How to choose a link building agency in the AI SEO era by uSERP

Remember when a handful of links from sites in your niche could drive steady organic traffic? That era is over.

Today, Google’s AI Overviews and the rise of answer engines like ChatGPT raise the bar. You have to do more to stay visible. Hiring an experienced link building agency is one efficient way to meet that challenge.

It’s also one of the most important investments you’ll make. The right partner doesn’t just build links. They position your brand as a trusted, cited source in the AI era.

So how do you choose the right agency for your company?

While the interface has changed, the core ranking signals remain largely the same. What’s changed is their priority.

LLMs need credible sources to ground their answers. That makes authoritative link building more important than ever.

This article shows you how to vet and choose a link building agency that understands these new priorities and can help your brand win trust in the AI-driven SEO landscape.

How link building and SEO are changing

Gartner predicted search engine volume to drop by 25% as AI takes over more answers. That makes working with an agency that understands AI SEO essential.

But how do you know which agencies actually do?

The real indicators are holistic authority and AI visibility. Only one in five links cited in Google’s AI Overviews matched a top-10 organic result, according to an Authoritas study. Even more telling, 62.1% of cited links or domains didn’t rank in the top 10 at all.

The takeaway is simple. AI systems and search engines don’t evaluate websites the same way. We’re no longer building links just for Google’s crawler.

Link equity alone isn’t enough. Sites need topical authority, brand mentions, and real market presence. The goal is to build a footprint that AI models recognize and can’t ignore.

The new criteria: Evaluating a link building agency for AI SEO

Choosing the right link building agency comes down to how well they prioritize the factors that matter now.

This section shows you what to look for.

Prioritizing quality, relevance, and traffic

I see this mistake all the time. A marketing director evaluates link quality based only on Domain Rating (DR).

High DR matters, but at uSERP, we know it’s not the finish line. You should also look for:

  • Relevance: A link from a DR 60, niche-specific site in your industry often beats a DR 80 general news site that covers everything from crypto to keto.
  • Minimum traffic standards: If a site doesn’t rank for keywords or attract real traffic, its links won’t help you rank. That’s why strict traffic minimums matter.

When vetting an agency, ask for contractual site-traffic guarantees.

A confident agency won’t hesitate to sign a Statement of Work that guarantees every link comes from a site with a minimum traffic threshold, such as 5,000+ monthly organic visitors.

If they won’t put traffic minimums in writing, they’re likely planning to place links on “ghost town” sites. These domains appear strong, but they lack a real audience, which protects their margins rather than supporting your growth.

Look for a content-driven approach and digital PR

Links don’t exist in a vacuum. The strongest ones come from being part of a real conversation.

The best agencies no longer operate like traditional link builders. They act more like content marketing and digital PR teams. 

Instead of asking for links, the best agencies create linkable assets — data studies, expert commentary, and in-depth guides that journalists and publishers want to cite – because they understand:

  • Google’s algorithms and AI models are continually getting better at identifying paid placements. A content-led approach keeps links natural, editorial, and valuable to readers.
  • Guest posting in the AI SEO era isn’t about a disposable 500-word article. It’s about thought leadership that positions your CEO as a credible expert.

At uSERP, for example, we created — and continuously update — our State of Backlinks for SEO report.

Red flags: Recognizing outdated or dangerous tactics

Choosing the wrong partner doesn’t just waste your budget. It puts your brand reputation — and potentially your company’s future — at risk.

Here are the biggest red flags to avoid when hiring an agency:

Guaranteed rankings

No one can guarantee a number-one ranking on Google. Any agency that promises specific keyword positions on a fixed timeline is likely doing one of two things:

  • Using risky, short-term tactics to force a temporary spike.
  • Selling you snake oil.

These agencies often rely on private blog networks (PBNs) or aggressive anchor text manipulation to manufacture fast results.

You might see an early jump, but the crash that follows—and the risk of a penalty when Google’s spam systems catch up—is never worth it.

Lack of transparency

If an agency won’t explain how they earn links or where placements will come from before you pay, walk away.

Reputable agencies are transparent. They’ll show real examples of past placements and share relevant case studies from your industry.

Agencies that hide their inventory usually do it for a reason. Those sites are often part of a low-quality network or link farm.

Self-serve link portfolios

If you’re a marketer or SEO on LinkedIn, chances are you’ve received a message like this:

This is a common tactic among low-quality link builders: reselling backlinks from a shared inventory. I understand the appeal.

Strategic link acquisition is hard. Buying and flipping links is easy.

The problem — for you — is the footprint. If an agency can secure a link by filling out a form, anyone can. That includes casino affiliates, gambling sites, adult content, and outright scammers.

That’s not a natural link profile. Google has almost certainly already identified and burned those domains.

In the best case, you pay for a link that passes zero authority. In the worst case, Google flags your site as part of a link scheme.

Dirt-cheap packages

SEO and link building deliver incredible ROI, but they aren’t cheap.

You can’t buy a high-quality article with a real, earned link from an authoritative site for $50. Speaking as someone who runs an AI SEO agency, the true cost of quality content, editing, outreach, and relationship building is at least an order of magnitude higher.

That’s why cheap packages that promise multiple high-authority links are a major red flag. They almost always rely on:

  • Fully AI-generated, barely edited content.
  • Low-value link farms or resold inventory.
  • Toxic backlinks.

None of those will help you show up on AI search engines or Google.

Partnering with a link building agency for a sustainable market presence

Link building in the AI era is a long-term investment. It’s about building a durable market presence, not chasing quick wins.

The right partner sees themselves as an extension of your team. They care about:

  • Your backlink gap compared to competitors.
  • Your brand mentions across LLMs.
  • Your overall search and AI visibility.

They help you navigate content syndication, backlink audits, content marketing, and modern link building strategies with a unified approach.

If you’re ready to move past vanity metrics and start building authority that drives revenue and AI citations, it’s time to be selective about who you trust with your domain.

The right link building agency is out there. You just need to know how to spot them.

Read more at Read More

News publishers expect search traffic to drop 43% by 2029: Report

News executives expect search referrals to drop by more than 40% over the next three years, as search engines continue evolving into AI-driven answer engines, according to a new Reuters Institute report. That shift is squeezing publisher traffic and accelerating a move away from classic SEO toward AEO and GEO.

Why we care. Google’s AI Overviews and chatbot-style search are changing how people get information, often without clicking through. SEO visibility, attribution, and ROI models built on old playbooks are breaking fast.

What’s happening. Publishers expect search traffic to nearly halve. Survey respondents forecast search engine traffic down 43% within three years, with a fifth of respondents expecting losses above 75%.

  • Google referrals are already falling. Chartbeat data cited in the report show organic Google search traffic down 33% globally from November 2024 to November 2025, and down 38% in the U.S. over the same period.
  • AI Overviews are a major factor. Google’s AI Overviews appear at the top of roughly 10% of U.S. search results, with studies showing higher zero-click behavior when they appear, according to the report.
  • The impact is uneven. Lifestyle and utility content (e.g., weather, TV guides, horoscopes) appear to be the most exposed, while hard news queries have been more insulated so far.

SEO to AEO and GEO. The Reuters Institute expects rapid growth in answer engine optimization (AEO) and generative engine optimization (GEO) as publishers and agencies adapt to AI-led interfaces.

  • AEO and GEO services are set to surge. Agencies are repurposing SEO playbooks for chatbots and overview boxes, with new demands on how content is written, structured, and surfaced.
  • Publishers are dialing back traditional SEO. Many survey respondents plan to reduce investment in classic Google SEO and focus more on distribution through AI platforms like ChatGPT, Gemini, and Perplexity.

Between the lines. This is about more than rankings. It’s about distribution inside platforms that publishers do not control.

  • Chat referrals are growing, but remain small. Traffic from ChatGPT is rising quickly, but the report calls it a rounding error compared with Google.
  • Attribution is getting murkier. If AI agents summarize content and complete tasks for users, it becomes unclear what counts as a visit and how monetization works.
  • Licensing is becoming a parallel strategy. As referral risk grows, publishers are turning to AI licensing, revenue-sharing deals, and negotiated citation or prominence as another path to value.

What to watch. A new KPI stack is emerging. Metrics like share of answer, citation visibility, and brand recall may matter as much as clicks.

  • Utility content faces the biggest squeeze. Categories built for fast answers are easiest for AI systems to commoditize.
  • A measurement arms race is coming. Expect new tools to separate human visits from agent consumption and to measure value beyond raw traffic.

Bottom line. Publishers are bracing for a world where search still matters, but clicks matter less. The report’s message is clear: when AI answers become the interface, AEO, GEO, and attribution strategy are no longer optional. They are a core modern search strategy.

The report. Journalism, media, and technology trends and predictions 2026

Read more at Read More

Google opens Olympic live sports inventory to biddable CTV buys

Live sports advertising is getting more programmatic — and more measurable.

Driving the news. Google is expanding biddable live sports in Display & Video 360, giving advertisers programmatic access to NBCUniversal’s Olympic Winter Games inventory ahead of a crowded 2026 global sports calendar.

Why we care. Live sports remain one of the few media environments that consistently deliver massive, attentive audiences. By moving premium sports inventory into biddable CTV, Google gives advertisers more control, stronger measurement, and simpler activation — without sacrificing reach.

What’s new. Advertisers can now pair Google audience signals with NBCUniversal’s live sports CTV inventory to reach fans on the big screen and re-engage them across YouTube and other Google surfaces.

  • New household-level frequency management reduces overexposure, while Google’s AI-powered cross-device conversion tracking connects CTV impressions to downstream purchases at no additional cost.
  • Google is also streamlining access to live sports with a redesigned Marketplace.
  • You can activate curated sports packages in just a few clicks instead of managing fragmented media buys.

The big picture. As fans move fluidly between connected TV, YouTube, Search and social feeds, advertisers are under pressure to follow attention across screens. Google is positioning Display & Video 360 as the hub that connects those moments, from the living room to mobile.

Bottom line: By unlocking Olympic and live sports inventory inside Display & Video 360, Google is making premium sports advertising easier to buy, easier to measure, and far more accountable.

Read more at Read More

Google expands Shopping promotion rules ahead of 2026

Inside Google Ads’ AI-powered Shopping ecosystem: Performance Max, AI Max and more

Google is broadening what counts as an eligible promotion in Shopping, giving merchants more flexibility heading into next year.

Driving the news. Google is update its Shopping promotion policies to support additional promotion types, including subscription discounts, common promo abbreviations, and — in Brazil — payment-method-based offers.

Why we care. Promotions are a key lever for visibility and conversion in Shopping results. These changes unlock more promotion formats that reflect how consumers actually buy today, especially subscriptions and cashback offers. Greater flexibility in promotion types and language reduces disapprovals and makes Shopping ads more competitive at key decision moments.

For retailers relying on subscriptions or local payment incentives, this update creates new ways to drive visibility and conversion on Google Shopping.

What’s changing. Google will now allow promotions tied to subscription fees, including free trials and percent- or amount-off discounts. Merchants can set these up by selecting “Subscribe and save” in Merchant Center or by using the subscribe_and_save redemption restriction in promotion feeds. Examples include a free first month on a premium subscription or a steep discount for the first few billing cycles.

Google is also loosening restrictions on language. Common promotional abbreviations like BOGO, B1G1, MRP and MSRP are now supported, making it easier for retailers to mirror real-world retail messaging without risking disapproval.

In Brazil only, Google will now support promotions that require a specific payment method, including cashback offers tied to digital wallets. Merchants must select “Forms of payment” in Merchant Center or use the forms_of_payment redemption restriction. Google says there are no immediate plans to expand this change to other markets.

Between the lines. These updates signal Google’s intent to better align Shopping promotions with modern retail models — especially subscriptions and localized payment behaviors — while reducing friction for merchants.

The bottom line. By expanding eligible promotion types, Google is giving advertisers more room to compete on value, not just price, when Shopping policies update in January 2026.

Read more at Read More

Apple is finally upgrading Siri, and Google Gemini will power it

Apple

Apple is teaming up with Google to power its next generation of AI features, including a long-awaited Siri upgrade.

What’s happening: Apple will use Google’s Gemini AI models and cloud infrastructure to support future Apple Foundation Models. The multi-year partnership is expected to roll out later this year.

Why we care. With Gemini powering Siri, Apple’s assistant should become a true AI answer engine. That will likely change how millions of iOS users find information, ask questions, and interact with search.

Driving the news. Apple said it chose Google after a “careful evaluation,” calling Gemini the “most capable foundation” for its AI ambitions.

  • We learned in September that Apple was in talks to use a custom Gemini model to power a revamped Siri.
  • Apple delayed its Siri AI upgrade last year, despite marketing the feature. The delay intensified scrutiny of Apple’s AI strategy.

What they’re saying. Here’s a statement Google shared via X:

Apple and Google have entered into a multi-year collaboration under which the next generation of Apple Foundation Models will be based on Google’s Gemini models and cloud technology. These models will help power future Apple Intelligence features, including a more personalized Siri coming this year. After careful evaluation, Apple determined that Google’s Al technology provides the most capable foundation for Apple Foundation Models and is excited about the innovative new experiences it will unlock for Apple users. Apple Intelligence will continue to run on Apple devices and Private Cloud Compute, while maintaining Apple’s industry-leading privacy standards.

The bigger picture. Google briefly crossed a $4 trillion market cap last week, surpassing Apple for the first time since 2019.

  • Google’s Gemini 3 model launched late last year as part of its broader AI push.
  • Apple largely stayed out of the AI arms race that followed ChatGPT’s launch in late 2022 while rivals poured billions into models, chips, and cloud infrastructure.

Read more at Read More

3 PPC myths you can’t afford to carry into 2026

SEO myths vs facts

PPC advice in 2025 leaned hard on AI and shiny new tools. 

Much of it sounded credible. Much of it cost advertisers money. 

Teams followed platform narratives instead of business constraints. Budgets grew. Efficiency did not.

As 2026 begins, carrying those beliefs forward guarantees more of the same. 

This article breaks down three PPC myths that looked smart in theory, spread quickly in 2025, and often drove poor decisions in practice. 

The goal is simple: reset priorities before repeating expensive mistakes.

Myth 1: Forget about manual targeting, AI does it better

We have seen this claim everywhere: 

AI outperforms humans at targeting, and manual structures belong to the past. 

Consolidate campaigns as much as possible. 

Let AI run the show.

There is truth in that – but only under specific conditions. 

AI performance depends entirely on inputs. No volume means no learning. No learning means no results. 

A more dangerous version of the same problem is poor signal quality. No business-level conversion signal means no meaningful optimization.

For ecommerce brands that feed purchase data back into Google Ads and consistently generate at least 50 conversions per bid strategy each month, trusting AI with targeting can make sense. 

In those cases, volume and signal quality are usually sufficient. Put simply, AI favors scale and clear outcomes.

That logic breaks down quickly for low-volume campaigns, especially those optimizing to leads as the primary conversion. 

Without enough high-quality conversions, AI cannot learn effectively. The result is not better performance, but automation without improvement.

How to fix this

Before handing targeting decisions entirely to AI, you should be able to answer “yes” to all three of the questions below:

  • Are campaigns optimized against a business-level KPI, such as CAC or a ROAS threshold?
  • Are enough of those conversions being sent back to the ad platforms?
  • Are those conversions reported quickly, with minimal latency?

If the answer to any of these is no, 2026 should be about reassessing PPC fundamentals.

Do not be afraid to go old school when the situation calls for it. 

In 2025, I doubled a client’s margin by implementing a match-type mirroring structure and pausing broad match keywords.

It ran counter to prevailing best practices, but it worked. 

The decision was grounded in historical performance data, shown below:

Match type Cost per lead Customer acquisition cost Search impression share
Exact €35 €450 24%
Phrase €34 €1,485 17%
Broad €33 €2,116 18%

This is a classic case of Google Ads optimizing to leads and delivering exactly what it was asked to do: drive the lowest possible cost per lead across all audiences. 

The algorithm is literal. It does not account for downstream outcomes, such as business-level KPIs.

By taking back control, you can direct spend toward top-performing audiences that are not yet saturated. In this case, that meant exact match keywords.

If you are not comfortable with older structures like match-type mirroring – or even SKAGs – learning advanced semantic techniques is a viable alternative. 

Those approaches can provide a more controlled starting point without relying entirely on automation.

Myth 2: Meta’s Andromeda means more ads, better results

This myth is particularly frustrating because it sounds logical and spreads quickly. 

The claim is simple: more creative means more learning, which leads to better auction performance. 

In practice, it far more reliably increases creative production costs than it improves results – and often benefits agencies more than advertisers.

Creative volume only helps when ad platforms receive enough high-quality conversion signals. 

Without those signals, more ads simply mean more assets to rotate. The AI has nothing meaningful to learn from.

Andromeda generated significant attention in 2025, and it gave marketers a new term to rally around. 

In reality, Andromeda is one component of Meta’s ad retrieval system:

  • “This stage [Andromeda] is tasked with selecting ads from tens of millions of ad candidates into a few thousand relevant ad candidates.”

That positioning coincided with Meta’s broader pivot from the metaverse narrative to AI. It worked. 

But it also led some teams to conclude that aggressive creative diversification was now required – more hooks, more formats, more variations, increasingly produced with generative AI.

Similar to Google Ads’ push around automated bidding, broad match, and responsive search ads, Andromeda has become a convenient justification for adopting Advantage+ targeting and Advantage+ creative. 

Those approaches can perform well in the right conditions. They are not universally reliable.

Get the newsletter search marketers rely on.


How to fix this

Creative diversification helps platforms match messages to people and contexts. That value is real. It is also not new. The same fundamentals still apply:

  • Creative testing requires a strategy. Testing without intent wastes resources.
  • Measurement must be planned in advance. Otherwise you’re setting yourself up for failure.
  • Business-level KPIs need to exist in sufficient volume to matter.

This myth breaks down most clearly when resources are limited – budget, skills, or time. In those cases, platforms often rotate ads with little signal-driven direction.

When resources are constrained, CRO is a better use of your resources:

  • Review tracking. More tracked conversions improve performance.
  • Improve the customer journey to increase conversion rates and signal volume.
  • Map higher-margin products to support more efficient spend.
  • Test new channels or networks using budget saved from excessive creative production.

The pattern is consistent. Creative scale follows signal scale, not the other way around.

Myth 3: GA4 and attribution are flawed, but marketing mix modeling will provide clarity

Can you think of 10 marketers who believe GA4 is a good tool? Probably not. 

That alone speaks to how poorly Google handled the rollout. 

As a result, more clients now say the same thing: GA4 does not align with ad platform data, neither feels trustworthy, and a more “serious” solution must be needed. 

More often than not, that path leads to higher costs and average results. 

Most brands simply do not have the spend, scale, or complexity required for MMM to produce meaningful insight. 

Instead of adding another layer of abstraction, they would be better served by learning to use the tools they already have.

For most brands, the setup looks familiar:

  • Media spend is concentrated across two or three channels at most – typically Google and Meta, with YouTube, LinkedIn, or TikTok as secondary options.
  • The business depends on a recurring but narrow customer base, which creates long-term fragility.
  • Outside that core audience, marketing is barely incremental, if incremental at all.

In those conditions, MMM does not add clarity. It adds abstraction. 

With such a limited channel mix, the focus should remain on fundamentals. 

The challenge is not modeling complexity, but identifying what is actually impactful. 

How to fix this

The priorities below deliver more value than MMM in these scenarios:

  • Differentiate clearly from competitors.
  • Increase margins, even basic budget planning can move the needle.
  • Build a solid data foundation, including tracking, CRO, and conversion pipelines.
  • Diversify channels or ad networks.
  • Lock creative execution to real customer pain points.
  • Fix marketing execution wherever it breaks.

MMM – like any advanced tool – becomes useful once complexity demands it. Not before. 

Used too early, it replaces accountability with abstraction, not insight.

The reality behind the myths

The common thread across these three myths is not AI, creative, or analytics. It is misuse. 

Platforms do exactly what they are asked to do. They optimize against the signals provided, within the constraints of budget and structure.

When business fundamentals break, AI cannot fix the problem. 

2026 is not about chasing the next abstraction. It is about business and ops focus, paired with disciplined execution, to scale profitably.

Read more at Read More

Why copywriting is the new superpower in 2026

Why copywriting is the new superpower in 2026

For the last few years, copywriting has been quietly written off.

Not with outrage. Not with ceremony.

Just sidelined. Replaced. Automated.

Words – the core material of SEO, landing pages, ads, and persuasion – were demoted during the traffic rush and later the AI gold rush.

Blog posts were generated. Product descriptions were bulked out. Landing pages were templated.

Content teams shrank. Freelancers disappeared. And a convenient narrative emerged to justify it all:

“AI can write now, so writing doesn’t matter anymore.”

Then Google made it worse.

The helpful content update, followed by AI Overviews and conversational search, didn’t just hurt SEO. It hurt the broader web.

It gutted an entire economy built on informational arbitrage – niche blogs, affiliate sites, ad-funded publishers, and content-led SEO businesses that had learned how to monetize curiosity at scale.

Now, large language models are finishing the job. Informational queries are answered directly in search. The click is increasingly optional. Traffic is evaporating.

So yes, on the surface, it sounds mad to say this:

Copywriting is once again becoming the most important skill in digital marketing.

But only if you confuse copywriting with the thing that just died.

AI didn’t kill copywriting

What AI destroyed was not persuasion. 

It destroyed low-grade informational publishing – content that existed to intercept search demand, not to change decisions.

  • “How to” posts.
  • “Best tools for” roundups.
  • Explainers written for algorithms, not people.

LLMs are exceptionally good at this kind of work because it never required judgment. It required:

  • Synthesis. 
  • Summarization. 
  • Pattern matching. 
  • Compression.

That’s exactly what LLMs do best.

This content was designed to intercept purchase decisions by giving users something else to click before buying, often with the hope that a cookie would track the stop in the journey and reward the page for “influencing” the buyer journey.

That influence was rewarded either through analytics for the SEO team or through an affiliate’s bank account.

But persuasion – real persuasion – has never worked like that.

Persuasion requires:

  • A defined audience.
  • A clearly articulated problem.
  • A credible solution.
  • A deliberate attempt to influence choice.

Most SEO copy never attempted any of this. It aimed to rank, not to convert.

So when people say “AI killed copywriting,” what they really mean is this: AI exposed how little real copywriting was being done in the first place.

And that matters, because the environment we’re moving into makes persuasion more important, not less.

Dig deeper: SEO copywriting: 5 pillars for ranking and relevance

GEO isn’t about rankings

Traditional search engines forced users to translate their problems into keywords.

Someone didn’t search for “I’m an 18-year-old who’s just passed my test and needs insurance without being ripped off.” They typed [cheap car insurance] and hoped Google would serve the best results.

This created a monopoly in SEO. Those who could spend the most on links usually won once a semi-decent landing page was written.

It also created a sea of sameness, with most ranking websites saying exactly the same thing.

LLMs reverse this process. They:

  • Start with the problem.
  • Understand context, constraints, and intent. 
  • Decide which suppliers are most relevant.

That distinction is everything.

LLMs are not ranking pages. Instead, they seek and select the best solutions to solve users’ problems.

And selection depends on one thing above all else – positioning.

Not “position on Google,” but strategic positioning.

  • Who are you for?
  • What problem do you solve?
  • Why are you a better or different choice than the alternatives?

If an LLM cannot clearly answer those questions from your website and third-party information, you will not be recommended, no matter how many backlinks you have or how “authoritative” your content once looked.

This is why copywriting suddenly sits at the center of SEO’s future.

Dig deeper: The new SEO imperative: Building your brand

From SEO to GEO: Availability beats visibility

Search engine optimization was about visibility.

Generative engine optimization is about AI availability.

Availability means increasing the likelihood that your business will be surfaced in a buying situation.

That depends on whether your relevance is legible.

Most businesses still describe themselves in static, categorical terms:

  • “We’re an SEO agency in Manchester.”
  • “We’re solicitors in London.”
  • “We’re an insurance provider.”

These descriptions tell you what the business is. 

They do not tell you what problem it solves or for whom it solves that problem. They are catchall descriptors for a world where humans use search engines.

This is where most companies miss the opportunity in front of them.

The vast majority of “it’s just SEO” advice centers on entities and semantics. 

The tactics suggested for AI SEO are largely the same as traditional SEO: 

  • Create a topical map.
  • Publish topical content at scale.
  • Build links.

This is why many SEOs have defaulted to the “it’s just SEO” position.

If your lens is meaning, topics, context, and relationships, everything looks like SEO.

In contrast, the world in which copywriters and PRs operate looks very different.

Copywriters and PRs think in terms of problems, solutions, and sales.

All of this stems from brand positioning.

Positioning is not a fixed asset

A strategic position is a viable combination of:

  • Who you target.
  • What you offer.
  • How your product or service delivers it

Change any one of those, and you have a new position.

Most firms treat their current position as fixed. 

They accept the rules of the category and pour their effort into incremental improvement, competing with the same rivals, for the same customers, in the same way.

LLMs quietly remove that constraint.

If you genuinely solve problems – and most established businesses do – there is no reason to limit yourself to a single inherited position simply because that’s how the category has historically been defined.

No position remains unique forever. Competitors copy attractive positions relentlessly. 

The only sustainable advantage is the ability to continually identify and colonize new ones.

This doesn’t mean becoming everything to everyone. Overextension dilutes brands.

It means being honest and explicit about the problems you already solve well.

This is something copywriters understand well. 

A good business or marketing strategist can help uncover new positions in the market, and a good copywriter can help articulate them on landing pages.

This is a key shift from semantic SEO to GEO.

You want LLMs to recommend your business to solve those problems.

Get the newsletter search marketers rely on.


From SEOs’ ‘what we are’ to GEOs’ ‘what problem we solve’

Take insurance as a simple example.

A large insurer may technically offer “car insurance.” But the problems faced by:

  • An 18-year-old new driver.
  • A parent insuring a second family car.
  • A courier using a vehicle for work.
  • Are completely different.

Historically, these distinctions were collapsed into broad keywords because that’s how search worked. 

LLMs don’t behave like that. They start with the user problem to be solved.

If you are well placed to solve a specific use case, it makes strategic sense to articulate that explicitly, even if no one ever typed that exact phrase into Google.

A helpful way to think about this is as a padlock.

Your business can be unlocked by many different combinations. 

Each combination represents a different problem, for a different person, solved in a particular way.

If you advertise only one combination, you artificially restrict your AI availability.

Have you ever had a customer say, “We didn’t know you offered that?”

Now you have the chance to serve more people as individuals.

Essentially, this makes one business suitable for more problems.

You aren’t just a solicitor in Manchester.

You’re a solicitor who solves X by Y.

You’re a solicitor for X with a Y problem.

The list could be endless.

Why copywriting becomes infrastructure again

This is where copywriting returns to its original job.

Good copywriting has always been about creating a direct relationship with a prospect, framing the problem correctly, intensifying it, and making the case that you are the best place to solve it.

That logic hasn’t changed.

What has changed is that the audience has expanded.

You now have to persuade:

  • A human decision-maker.
  • A LLM acting as a recommender.

Both require the same thing: clarity.

You must be explicit about:

  • The problem you solve.
  • Who you solve it for.
  • How you solve it.
  • Why your solution works.

You must also support those claims with evidence.

This is not new thinking. It comes straight out of classic direct marketing.

Drayton Bird defined direct marketing as the creation and exploitation of a direct relationship between you and an individual prospect. 

Eugene Schwartz spent his career explaining that persuasion is not accidental – benefits must be clear, claims must be demonstrated, and relevance must be immediate.

The web environment made it possible to forget these fundamentals for a while.

AI brings them back.

Dig deeper: Why ‘it’s just SEO’ misses the mark in the era of AI SEO

Less traffic doesn’t mean less performance

Traffic is going to fall.

Informational traffic is being stripped out of the system.

Traffic only became a problem when it stopped being a measure and became a target. 

Once that happened, it ceased to be useful. Volume replaced outcomes. Movement replaced progress.

In an AI-mediated world, fewer clicks does not mean less opportunity.

It means less irrelevant traffic.

When GEO and positioning-led copy work, you see:

  • Traffic landing on revenue-generating pages.
  • Brand-page visits from pre-qualified prospects.
  • Fewer exploratory visits and more decisive ones

No one can buy from you if they never reach your site. Traffic still matters, but only traffic with intent.

In this environment, traffic stops being a vanity metric and becomes meaningful again.

Every click has a purpose.

What measurement looks like now

The North Star is no longer sessions. It is commercial interaction.

The questions that matter are:

  • How many clicks did we get to revenue-driving pages this month versus last?
  • How many of those visits turned into real conversations?
  • Is branded demand increasing as our positioning becomes clearer?
  • Are lead quality and close rates improving, even as traffic falls?

Share of search still has relevance – particularly brand share – but it must be interpreted differently when the interface doesn’t always click through.

AI attribution is messy and imperfect. Anyone claiming otherwise is lying. But signals already exist:

  • Prospects saying, “ChatGPT recommended you.”
  • Sales calls referencing AI tools.
  • Brand searches rising without content expansion.
  • Direct traffic increasing alongside reduced informational content

These are directional indicators. And they are enough.

The real shift SEO needs to make

For a decade, SEO rewarded people who were good at publishing.

The next decade will reward people who are good at positioning.

That means:

  • Fewer pages, but sharper ones.
  • Less information, more persuasion.
  • Fewer visitors, higher intent.

It means treating your website not as a library, but as a set of sales letters, each one earning its place by clearly solving a problem for a defined audience.

This is not the death of SEO.

SEO is growing up.

The reality nobody wants, but everyone needs

Copywriting didn’t die.

Those spending a fortune on Facebook ads embraced copywriting. Those selling SEO went down the route of traffic chasing.

The two worlds had different values.

  • The ad crowd embraced copy.
  • The SEO crowd disowned it.

One valued conversion. The other valued traffic.

We are entering a world with less traffic, fewer clicks, and an intelligent intermediary between you and the buyer.

That makes clarity a weapon. That makes good copy a weapon.

In 2026, the brands that win will not be the ones with the most content.

They will be the brands that return to the basics of good copy and PR.

The information era of SEO is over.

It’s time to get back to marketing.

Read more at Read More

Not all MMM tools are equal: Meridian, Robyn, Orbit, and Prophet explained

Not all MMM tools are equal: Meridian, Robyn, Orbit, and Prophet explained

Marketing mix modeling (MMM) has shifted from an enterprise luxury to an essential measurement tool. 

Tech giants like Google, Meta, and Uber have released powerful open-source MMM frameworks that anyone can use for free. 

The challenge is understanding which tool actually solves your problem and which require a PhD in statistics to implement.

Open-source MMM tools are often grouped together but solve different problems

The landscape can be confusing because these tools serve fundamentally different purposes despite being mentioned together. 

Google’s Meridian and Meta’s Robyn are complete, production-ready MMM frameworks that take your marketing data and deliver actionable budget recommendations. 

They include everything needed: 

  • Data transformations that model advertising decay.
  • Saturation curves that capture diminishing returns.
  • Visualization dashboards and budget optimizers that recommend spend allocation.

Uber’s Orbit and Facebook’s Prophet occupy different niches. 

Orbit is a time-series forecasting library that can be adapted for MMM, but it requires months of custom development to build MMM-specific features. 

Prophet is a forecasting component used within other frameworks, not a standalone MMM solution. 

Think of it like transportation: 

  • Meridian and Robyn are complete cars you can drive today. 
  • Orbit is a high-performance engine that requires you to build the transmission, body, and wheels. 
  • Prophet is the GPS system that goes inside the car.

Dig deeper: Marketing attribution models: The pros and cons

Robyn: The accessible powerhouse

Meta built Robyn specifically to democratize MMM through automation and accessibility. 

The framework uses machine learning to handle model building that traditionally required weeks of expert tuning. 

Upload your data, specify channels, and Robyn’s evolutionary algorithms explore thousands of configurations automatically.

What makes Robyn distinctive is its approach to model selection. 

Rather than claiming one “correct” model, it produces multiple high-quality solutions that show trade-offs between them. 

Some fit historical data better but recommend dramatic budget changes. 

Others have slightly lower accuracy but suggest more conservative shifts. 

Robyn presents this range, allowing decisions based on business context and risk tolerance.

Budget allocation with Robyn

The framework also excels at incorporating real-world experimental results. 

If you have run geo-holdout tests or lift studies, you can calibrate Robyn using those results. 

This grounds statistical analysis in experiments rather than pure correlation, improving accuracy and giving skeptical executives evidence to trust the outputs.

However, Robyn assumes marketing performance remains constant throughout the analysis period. 

In practice, algorithm updates, competitive changes, and optimization efforts mean channel effectiveness often varies over time.

Meridian: The statistical heavyweight

Meridian represents Google’s Bayesian causal inference approach to MMM. 

Unlike Robyn’s pragmatic optimization, Meridian models the mechanisms behind advertising effects, including decay, saturation, and confounding variables. 

This theoretical rigor allows Meridian to better answer, “What would happen if we changed budget allocation?” rather than simply, “What patterns existed in the past?”

Its standout capability is hierarchical, geo-level modeling. 

While most MMMs operate at a national level, Meridian can model more than 50 geographic locations simultaneously using hierarchical structures that share information across regions. 

Advertising may perform well in urban coastal markets but struggle in rural areas. 

National models average these differences away. 

Meridian’s geo-level approach identifies regional variation and delivers market-specific recommendations that national models can’t.

Meridian insights on channel contribution

Another distinguishing feature is its paid search methodology, which addresses a fundamental challenge: when users search for your brand, is that demand driven by advertising or independent of it? 

Meridian uses Google query volume data as a confounding variable to separate organic brand interest from paid search effects. 

If brand searches spike because of viral news or word-of-mouth, Meridian isolates that activity from the impact of search ads.

The technical complexity, however, is significant. 

Meridian requires deep knowledge of Bayesian statistics, comfort with Python, and access to GPU infrastructure. 

The documentation assumes a level of statistical literacy most marketing teams lack. 

Concepts such as MCMC sampling, convergence diagnostics, and posterior predictive checks typically require graduate-level training.

Dig deeper: How Bayesian testing lets Google measure incrementality with $5,000

Get the newsletter search marketers rely on.


Uber Orbit: The time-varying specialist

Orbit is not technically an MMM tool. 

It’s a time-series forecasting library from Uber with a notable feature: Bayesian time-varying coefficients, or BTVC, which address a fundamental MMM challenge.

Imagine presenting MMM results to your CEO, who asks, “This assumes Facebook ads had the same ROI in January and December? But iOS 14 hit in April, and we spent months recovering. How can one number represent the whole year?” 

That is the credibility-breaking moment practitioners fear because it exposes a simplifying assumption executives correctly recognize as unrealistic.

Traditional MMM frameworks assign one coefficient per channel for the entire analysis period, producing a single ROI or effectiveness estimate. 

  • For stable channels like TV, this can work. 
  • For dynamic digital channels, where teams constantly optimize, respond to algorithm changes, and face shifting competition, assuming static performance is clearly flawed. 

Orbit’s BTVC allows channel effectiveness to change week by week. 

Facebook ROI in January can differ from December, while the model keeps estimates stable unless the data shows clear evidence of real change.

The reality, however, is that while time-varying coefficients are powerful, Orbit lacks the other components required for a complete MMM solution. 

Orbit makes sense only for data science teams building proprietary frameworks that require advanced capabilities and have the resources for significant custom development. 

For most organizations, the cost-benefit tradeoff does not justify that investment. 

Teams are better served using Robyn or Meridian while acknowledging their limitations, or working with commercial MMM vendors that have already built time-varying capabilities into production-ready systems.

Facebook Prophet: The misunderstood component

Prophet is Meta’s time-series forecasting tool. 

It’s highly effective at its intended purpose but is often misrepresented as an MMM solution, which it is not.

Prophet decomposes time-series data into trend, seasonality, and holiday effects. 

It answers questions, such as:

  • “What will our revenue be next quarter?” 
  • “How do Black Friday spikes affect baseline performance?” 

This is forecasting, or predicting future values based on historical patterns, which is fundamentally different from attribution. 

Prophet can’t identify which marketing channels drove results or provide guidance on budget optimization. 

It detects patterns but has no concept of marketing cause and effect.

Prophet’s primary role is as a preprocessing component within larger systems. 

Robyn uses Prophet to remove seasonal patterns and holiday effects before applying regression to isolate media impact. 

Revenue often rises in December because of holiday shopping rather than advertising. 

Prophet identifies and removes that seasonal effect, making it easier for regression models to detect true media impact.

This preprocessing is valuable, but Prophet addresses only one part of the overall attribution problem. 

Marketing teams should use Prophet for standalone KPI forecasting or as a component within custom MMM frameworks, not as a complete attribution or budget optimization solution.

Dig deeper: MTA vs. MMM: Which marketing attribution model is right for you?

Making the right choice for your team

Making the right choice for your team

Choosing between these tools requires an honest assessment of your organization’s capabilities, resources, and needs. 

  • Do you have data scientists comfortable with Bayesian statistics and complex Python? 
  • Or marketing analysts whose statistical training ended with basic regression? 

The answer determines which tools are viable options and which are aspirational.

For about 80% of organizations, Meta’s Robyn is the right choice. 

This includes:

  • Teams without deep data science resources but still need rigorous MMM insights.
  • Digital-heavy advertisers seeking attribution without lengthy implementations. 
  • Organizations that require insights in weeks rather than quarters. 

The learning curve is manageable, implementation takes weeks rather than months, and outputs are presentation-ready. 

A large, active user community also shares solutions when challenges arise.

Google’s Meridian suits:

  • Small and midsize businesses and enterprise organizations with dedicated data science teams comfortable working in Bayesian frameworks. 
  • Multi-regional operations where geo-level insights would meaningfully influence budget decisions.
  • Complex paid search programs requiring more precise attribution.
  • Stakeholders who prioritize causal inference over pragmatic correlations can justify Meridian’s added complexity.

Uber Orbit is appropriate only for data science teams building proprietary frameworks with requirements that Robyn and Meridian can’t meet. 

The opportunity cost of spending months on custom infrastructure rather than using existing tools is substantial unless proprietary measurement itself provides a competitive advantage. 

Facebook Prophet should be used for KPI forecasting or as a preprocessing component within larger systems, never as a complete attribution solution.

Matching MMM tools to real-world team capabilities

The most advanced tool delivers little value if it can’t be implemented effectively. 

A well-executed Robyn implementation running consistently provides more value than an abandoned Meridian project that never progressed beyond a pilot. 

Tools should be chosen based on what teams can realistically use and maintain, not on the most impressive feature set.

For most marketing teams, Robyn and Meridian represent pragmatic choices that balance performance with accessibility. 

Automation handles much of the statistical work, allowing analysts to focus on insights rather than debugging code. 

Strong community support and documentation reduce friction, and teams can move from zero to actionable insights in weeks instead of months, which matters when executives want answers quickly.

For enterprises with substantial technical resources and multi-regional operations, Google Meridian can deliver returns through more reliable causal estimates and geo-level granularity that materially improve budget allocation. 

The investment in infrastructure, expertise, and implementation time is significant, but at a sufficient scale, better decision-making can justify the cost.

Uber Orbit offers advanced capabilities for organizations that truly need time-varying performance measurement and have the resources to build complete MMM systems around it. 

For most teams, commercial vendors that have already incorporated time-varying capabilities into production-ready platforms are more cost-effective than extended custom development.

These open-source frameworks have made marketing measurement accessible beyond Fortune 500 companies. 

The priority is choosing the tool that fits current capabilities, implementing it well to earn stakeholder trust, and using insights to make better decisions. 

Competitive advantage comes from allocating budgets more effectively and faster than competitors, not from maintaining a technically impressive system that is too complex to sustain.

Dig deeper: How to avoid marketing mix modeling mistakes that derail results

Read more at Read More

Google to require separate product IDs for multi-channel items

Google Shopping Ads - Google Ads

Starting in March 2026, Google Merchant Center will enforce a new system for multi-channel products — items sold both online and in physical stores — requiring advertisers to use separate product IDs when those products differ by channel.

What’s changing. Under the new approach, online product attributes will become the default. If a product’s in-store details differ, advertisers will need to create a second version with a distinct product ID and manage it independently in their feeds.

What advertisers should do. Google has started emailing affected accounts, flagging products that need updates ahead of the March deadline. Retailers should review their product data feeds now to ensure online and in-store items are properly segmented — especially if they rely on Local Inventory Ads or sell across multiple Google surfaces.

Why we care. Many retailers currently manage online and in-store versions of the same product under a single ID. Google’s update changes that assumption, pushing advertisers to explicitly separate products when attributes like price, availability, or condition aren’t identical.

The big picture. This update gives Google cleaner, more consistent product data across channels, but shifts more feed management responsibility onto advertisers — particularly large retailers with complex inventories.

First seen. The update and news of Google’s comms was first mentioned by PPC News Feed founder Hana Kobzová.

Bottom line. If your online and in-store products aren’t truly identical, Google will soon require you to treat them as separate items, or risk issues with visibility and eligibility.

Dig Deeper. Update of multi-channel product system from Google.

Read more at Read More

Google to allow Prediction Markets ads under strict rules

Why phrase match is losing ground to broad match in Google Ads

Google is updating its advertising policies to allow ads for Prediction Markets in the U.S. starting January 21st — but only for federally regulated entities.

Who qualifies. Eligibility is limited to entities authorized by the Commodity Futures Trading Commission (CFTC) as Designated Contract Markets (DCMs) whose primary business is listing exchange-listed event contracts, or brokerages registered with the National Futures Association (NFA) that offer access to products listed by qualifying DCMs. Advertisers must also apply for Google certification to run ads in the U.S.

Why we care. Prediction markets have long been restricted on Google Ads. This change opens a new advertising channel while keeping tight controls around compliance and regulation. The narrow eligibility and certification requirements mean only compliant, federally regulated players can participate, potentially reducing competition. For qualifying advertisers, this offers earlier access to a high-intent audience within a tightly controlled ad environment.

The fine print. All ads, products, and landing pages must comply with applicable local laws, financial regulations, industry standards, and Google Ads policies. The new policy will appear in the Advertising Policies Help Center, with references in the Financial Services and Gambling and Games sections, and is available now for preview.

The big picture. Google is cautiously expanding access for prediction markets by recognizing them as regulated financial products — while continuing to block unregulated platforms.

Bottom line. Prediction market ads are coming to Google, but only for advertisers that meet strict federal and platform-level requirements.

Read more at Read More