Many organisations still treat the website as the source of truth about what they offer. Upstream signals are seen as distribution channels or brand space, not as places where meaning is formed.
In reality, search results, AI summaries, reviews, help content and third party listings are already telling a story about you before anyone arrives. When that story does not match what people find on site, they notice. So do AI systems that rely on the same sources.
WHO IS THIS NOTE FOR
This note is for product, marketing and digital teams who:
See good ratings or positive upstream activity but fragile on site performance
Suspect that search results and reviews are describing a different product from the one their journeys assume
Have been told that AI summaries of their organisation feel slightly wrong, but are not sure where that comes from
This field note looks at how those disagreements show up, why they are easy to miss from inside the organisation and what to do about them.
There are many ways for the story upstream to slide away from the one you are telling on the site. Some common patterns:
AI systems are trained on all of this. People read it before they ever see your navigation.
Consider a subscription product where:
On your site, the main journey:
From inside the organisation it may feel as if you are being clear. From the outside it feels like two stories:
Even if you are technically compliant, the mismatch erodes trust and increases support load.
EXAMPLE:
A team selling a subscription product saw strong reviews that praised flexibility and “cancel any time”, and an AI summary that repeated those phrases. On site, the main journey hid cancellation details in legal copy and footer links.
Support tickets about “misleading terms” kept rising even after minor copy tweaks. The problem was not that the information was missing. The problem was that the upstream story and the on site journey were telling two different versions of the same promise.
This kind of disagreement rarely announces itself neatly. It tends to appear as:
Inside teams, it can show up as arguments:
Organisations do react to these inconsistencies. The usual responses are not enough on their own.
Teams adjust headings and microcopy on key pages to address complaints, without changing the upstream content that caused the expectations in the first place.
Result:
Reviews and third party comparisons are sometimes left to “brand” or ignored, especially if they are positive overall.
Result:
Content teams respond by adding FAQs, explainer pages and resource hubs.
Result:
Instead of asking “how do we make the site clearer”, it is often more useful to ask:
These questions move the focus from the individual page to the whole system of signals.
A practical approach is to map the journey from the user’s perspective, not from the site map.
For a given task or decision:
The result is not a perfect map, but it gives a concrete view of where the stories diverge.
Once you can see the split, the job of your on site journeys changes.
It is no longer enough for them to be internally coherent. They also need to:
Recognise that people arrive with prior information, not as blank slates.
Reinforce accurate expectations, rather than resetting everything.
Address gaps, misunderstandings and outdated claims directly, with evidence.
In practice that might mean:
This is less glamorous than a full visual redesign. It is usually more effective.
NOTE
When upstream signals and on site journeys disagree, users do not average them. They decide which one they trust. If your site contradicts what they just saw in three independent places, it is unlikely to win the argument.
AI systems that summarise you are blending:
When those inputs disagree, the AI will still produce a confident answer. It may:
Aligning upstream signals and on site journeys is not about feeding the algorithm. It is about creating a consistent, truthful story that both humans and AI systems can work with.
From a Corpus perspective, split signals are a sign that different parts of the system have been allowed to drift.
When we work with teams on this, we typically:
The aim is not to create a perfectly controlled narrative. It is to reduce unnecessary friction and confusion so that attention can go on the quality of the product or service, not on decoding conflicting stories.
