Take5.com recently went live with almost 2,500 unique URLs, built from twelve templates. That ratio came up early when Matthew Flanagan sat down with Sean Winter, Head of Value Engineering at Contentful, for the first episode of Elevate Digital's new LinkedIn Virtual Event, Robots, Crawlers, and Humans. It's also the clearest evidence for the argument the two spent the next hour building out: a website doesn't have one audience anymore.
The name for the series traces back to an internal strategy session. Lindsay Ogden, one of Elevate's enterprise architects, was presenting the plan for the Take 5 rebuild to the Driven Brands team and laid out the three groups the new site had to serve: AI language models, search crawlers, and the person whose oil light just came on. Flanagan heard "robots, crawlers, and humans," and the framing stuck around long enough to become a working thesis for both a series and how Elevate approaches every rebuild now.
Take 5's business depends on showing up when someone asks their phone where the nearest oil change is. That's the test now: not "are we ranking," but "does ChatGPT, Gemini, or Perplexity know we exist when it matters." Winter said Contentful customers increasingly show up describing the same symptom: strong SEO on paper, invisible in answer engines.
Contentful built two tools to diagnose it. The first, Page Scope, came out of an internal hackathon and scored a domain based on how well its content structure showed up across language models. It became useful enough that Contentful built a full product around it, Palmata, which now audits AI discoverability and gives specific recommendations, not a roadmap for some future content model but steps a team can take today.
For Take 5, that meant a unified schema, a canonical AI feed, and server-side rendering across all 2,500 pages, plus one handwritten, roughly 70-line file that plainly states what Take 5 is, which service pages matter most, and where its locations are. No answer engine has confirmed that a file like this moves rankings. With the site recently live, holding search position with no drop-off is the first thing Elevate will be watching for.
Duplicate content, weak schema, inconsistent metadata, content trapped in page templates: these were always content architecture problems wearing an SEO label. Answer engines just raised the cost of ignoring them, because now they're not only indexing a page–they're extracting facts, comparing sources, and deciding which brands are credible enough to cite.
That reframes an old argument Elevate runs into constantly: build for search first, or build for the customer first? The SEO-first playbook (keyword density, tag stuffing, content written for a crawler instead of a person) was never Elevate's take. A customer-first experience, backed by clean structure underneath it, is what both search engines and AI systems now reward. For Take 5, that showed up as a specific punch list: a unified entity graph, consistent page structures, and server-side rendering for every fact the business needed a machine to find.
Flanagan put the shift in blunter terms: "Site traffic is down. SERP is down. Asking questions, which is more human anyway, is up. And people who are doing it right are seeing conversion go up."
The third audience is the one the site was originally built for, and it's easy to lose sight of once the first two get complicated. Winter described her clearly: she's not browsing, she's not comparing oil change philosophies, her light has been on for a while, and she needs the nearest location, a short wait, and a reason to act now.
"For the human audience, relevancy is about the next best action," Winter said. Contentful's model for this is to stop treating content as static pages and start treating it as reusable pieces: services, locations, offers, proof points, calls to action, assembled based on where someone is and what they're trying to do. As Winter put it, "AI may introduce the brand, but the site still has to provide the experience, the proof, and a simple way to turn discovery into an actual relationship."
Elevate built the Take 5 site around that from the start rather than adding it later. Feature flags for testing live in the foundation, not bolted on after launch. Analytics stream into BigQuery from day one so personalization has something real to learn from.
Localization for Take 5's US Hispanic customers goes past translation, too. Contentful's models handle the initial pass, a person reviews it, and the plan is to test imagery and messaging separately over time, because that customer's path to the shop isn't identical to the general-market one.
One shift both Flanagan and Winter noted without much prompting: the platform decision now sits at the friction point between marketing and IT, driven by a straightforward question: why are our engineers spending hours adjusting a page instead of doing anything strategic? Marketing is chasing revenue and speed; IT is chasing lower operating costs and fewer support tickets. Those goals aren't in tension, but they get expressed in different language, and a lot of stalled platform decisions come down to nobody translating between the two.
There's a related pattern worth naming: companies forced into a second migration two or three years after the first one, often because a "free" platform bundled into a larger enterprise agreement turned out to have real costs once a team tried to build on it. Picking harder next time isn't the answer. What matters is choosing infrastructure flexible enough that the next digital transformation isn't a transformation at all, just an update.
Winter closed with two questions he thinks belong in front of any CMO or CIO before they sign anything:
That second question is the one Take 5's rebuild was built to answer, and the recent launch is the first real test of it. The more interesting answer comes later, once enough of the customer journey (search, site visit, coupon, store visit) connects well enough to show which audience is filling the bay with cars getting serviced on time.
What does the "three audiences" framework mean for website content? It means a website's content now has to work for three distinct readers at once: AI systems and answer engines, search crawlers, and the human visitor. Each one reads content differently, so the underlying structure, not just the page design, has to serve all three without creating separate versions of the same content.
How do you structure a website so AI answer engines can find and cite it? Start with unified schema, consistent metadata, and server-side rendering so machines can read the facts on a page as easily as a person can. Tools like Contentful's Palmata can audit a domain and flag specific improvements, and some brands are pairing that with a short, plain-language file describing what the business is and where its most important pages live.
Should a website be designed for SEO first or for the customer first? Customer-first, with structured content underneath it. Search engines and AI answer engines are both now rewarding sites that are genuinely useful to a person, not just optimized for a crawler, which means a strong customer experience and strong machine-readability come from the same underlying content architecture.
Why are IT teams increasingly leading CMS and platform decisions? Operating cost. IT organizations are watching engineers spend hours on manual page updates and evaluating whether a different platform, often a headless CMS, could reduce that dependency while giving marketing more room to move independently.
What should a company ask before choosing its next content platform? Two things: what the business is trying to accomplish, defined without reference to the technology itself, and whether the platform can structure content once and deliver it to every channel that needs it, including channels that don't exist yet.
Considering a Migration to Contentful Headless CMS? Check out our Migration Readiness Assessment
Did you miss the event? The replay will continue to be available on LinkedIn. Follow for more events like this one.