Brittany Hoversten

Let’s talk

Tag: AI in Marketing

  • AI Search Optimization: A Practical Guide to Getting Found

    Ask ChatGPT about your organization. Go ahead, I’ll wait.

    If you’re like most of the leaders I talk with, the answer comes back close but not quite right. An old program name. A service you stopped offering. A description that sounds like your website from five years ago, because that’s the last time anyone wrote it down clearly.

    That gap is what AI search optimization is about. People are asking AI tools the questions they used to type into a search bar, and those tools answer by pulling from pages they can find, read and trust. If your best information isn’t on one of those pages, someone else’s version of you gets used instead.

    The good news is that most of the work is the same fundamentals good marketers already know. The bad news is that a lot of organizations skipped those fundamentals the first time around.

    What AI search optimization actually means

    AI search optimization is the practice of making your content easy for AI-powered search tools to find, understand and cite. That includes Google’s AI Overviews and AI Mode, ChatGPT search, Perplexity, Microsoft Copilot and whatever launches next quarter.

    You’ll see it called GEO, AEO, LLMO and a few other acronyms. I don’t much care what we call it. The job underneath is the same, and it has four parts:

    • Be reachable. The crawlers these tools rely on have to be allowed to read your site.
    • Be clear. Your pages need to state plainly who you are, what you do, where you do it and who it’s for.
    • Be consistent. What your site says should match what your profiles, directories and partners say about you.
    • Be worth citing. Specific, useful answers get pulled into responses. Vague brochure copy doesn’t.

    Notice what’s not on that list: tricks. There’s no secret file and no magic phrasing.

    What the platforms themselves say

    Before you buy a course or a new tool, read the source documentation. It’s short, and it cuts through a lot of noise.

    Google’s guidance on AI features and your website is refreshingly blunt. There are no additional requirements to appear in AI Overviews or AI Mode, and you don’t need new machine-readable files, AI text files or special markup. A page needs to be indexed and eligible to show with a snippet. The rest is ordinary SEO, done well.

    OpenAI’s crawler documentation adds one practical step for ChatGPT search: allow OAI-SearchBot in your robots.txt. OpenAI says sites that opt out of that crawler won’t be shown in ChatGPT search answers. It also keeps the search crawler separate from GPTBot, the one used for training, so you can allow one and block the other.

    That distinction matters for many of the nonprofits and small businesses I work with. Plenty of them blocked “AI bots” as a group at some point, sometimes through a hosting setting nobody remembers turning on. Reasonable instinct. But a blanket block can also take you out of the answers your audience is now reading.

    If you want a refresher on the foundation underneath all of this, start with the SEO basics every marketing leader should own. AI search sits on top of that foundation. It doesn’t replace it.

    Step 1: Make sure the machines can reach you

    This is the unglamorous part, and it’s where I find the most problems. Sit down with whoever manages your website and walk through this list together:

    1. Open yoursite.com/robots.txt and actually read it. Look for rules that block Googlebot, Bingbot or OAI-SearchBot. Decide on purpose which AI crawlers you allow and which you don’t.
    2. Check your host, CDN and security plugins. Some block bot traffic by default, and a friendly robots.txt can’t override a firewall that turns crawlers away.
    3. Verify your site in Google Search Console and Bing Webmaster Tools. Bing’s index feeds Copilot and other tools, so Bing isn’t optional anymore.
    4. Hunt for accidental noindex tags, especially on pages that were staged, redesigned or migrated.
    5. Get important content out of PDFs and images. If your service list lives in a scanned brochure, assume nothing will read it. Put it on a real page in real text.
    6. Fix your internal links. A page nobody links to is hard for anyone, human or crawler, to find.

    None of this is new. That’s the point. A lot of AI search visibility comes down to technical hygiene that got pushed to “someday.”

    Step 2: Write pages that answer real questions

    AI tools answer questions. So the pages that show up in their answers tend to be the pages that answer questions directly.

    I use a simple test. For each important page, can a stranger read the first two sentences and know who this is for, what it is and why it matters? If not, rewrite those two sentences before you touch anything else.

    Structure your pages for answers

    • Use headings that match how people ask. “How much does respite care cost?” beats “Our Approach.”
    • Answer first, then explain. Put the short answer in the first sentence under the heading.
    • Be specific. Name the counties you serve, your hours, who qualifies and what it costs. Specifics are what get quoted.
    • Keep one main topic per page. A page that tries to cover six services usually explains none of them well.
    • Add a short FAQ where it genuinely helps, and make sure any structured data matches what’s visible on the page.

    Say who you are, plainly

    Your About page is doing more work than you think. It should state your full organization name, what you do, where you do it, how long you’ve done it and who leads it. Write it the way you’d want it summarized, because it will be.

    This is also where brand discipline pays off. If one program goes by three different names across your site, don’t be surprised when an AI answer invents a fourth.

    Step 3: Fix the facts about you everywhere else

    AI answers rarely come from your website alone. They draw on directories, review sites, news coverage, partner pages, LinkedIn and business profiles. When those sources disagree, the tools have to guess, and they don’t always guess in your favor.

    So do a consistency pass. Start here:

    • Google Business Profile: name, address, hours, categories, services and description.
    • Bing Places, Apple Business Connect and the major directories in your sector.
    • Your LinkedIn company page and your leaders’ profiles.
    • Partner, funder and chamber sites that describe you. Send them updated copy. Most are glad to get it.
    • Old press releases and event pages that still rank for your name.

    Anyone who has led a rebrand knows this kind of cleanup. It’s tedious. It’s also what makes everything after it work. If you’re curious what that looks like in practice, some of my brand and web work is in my portfolio.

    Step 4: Measure what you can, and be honest about the rest

    Measurement here is imperfect. Say that out loud to your leadership team, then measure what you can:

    • Search Console. Google counts AI Overviews and AI Mode traffic inside the standard Performance report under the Web search type. It isn’t broken out on its own, so watch trends on the specific pages you’ve improved.
    • GA4 referrals. Build an exploration or a custom channel group for referral traffic from AI tools, such as chatgpt.com, perplexity.ai and copilot.microsoft.com. The volume may be small. The visitors are often further along in their decision.
    • A monthly prompt audit. Keep a list of 10 to 20 questions people actually ask about your organization or category. Run them in two or three tools once a month. Log whether you’re mentioned, whether the facts are right and which sources get cited.

    That last one is low-tech, and it’s the most useful of the three. The cited sources tell you exactly which pages and profiles to fix next.

    One more opinion. Don’t respond to AI search by flooding your site with AI-written filler. Thin pages written to catch every possible question make you less trustworthy, not more. Ten clear, accurate, specific pages will do more for you than a hundred generic ones.

    Key takeaways

    • AI search optimization is mostly SEO fundamentals, done consistently and on purpose.
    • Decide which crawlers you allow. Opting out of OAI-SearchBot keeps you out of ChatGPT search answers.
    • Write pages that answer real questions in the first sentence, with real specifics.
    • Make your facts match everywhere they appear, not only on your own site.
    • Run a monthly prompt audit and fix the sources the tools cite.

    FAQ

    Is AI search optimization different from SEO?

    Mostly, no. Google says there are no extra requirements for AI Overviews or AI Mode beyond being indexed and eligible for a snippet. The difference is emphasis: clear answers, consistent facts across the web and deliberate crawler access matter more than they used to.

    Do I need an llms.txt file or special schema for AI search?

    Not for Google. Its documentation says you don’t need new machine-readable files, AI text files or special markup to appear in its AI features. Use structured data where it already fits your content, and keep it matched to what’s visible on the page.

    How do I know if AI tools are mentioning my brand?

    Combine three signals: AI referral traffic in GA4, page-level trends in Search Console and a monthly manual check of the questions your audience actually asks. The manual check also shows which sources get cited, which tells you what to fix first.

    / Work with Brittany

    Want to know how your brand shows up in AI answers?

    Hiring a marketing leader or need a hands-on partner for a project? See how we can work together or send me a note.

  • Using Responsible AI in Marketing: A Guardrails Checklist

    Somebody on your team is already using AI. Maybe it’s drafting social captions. Maybe it’s summarizing meeting notes or cleaning up a grant narrative at 10 p.m. The question isn’t whether AI is in your marketing work. It’s whether anyone has set the rules. Responsible AI in marketing starts with admitting that the tools showed up before the policy did.

    I work with AI every week. Analytics, strategic communications and AI are the deep end of what I do, so I’m not writing this as a skeptic. I’m writing it as someone who has seen how fast a helpful shortcut turns into a brand problem when nobody owns the guardrails.

    Good guardrails don’t slow a team down. They let people move faster because they know where the edges are.

    Why responsible AI in marketing is a leadership job

    It’s tempting to treat AI as an individual productivity choice. One person likes it, another doesn’t, everybody does their own thing.

    That works right up until:

    • Someone pastes donor or client details into a free tool with unclear data terms.
    • A generated statistic ends up in a press release, and nobody can find where it came from.
    • An AI image of “a family we served” runs in a campaign, and the family doesn’t exist.
    • Your newsletter starts sounding like everyone else’s newsletter.

    Every one of those is a trust problem, and trust is the asset marketing is supposed to protect. That makes AI governance a leadership responsibility, not a tech-support ticket. If you lead a team, the rules are yours to write, even if you write them in an afternoon.

    The guardrails checklist

    Here’s the checklist I’d hand any marketing team, from a solo nonprofit communicator to a department of twenty. Adjust the details. Keep the categories.

    1. Data: what never goes in

    • No personal information about clients, donors, patients, students or employees in any tool your organization hasn’t approved.
    • No confidential financials, contracts, legal matters or unreleased plans.
    • Know each approved tool’s data terms: does it store prompts, and can it train on them?
    • When in doubt, anonymize or summarize before you paste.

    2. Truth: what always gets checked

    • Every fact, number, name, date and quote gets verified against a real source before it’s published.
    • AI-suggested statistics are treated as leads, not facts. If you can’t find the original source, the number doesn’t run.
    • Summaries of long documents get spot-checked against the original, especially anything you’ll repeat to leadership.

    3. People: who stays in charge

    • A named human reviews and approves anything that goes public.
    • Sensitive communications (crisis statements, layoffs, condolences, anything about a specific person) are written by people. AI can help you outline. It shouldn’t hold the pen.
    • Final judgment calls on tone and timing stay with the communicator who knows the audience.

    4. Images and likeness: no synthetic “real people”

    • Never generate an image that implies a real client, donor, employee or event.
    • Don’t alter real photos in ways that change what happened.
    • If you use generated visuals for illustration, make sure they read as illustration.

    This one matters a lot in mission-driven work. I’ve written about ethical storytelling with consent and dignity, and the same principle applies here. A made-up face standing in for a real person’s story isn’t a shortcut. It’s a misrepresentation.

    5. Brand: does it still sound like you?

    • Keep a short voice guide that people can paste into a prompt or check drafts against.
    • Read AI-assisted drafts out loud. If it sounds like a brochure for nobody in particular, rewrite it.
    • Hold AI output to the same brand standards as anything else. Templates, colors and messaging rules still apply.

    6. Disclosure: when do you say AI helped?

    There isn’t one universal rule here, so decide yours and write it down. A reasonable starting point: disclose when AI generated a substantial part of an image, a voice or content presented as someone’s personal words. Don’t put a quote in a leader’s mouth that they didn’t write and approve.

    7. Tools: who decides what’s approved

    • Keep a short list of approved tools and what each is approved for.
    • Assign one person to review new tools before they’re used with real work.
    • Revisit the list a couple of times a year, since terms and features change.

    How to roll it out without a 40-page policy

    You don’t need a committee and six months. You need a page people will actually read.

    1. Inventory what’s already happening. Ask your team, without judgment, which tools they use and for what. You’ll learn more in a 20-minute conversation than in a formal audit.
    2. Draft a one-page guide. Use the seven categories above. Plain language, short bullets.
    3. Get a quick review from whoever owns risk. In a small nonprofit, that might be the executive director. In a larger organization, legal, HR or IT.
    4. Train with real examples. Show a good use and a bad use for each category. Examples stick better than rules.
    5. Make it easy to ask. A shared channel or a named person for “is this okay?” questions prevents a lot of quiet guessing.
    6. Review it twice a year. The tools will change. Your principles probably won’t, but the details will.

    If you want a deeper framework to borrow from, the NIST AI Risk Management Framework is a solid, free reference. You don’t need to adopt the whole thing to get value from its way of thinking about risk.

    Where AI genuinely helps (and where I keep it out)

    Guardrails are only half the picture. The other half is using AI where it actually earns its keep.

    Good fits

    • First drafts of routine content you’ll heavily edit, like event recaps or internal updates.
    • Brainstorming angles, headlines and subject lines to react to.
    • Summarizing survey comments or long reports so you can find themes faster, followed by your own read.
    • Repurposing approved content into different formats and lengths.
    • Helping with formulas, scripts and small automations when you’re building your own tools.

    Where I keep it out

    • Anything that requires knowing a specific person’s story.
    • Final messaging for sensitive or high-stakes moments.
    • Strategy itself. AI can help you pressure-test a plan, but it doesn’t know your board, your community or your history.

    AI is one tool among many. The marketers who’ll do best with it are the ones who already have strong judgment about audience, message and brand. The tool speeds up the work. It doesn’t replace knowing what good looks like.

    What to do this week

    If you have nothing in place yet, start small:

    • Send your team a three-line note: no personal data in unapproved tools, verify every fact, a human approves anything public.
    • Book 20 minutes to ask what tools people are using.
    • Draft the one-page guide using the checklist above.
    • Add AI review to your existing approval process, so nothing new needs to be invented.

    Those four steps cover most of the real risk. Everything else is refinement.

    If your team already uses brand templates, put the AI guide right next to them. Guardrails work best when they live where the work happens, a lesson I’ve learned building Canva brand templates for teams who aren’t designers.

    Key takeaways

    • AI is already in your team’s workflow, so the real question is who sets the rules.
    • Protect data, verify facts and keep a named human in charge of anything public.
    • Never use generated images or words to stand in for real people.
    • A one-page guide people read beats a long policy nobody opens.
    • Use AI where it saves time on routine work, and keep strategy and sensitive stories human.

    FAQ

    What does responsible AI in marketing mean?

    Responsible AI in marketing means using AI tools in ways that protect people’s data, keep content accurate, stay honest about what’s real and keep humans accountable for what gets published. It’s a set of team guardrails, not a single tool or setting.

    Do small marketing teams need an AI policy?

    Yes, but it can be one page. Small teams often carry the most risk because one person handles sensitive data, public content and approvals at once. A short written guide gives that person clear rules and protects the organization.

    Should marketers disclose when they use AI?

    Decide on a clear internal rule and apply it consistently. A sensible baseline is to disclose when AI substantially created an image, a voice or words presented as a real person’s, and to never publish AI-written quotes as if someone said them.

    Curious how I approach marketing, analytics and AI together? You can read more about my background.

    / Work with Brittany

    Want AI to save your team time without costing you trust?

    Hiring a marketing leader or need a hands-on partner for a project? See how we can work together or send me a note.