As AI commoditizes, what still appreciates? Martech Futurist | August 20, 2026

Tuesday's edition argued that routing is the marketing operations discipline of the 2027 planning cycle, deciding per workflow which mechanism and how much scrutiny a task deserves. That was a question about how you spend inside a workflow.

This week the question moves up to the stack itself. Four items published between August 18 and August 19 converge on where value settles once every vendor can perform the same tasks. Forrester published a market model that separates technology categories by whether AI accelerates, reshapes, or replaces them. Experian published survey data showing consumers pulling back from AI at precise, predictable points. Marketing Week documented one B2B brand rebuilding its search strategy around machine retrieval and reported the result.

Read together, they describe a stack where the capability layer commoditizes and the conditions underneath it appreciate.

Read the Forrester model as a budget instruction for your own stack

Forrester introduced its AI Disruption Model on August 19, applying it to 17 technology and service categories spanning more than 200 markets. Craig Le Clair and Ted Schadler, both VP and principal analyst at Forrester, built it on nine drivers including AI substitutability, labor intensity, support for agentic workloads, commercial model, regulatory friction, and switching costs. The model sorts each market into disrupted, neutral, contested, or accelerated.

Three categories come out positioned for clear growth: infrastructure, data and AI, and identity, access, and network security. Le Clair states the finding plainly: "Every technology and service market is facing an AI overhaul."He adds that the benefits arrive unevenly, and that categories outside those three will have to adapt.

Marketing technology sits in the middle group. Forrester's analysis puts business applications, governance and compliance tools, process automation, customer experience systems, and martech in the reshaped category, held in place by embedded workflows, regulatory requirements, and switching costs even as AI changes how people use them and what they cost. The categories under real pressure are the labor-intensive services: implementation, custom development, creative services, localization, and training.

Schadler frames the practical question for vendors as one of market economics, and points providers toward repricing and redesigning where AI removes manual work. Turn that around and it becomes a buyer's instruction. Your martech vendors are about to reprice. The line items that hold value are the ones your organization would have to rebuild from scratch if the vendor disappeared: your data model, your identity graph, your consent records, your audit trail.

This maps onto the capability framework I use with clients. The four pillars name what AI does. The layers name what the doing requires: interface, memory and context, trust and verification, identity and permissions. Forrester's growth categories are the layers. Its pressured categories are the pillars, delivered as labor.

It reminds me of what Michelle Boockoff-Bajdek, CMO from Sitecore, said when I interviewed her on The Agile Brand podcast: "We're seeing a critical reframe, moving from understanding the customer as a static identity or a segment to understanding their dynamic intent through real-time signals. This requires bringing together data from clicks, searches, and other interactions to create a living picture of the customer at that exact moment." The living picture is the asset. The tool that renders it is the commodity.

Assume your customers withhold trust exactly where your margin sits‍ ‍

Experian released its 11th annual Identity and Fraud Report on August 19, based on April 2026 surveys conducted with HarrisX covering more than 2,000 US consumers and more than 200 US businesses.

The gradient in the data is the finding. Thirty-one percent of consumers have already used AI tools to shop and transact online, and another 23% would consider it. Comfort drops to 21% for completing travel purchases and 17% for financial services decisions. More than half report concern about AI-enabled scams.

That curve is the thesis of my capability framework rendered as consumer behavior. AI fades into the background for low-stakes work and stays explicitly named where being wrong carries cost. Consumers are drawing that line themselves, and they are drawing it at the exact point where transaction value concentrates. The Agile Brand Guide covers this pattern as the fade, or uneven disappearance.

Experian's identity findings show the lever. Seventy-one percent of consumers say it matters that businesses recognize them accurately online, close to half report greater trust in organizations that manage this without repeated authentication, and 84% will complete additional security steps when fraud prevention requires it. Kathleen Peters, Chief Innovation Officer at Experian, ties the two together: <cite index="64-1">"the organizations that will succeed will be the ones that make trust visible."</cite>

Work the math on a travel brand. Say you run 100,000 online bookings a quarter at an average value of $1,200. Apply the Experian gradient and roughly 31,000 of those buyers arrive having used an agent to research. About 21,000 would let the agent complete the purchase. The remaining 10,000 hand the transaction back to a human at the decision point, and that handoff governs $12 million a quarter. If your checkout forces those returning buyers to re-authenticate and 15% abandon there, you lose 1,500 bookings and $1.8 million in a quarter. The identity layer converts that abandonment, and it costs a fraction of $1.8 million to build.

Booking value and abandonment rate in that example are mine, not Experian's. Swap your own numbers in. The structure holds at any scale: the higher your average transaction, the more the identity layer pays.

It reminds me of what Christian Nelissen, Chief Data and Analytics Officer from National Australia Bank, said when I interviewed him on The Agile Brand podcast: "We have a rule… don't be creepy. The ability to use our customers' data to help them is based entirely on the trust that they have in us. That's more valuable than any given sale." Agile Brand Principle 4 says the same thing in shorter form: respect customers and their data. Experian just priced it.

Make your brand legible to retrieval before you fund more persuasion

Marketing Week reported on August 18 that Akamai rebuilt its search strategy around large language models and recorded a 133% increase in ChatGPT visibility against competitors. The framing in the piece treats LLMs as "VIP customers," a category of visitor the brand structures its content to serve directly.

The mechanism deserves attention more than the number. Akamai did not buy visibility. The team changed how information about the company is structured so retrieval systems could parse and cite it. That work lives in the content and data layer, which is exactly where Forrester's model puts the growth.

For B2B, the timing compounds. An LLM shapes a shortlist before a buyer contacts anyone. Illegible product data means the model reconstructs your positioning from third-party sources, and you inherit whatever it finds. The Agile Brand Guide tracks this as identity legibility within the brand visibility framework for agentic commerce.

One caution on the Akamai figure. Visibility measured against competitors on a set of prompts is a relative index on a sampled query set. Ask what prompts, how many, and over what period before you set a target against it.

Keep funding the human signals machines cannot synthesize

Marketing Week also published analysis on August 19 on how B2B influencer strategy is changing as LLMs take a larger role in purchase decisions, alongside digital communities and peer advocacy.

This is the counterweight to the previous section, and it belongs in the same budget conversation. Structured data earns retrieval. Practitioner testimony, community credibility, and peer advocacy generate the source material worth retrieving. A model summarizing a category draws on what people published about you. Fund the second, and the first improves without further optimization.

Agile Brand Principles 5 and 6 hold up well here: focus on the relationship over the transaction, and treat dialogue as something to learn from. Both remain outside what a retrieval system can manufacture.

Featured insights

Forrester AI Disruption Model, August 19.Le Clair and Schadler sort more than 200 markets by AI exposure and place martech among the categories AI reshapes while infrastructure, data, and identity security accelerate. Practitioner takeaway: audit your renewals for line items that price capability and separate them from line items that hold your data model, identity graph, and audit trail. Expect the first group to reprice downward and the second to carry the leverage.

Experian Identity and Fraud Report, August 19.Consumer comfort with AI falls from 31% for general shopping to 21% for travel purchases to 17% for financial decisions, while 71% want accurate recognition and 84% accept extra security steps when fraud is the reason. Practitioner takeaway: map your journeys against that gradient and find the point where customers take the transaction back from the agent. Instrument that handoff before you invest anywhere else in the funnel.

Akamai search rebuild, Marketing Week, August 18.A B2B cybersecurity firm restructured its content for machine retrieval and reported a 133% relative gain in ChatGPT visibility. Practitioner takeaway: run twenty category prompts your buyers would plausibly type, record what the models say about you, and treat every factual error as a data defect with an owner and a due date.

B2B influencer strategy, Marketing Week, August 19.Creator and peer-advocacy programs are being rebuilt for a market where LLMs shape shortlists ahead of vendor contact. Practitioner takeaway: brief your creators and analysts to publish specifics, named use cases, measurable outcomes, and stated limitations, since those are the passages models quote.‍ ‍

Key takeaways‍ ‍

Budget the layers your vendors cannot repossess. Forrester's model separates markets AI accelerates from markets it reshapes. Your data model, identity resolution, consent records, and audit trail sit in the first group. Feature parity sits in the second.

Instrument the point where trust breaks. Experian's gradient from 31% to 21% to 17% marks where customers stop delegating. That point sits near your highest-value transactions, and it is measurable in your own funnel this quarter.

Treat identity as revenue infrastructure. Close to half of consumers report more trust in businesses that recognize them without repeated authentication. In a high-value category that single capability converts abandonment worth more than the build.

Fund the signals models cannot generate. Structured data earns citation, and human testimony supplies what gets cited. Both belong in the same line of the 2027 plan.

A closing note

I have sold two companies and acquired three while growing others. Diligence teaches you quickly which line items a buyer actually pays for. Nobody paid us a premium for our tooling. Buyers paid for the customer data, the contracts, the documented processes, and the operational knowledge that would have taken them years to rebuild. The software we ran got a polite look and a discount.

This week's research describes the same sorting. Le Clair and Schadler put the growth in infrastructure, data, and identity security. Experian's survey work locates the point where customers stop delegating. The Akamai team earned visibility by restructuring data. All three describe conditions.

Across more than 900 conversations with senior leaders on the podcast, the organizations pulling ahead read their own stack the way an acquirer would. They know which line items they would have to rebuild from scratch and which they could replace inside a quarter. They fund the first group accordingly.

Open your 2027 plan and find the largest line item that buys a capability your competitors can buy next quarter. Move part of it into the layer underneath.