ChatGPT, Perplexity and Gemini Are Shaping the First Impression of You: How to Influence the Answer

Search results at least gave you the illusion of control: a bad result could be pushed down to the second page. That doesn't work with AI assistant answers. When someone asks ChatGPT, Perplexity or Google Gemini about you or your company, they see one coherent paragraph that can't be scrolled past or pushed aside. In a piece dated October 13, 2026, TNW calls this answer the new first impression. It also makes the key point: you can influence it.

A caveat about the source: in the portion available to us, TNW gives only the thesis and an introduction, with no studies or figures. What follows is therefore an analysis of the underlying logic of the problem and the practical steps that follow from it. We do not reproduce any specific metrics from the publication.

Where the paragraph comes from

TNW's starting thesis: assistants don't invent a reputation from scratch, they assemble it from what has already been published. For ChatGPT with web search, Perplexity and Gemini, that means working with open sources. Perplexity shows links right next to its answer, Gemini relies on Google's infrastructure, and ChatGPT blends in search results and the knowledge baked in during training. Each system uses its own set of sources, but the principle is the same: the final text reflects what is easiest to find about you and what sounds most convincing.

An unpleasant conclusion follows. If your digital footprint is thin, the answer will be sparse. If it is contradictory, the assistant will smooth the contradiction into a single formulation, and not necessarily in your favor.

Where the answer most often gets distorted

The typical causes are visible from the mechanics of how an answer is assembled:

• an outdated profile or a bio that hasn't been updated in a long time;
• one loud negative piece that dominates everything else;
• namesakes and companies with similar names whose data gets merged with yours;
• discrepancies between your website, social media and directories in dates, job titles and figures;
• no authoritative source of your own: a website with a clear "About us" page.

All of these can be fixed, but not quickly. An assistant has no "delete" button; you have to correct the sources.

Where to start

The first step is an audit. Ask ChatGPT, Perplexity and Gemini the same questions: who this person is, what the company does, what it is known for, what complaints exist, how it differs from competitors. Save the answers with the date. Then, using the links Perplexity shows and the pages that surface in regular search, build a map of the sources that shape the picture.

From there, the work proceeds along three lines:

1. Fix the facts at the original sources. Update your website, your profiles in industry directories and your pages on professional networks. Wording, names and dates should match everywhere. 2. Create what is missing. A clear page about the company or person, a press kit, bylined articles, interviews on authoritative outlets. The assistant needs something to build on. 3. Mark up your data for machines. Structured markup on your site helps systems understand unambiguously who is who and tell you apart from namesakes.

A separate note on what not to do. Attempts to hide instructions for models on a page or to inflate fake mentions are risky: platforms are cracking down on such tricks, and the reputational damage from being exposed outweighs any possible benefit.

Finally, you need monitoring. Assistants' answers change as models and search indexes are updated, so the check is worth repeating regularly, at least once a month, and tracking how the answers shift. For press offices and marketing teams, this is a new routine, comparable to tracking search rankings. Only now the result is measured not by a position in a list, but by what exactly is said in a single paragraph.