everyone treats the pitch as “convince them it will work” when it should be “convince them the cost of not knowing is higher than the cost of finding out.” Executives don’t need certainty about impact – they need certainty that you’ll produce a decision with their money.

Making stakes crystal clear:

Your Point of View + Consequences = Stakes. Leaders need to know what happens if they don’t act.

Image Credit: Kevin Indig

The cost of passing on an AI search strategy can be simple and brutal:

1. Competitors who invest early in AI search visibility will build entity authority and brand presence.
2. Organic traffic stagnates and will drop over time while cost-per-click rises.
3. AI Overviews and AI Mode outputs will replace queries your brand used to win in Google.
4. Your influence on the next discovery channel will be decided without you.

AI search strategy builds brand authority, third-party mentions, entity relationships, content depth, pattern recognition, and trust signals in LLMs. These signals compound. They also freeze into the training data of future models.

If you aren’t shaping that footprint now, the model will rely on whatever scraps already exist based on whatever your competitors are feeding it.

3. Sell Controlled Experiments – Small, Reversible, And Time-Boxed

You’re asking for resources to discover the truth before the market makes the decision for you. This approach collapses resistance because it removes the fear of sunk cost and turns ambiguity into manageable, reversible steps.

A winning AI search strategy proposal sounds like:

• “We’ll run x tests over 12 months.”
• “Budget: ≤0.3% of marketing spend.”
• “Three-stage gates with Go/No-Go decisions.”
• “Scenario ranges instead of false-precision forecasts.”
• “We stop if leading indicators don’t move by Q3.”

45% of executives rely more on instinct than facts. Balance your data with a compelling narrative – focus on outcomes and stakes, not technical details.

I covered how to build a pitch deck and strategic narrative in how to explain the value of SEO to executives, but focus on selling learning as a deliverable under the current AI search landscape.

When presenting to leaders, they focus on three things only: money (revenue, profit, cost), market (market share, time-to-market), and exposure (retention, risk). Structure every pitch around these.

The SCQA framework (Minto Pyramid) guides you:

• Situation: Set the context.
• Complication: Explain the problem.
• Question: What should we do?
• Answer: Your recommendation.

This is the McKinsey approach – and executives expect it.

Featured Image: Paulo Bobita/Search Engine Journal