By TruePrime AI · Updated August 24, 2026
Every marketing service now claims to be “AI-powered.” Some are. Many bolted the word “AI” onto the same service they sold last year. These seven questions help you tell the difference before you sign a contract.
The AI marketing landscape in 2026 is crowded and confusing. Point tools that optimize one thing call themselves “AI marketing platforms.” Agencies that use ChatGPT for copywriting call themselves “AI-powered.” Full-stack growth engines that automate the entire pipeline exist alongside rebranded SEO tools. These questions cut through the branding to reveal what you’re actually buying.
This is the fastest filter. A vendor that uses its own product on a real, revenue-generating business has different incentives than one that only sells to others. They’ve hit the same problems you will — page quality, ranking stability, lead routing — and had to solve them because their own revenue depended on it.
Ask for specifics: which business, what domain, how long. If the answer is vague or defensive, that tells you something.
Testimonials can be cherry-picked, fabricated, or outdated. A vendor running the product on their own business creates a live, verifiable proof point you can check yourself. Visit their domain. Look at their pages. Check their search rankings. The evidence is either there or it isn’t — no testimonial required.
Search authority accrues to the domain hosting the content. If your AI marketing service publishes pages on their platform or subdomain, you’re building their SEO with your money. When you cancel, those rankings disappear because the pages weren’t yours.
The right answer: everything publishes on your domain. Your URL structure, your authority, your content — even if you leave.
This reveals the business model. Services that publish on your domain leave you with real assets — pages, content, schema — if you cancel. Services that host everything on their platform leave you with nothing. Neither is wrong, but you should know which you’re buying.
| Cancellation scenario | Domain-based service | Platform-based service |
|---|---|---|
| Content after cancellation | Stays on your domain, continues ranking | Disappears or redirects to vendor |
| SEO authority | Yours permanently | Returns to vendor’s domain |
| Schema/structured data | Remains on your pages | Removed with content |
| Lead capture forms | May need replacement, but pages stay | Gone entirely |
| Analytics history | In your analytics account | May be locked in vendor dashboard |
| Switching cost | Low — new service inherits your content | High — starting from zero |
If the answer is “what’s AEO?” or “we focus on Google,” that’s a service optimizing for 2023. A growing share of business discovery happens through AI assistants — ChatGPT, Perplexity, Gemini. An AI marketing service that doesn’t optimize for answer engines is ignoring a channel that’s growing every quarter.
Ask specifically: do they generate structured data, llms.txt, brand facts? Do they track citations in AI assistants?
| AEO capability | Strong answer | Red flag answer |
|---|---|---|
| llms.txt deployment | “We deploy and maintain it as a standard artifact” | “What’s an llms.txt?” |
| brand-facts.json | “Structured facts file, updated when business details change” | “We use Google My Business” |
| Schema markup | “JSON-LD on every page, using currently supported types only” | “We add some schema tags” |
| Citation tracking | “We query AI assistants on your target keywords and track citations over time” | “We don’t really track that yet” |
| Content structure | “Direct-answer format, question-based headers, factual and verifiable” | “Our content is optimized for Google” |
Bad reports measure activity: “we published 12 pages, wrote 8 blog posts, updated 3 meta descriptions.” Good reports measure outcomes: “keyword X moved from position 18 to 9, you received 4 new leads, ChatGPT cited your business for the first time on this query.”
Ask to see a sample report. If it reads like a timesheet instead of an intelligence briefing, the service is optimized for looking busy rather than producing results.
| Metric type | Activity metric (weak) | Outcome metric (strong) |
|---|---|---|
| Content | “Published 15 pages this month” | “3 pages entered top 10 rankings, 2 earning clicks” |
| SEO | “Optimized 20 meta descriptions” | “Average position improved from 24 to 16 across tracked keywords” |
| Leads | “Form installed on 5 pages” | “Captured 12 leads, 3 converted to consultations” |
| AEO | “Schema added to all pages” | “Cited by Perplexity for 2 target queries this week” |
| Competitive | “Monitored 5 competitors” | “Competitor X launched a page targeting your #3 keyword — here’s our response plan” |
This one separates experienced operators from newcomers. A stability rule means the service never regenerates or substantially changes a page that already ranks well. Why? Because Google re-evaluates changed pages, and rankings often drop during re-evaluation.
Services without this rule will “improve” your top-ranking pages and accidentally destroy what’s working. Ask how they handle pages that are already performing.
“AI-powered” shouldn’t mean “no humans available.” AI systems make mistakes — hallucinated facts, broken pages, wrong schema. When that happens, you need a human who can fix it. Ask about escalation paths and response times.
Use this scorecard to compare AI marketing services side by side. Score each provider 0–2 on each criterion (0 = no, 1 = partially, 2 = yes):
| Criterion | What to look for | Weight | Score (0–2) |
|---|---|---|---|
| Runs on own business | Vendor uses the product on a real, named, revenue-generating business | High | |
| Publishes on your domain | All content lives at your URL, not a vendor subdomain or platform | High | |
| Content portability | If you cancel, pages stay on your domain and continue ranking | High | |
| AEO included | Generates structured data, llms.txt, brand facts for AI assistants | High | |
| Outcome-focused reporting | Reports measure rankings, leads, citations — not activity counts | High | |
| Stability rule | Never regenerates pages that already rank well | Medium | |
| Human escalation path | Real humans available for errors, strategy, questions | Medium | |
| No lock-in contracts | Month-to-month, cancel anytime, no early termination fees | Medium | |
| Transparent pricing | All costs disclosed upfront, no hidden fees or required add-ons | Medium | |
| Content quality test | Every page passes “would a business owner learn something?” | Medium | |
| Security and data isolation | Client workspaces are isolated; your data never appears in another client’s output | High | |
| Compliance monitoring | Proactively adapts to Google/AI engine policy changes | Medium |
Scoring: 20–24 = strong candidate. 14–19 = proceed with caution. Below 14 = keep looking.
Not all AI marketing services are the same category. Understanding what you’re actually evaluating saves time:
| Type | What it actually is | Best for | Limitations |
|---|---|---|---|
| AI content tools | Software that generates text (Jasper, Copy.ai). You still do everything else. | Businesses with in-house marketing teams who need faster content creation | No SEO strategy, no deployment, no tracking, no lead capture |
| AI SEO platforms | Tools that analyze keywords and suggest optimizations (Surfer, Clearscope). You implement. | SEO professionals who want AI-assisted analysis | Analysis without execution — you still need to write, build, and deploy |
| Agencies with AI | Traditional agencies using AI tools to speed up their existing process | Businesses that want human account management with some AI efficiency gains | Still agency economics — AI reduces cost slightly but doesn’t change the model |
| AI growth engines | End-to-end automated systems that handle SEO, AEO, content, leads, and reporting | Businesses that want a complete marketing pipeline without managing tools or people | Don’t handle creative, paid ads, PR, or multi-channel campaigns |
When comparing vendors, make sure you’re comparing the same category. A $29/month content tool and a $999/month growth engine aren’t competitors — they solve different problems at different levels of the stack.
AI systems occasionally generate incorrect facts. A responsible service has review processes — automated fact-checking, human audits, correction workflows. If the vendor claims their AI never makes mistakes, that’s the biggest mistake of all.
Your brand facts, customer conversations, and leads should be completely separated from other clients. Ask specifically: are workspaces isolated? Can one client’s data ever appear in another client’s output? How is this verified?
Google rolls out core updates several times per year. A service that publishes content needs a compliance process: monitoring for policy changes, auditing existing pages against new guidelines, and fixing issues proactively — not reactively after rankings drop.
Don’t have time for a full evaluation? Ask this one question on the sales call: “Show me a real page you published for a real client in the last 30 days.”
Then evaluate what you see:
| What you see | What it means | Next step |
|---|---|---|
| A real page on a client’s domain with unique, substantive content | They actually produce work. Good sign. | Check if the page ranks for anything — production without results is still a problem |
| A page on the vendor’s subdomain (client.vendorplatform.com) | Content lives on their domain, not yours. You’re building their SEO equity. | Ask what happens to rankings if you cancel |
| A demo page or template with placeholder text | They’re showing capability, not production. Ask for a real example. | If they can’t show real output, they probably don’t have real output |
| “We can’t share client work due to NDAs” | Possibly legitimate, but ask: “Then show me the work you did on your own business.” | If they can’t show their own work either, there’s no proof the product works |
| A page that’s clearly AI-generated with no unique insight | They’re selling AI output without quality control. That’s a compliance risk for you. | Walk away — thin AI content can trigger Google’s scaled content abuse policy |
Here’s how the seven questions play out in practice. Pay attention to the difference between strong and weak answers:
“Yes — we run Go on ShopProp Realty and AskBeforeYouEat. Same agents, same process. Here’s the live site, here are the rankings.” Specific, verifiable, no hedging.
“We’ve tested it internally.” No business name, no domain, no results. “Tested internally” often means a developer ran it once on a staging site.
“Keywords tracked, positions with trends, organic leads attributed, cost per lead, AI citations by engine. Here’s a sample.” They hand you a real report with real numbers.
“We provide comprehensive monthly reporting with actionable insights.” No specifics. No sample. “Comprehensive” and “actionable” are filler words that mean nothing without evidence.
| Red flag | Why it’s a problem | What to do |
|---|---|---|
| Guaranteed rankings | No one controls Google’s algorithm. Guarantees mean black-hat tactics or lies. | Ask what specific tactics they use — if they can’t explain, walk away |
| Proprietary metrics only | If you can’t verify metrics independently, they’re probably designed to always look good | Require access to Google Analytics and Search Console directly |
| 12-month lock-in contracts | Suggests the service doesn’t retain customers on merit | Negotiate month-to-month or walk |
| AI-generated images and videos | Google’s spam policy covers AI media used to appear more authoritative | Ask about their media policy — text-only is safer in 2026 |
| No named portfolio companies | If they can’t name a single business they run the product on, the product may not work | Require at least one verifiable reference |
| “Thousands of clients” with no specifics | Volume claims without evidence suggest inflated numbers | Ask for 3 client domains you can check yourself |
The market has matured since we first published this checklist. Two new evaluation criteria have emerged from real buyer experiences:
The most important mental model for AI marketing is compounding. Content libraries, domain authority, AEO citations, and data feedback loops all accelerate over time. Month 1 results never represent what the system will produce at month 6 or month 12. Ask your vendor to explain the compounding timeline honestly — any service promising immediate results is either lying or selling paid ads under a different name.
The AI marketing market has split into two clear categories: tools you operate yourself and services that operate for you. A $99/month tool gives you capabilities. A $799/month service gives you outcomes. Neither is wrong, but comparing them directly is like comparing a gym membership to a personal trainer. Know which one you need before you start evaluating vendors.
We run Go on our own companies — ShopProp Realty and AskBeforeYouEat. Everything publishes on your domain. We include AEO in every program. Reports focus on rankings, leads, and citations. We enforce a stability rule on every page. And there are humans behind the AI — auditing weekly, available when things break.
Plans start at $499/month. First payment after 30 days. Cancel anytime.
Start your growth program