LinkedIn
By December 31, 2027, no agency managing five or more client domains will pay per tracked prompt for AEO. Pull up this post then and grade me. The scorecard is at the end.
I have seen this meter break before. I used to pay for rank tracking by the keyword, then watched that charge fall away once every tool could crawl Google cheaply enough. Prompt monitoring is on the same path. Checking what ChatGPT or AI Overviews said yesterday is getting cheaper. Doing something about the answer across ten CMS logins is not.
That is my claim for the next 12 to 24 months: prompt monitoring becomes a commodity, agency pricing settles around a flat rate per domain with uncapped prompts, and the valuable work moves to publishing fixes and building citations. Here are five predictions, including what would prove each one wrong.
Prediction 1: Per-prompt caps disappear from agency pricing by the end of 2027
Right now, caps are the business model. HubSpot AEO charges $50 a month for 25 prompts, Profound starts at $99 a month for 50 prompts on ChatGPT only, and Semrush sells AI Visibility at $99 a month per domain for 25 prompts with another $60 for 50 more. I watch these prices because I watched rank tracking the same way in 2015: when every vendor meters the same basic check, I start looking at how long that meter can hold.
The mechanism is straightforward. Running the same 50 prompts against five to seven models every day is compute plus parsing. Daily refresh is already standard, not premium, and the spread from Otterly Lite at 15 prompts to Scrunch Core at 125 prompts for $250 a month puts the price per daily prompt between $0.45 and $2.46. As the check gets cheaper, it gets harder to sell the check as the product. Uncapped tracking becomes the row you expect in the comparison table, not the feature you pay extra to turn on.
Why I’m betting against the meter
I paid per keyword for years because crawling SERPs used to be hard. Then every tool built the same crawler, the price fell through the floor, and rank tracking became the unlimited row nobody shops on. I used to tell clients keyword counts mattered. I was wrong. What mattered was who acted on the movement.
Prompt checks face the same problem. A count cannot tell an agency which prompts matter to revenue or which cited URL needs a fix. If the top three agency tools still charge per prompt block in December 2027 and agencies keep paying without churning, this prediction is wrong. Until then, buy the shortest term you can get on a capped plan.
Prediction 2: Flat per-domain pricing becomes the default for multi-client agencies
Agencies managing 5 to 15 domains will force this change. Semrush already charges another $99 for each extra user or domain, so a ten-client book can run into seat, domain, and prompt-pack charges at once. I once ran the math for a home services client with nine locations; the tracking line cost more than the content work. That is a hard invoice to defend.
A flat rate per domain makes the next client’s tooling cost predictable. You can add client eleven without reopening a spreadsheet to count prompts, users, and models. The vendor still charges for each domain, but you stop negotiating a new allowance every time the work grows.
The spreadsheet for a 5–15-domain book
Say you run ten clients at 50 prompts each. One analysis puts current AEO retainers at two to three times SEO per client because those prompts run against five to seven platforms. ChatGPT, Perplexity, Gemini, Claude, AI Overviews: the list adds up quickly, and so does the analyst time. Tooling alone adds $300 to $1,500 a month on top of the SEO stack.
I have lived inside that spreadsheet. The monitoring line can hit $500 to $1,000 per client before anyone writes a word. Multiply that across ten clients and you are paying a full-time salary for screenshots. The pressure is not abstract; it arrives every month as a choice between paying to observe more prompts and paying someone to act on what you have already observed.
I am wrong if agencies at this size still accept metered billing in late 2027 without demanding a flat domain rate. Watch what they buy, not what vendors announce. If a vendor cannot quote one number per domain with uncapped prompts, keep the contract short.
Prediction 3: Monitoring-only tools get bundled away or absorbed
A tool that shows the gap without doing the work has a short life in an agency. The pattern in 2026 agency tiers is already visible: basic monitoring at $500 to $1,500 a month carries 75%-plus margin, but reporting alone does not prevent churn. Budget cuts at 42% and lack of perceived value at 32% outrank reporting issues at 10%.
I learned the distinction selling PPC reports in 2017. Clients do not cancel because a chart looks ugly. They cancel because nothing changes after they see it. Monitoring without fixes is a churn machine.
A visibility score is not a work order
Credit-metered systems also bill you for curiosity. Every prompt times every model times every day draws from the same balance, even though daily refresh is the baseline. An agency needs more than a brand-level visibility score: it needs one workspace per client, the exact cited URL, and export or API access to connect citations to revenue.
Without that detail, you know a score moved but not which prompt changed, which source appeared, or what page to fix. I have sat in the meeting where a score went up four points and nobody could say what to do next. That is the monitoring product I expect to see bundled into something else, absorbed, or given away.
I am wrong if agencies still pay separately for monitoring-only plans at the end of 2027 and clients keep renewing them without a path from the data to a fix. Before paying for a dashboard, ask it to name the page you should change.
Prediction 4: CMS publishing and edge-delivered fixes become table stakes
Monitoring tells you ChatGPT picked a competitor. It does not change the page ChatGPT might cite next time. Across ten clients, that handoff is where an AEO plan stalls: someone has to push schema, metadata, and answer blocks live while working around each client’s developer queue.
I spent years waiting on that queue for PPC landing pages. AEO repeats the problem across every domain in the agency’s book. By the end of 2027, an agency tool without a shipping path will be classified as reporting.
The fix has to reach the live page
The bottleneck is access, not advice. Teams addressing it push generated schema through CMS APIs such as Contentful, Strapi, WordPress, or Shopify with field-level PATCH mapping. An approved fix can then reach the pages that need it without a fresh developer ticket for each change. That is what turns a cited-URL finding into work an agency can deliver.
For legacy builds where nobody will hand over CMS keys, the work can move to the edge. The same writeup describes injecting fixes through Cloudflare Workers or Vercel Edge Functions. It also describes tracking GPTBot, ClaudeBot, and PerplexityBot at the CDN or server level because GA4 misses bots that do not run JavaScript. If you cannot tell whether a crawler saw the fix, explaining what your fee bought gets harder.
Publishing is only part of the job. Agencies also have to use the cited-source detail from monitoring to decide where to build citations, rather than treating a higher visibility score as the deliverable. Find the prompt, inspect the cited URL, ship the page change, and check what gets picked up. That is a service; a screenshot is not.
I am wrong if most agency AEO contracts still exclude publishing rights at the end of 2027 and clients renew them anyway. Ask how a proposed fix gets from the dashboard to the live site.
Prediction 5: Agencies stop reselling dashboards and start reselling outcomes
By the end of 2027, no healthy agency will resell a login and call it AEO. A rebrandable dashboard still needs people to do the work. An agent platform does the work itself, and flat pricing protects margin as client count triples because per-seat and per-usage costs rise with the agency’s success.
That has the same feel as percentage-of-spend pricing in PPC: the bill grows even when the work does not grow at the same rate. When average agency net margin sat at 13% in 2025, holding service prices steady while agent costs sit at $49 to $500 a month is how shops pitch 60% to 80% gross margin on AI-delivered services. The margin argument is clear. The client argument still depends on whether the agency ships anything worth buying.
A logo on a PDF is not a delivered outcome. White-label options run from a static branded PDF to a live dashboard on your subdomain to a resellable platform. At the platform end, you need a custom domain with SSL, scheduled delivery, multi-client roles, and API or webhooks to pull raw citation data into your systems. Those features help an agency operate across clients. They do not, by themselves, earn the retainer.
Clients pay for fixes, which is why I keep pointing agencies to Stop Reselling AEO Dashboards. Clients Pay for the Fixes. I am wrong if agency proposals still lead with dashboard seats and prompt packs in late 2027, and clients keep buying them without asking what changed. If the deliverable is a chart, you are renting churn.
What I would demand from an AEO vendor now
If uncapped tracking, flat per-domain pricing, and shipped fixes are where this ends up, your contract should already reflect that. I would not sign for a platform that meters curiosity and leaves publishing to someone else. Before moving 5 to 15 domains, get four answers in writing and run them against Before You Resell AEO Across 10 Clients: A 16-Point Vendor Checklist:
- Price: Are prompts uncapped in the base per-domain fee? Are there prompt packs or per-seat charges?
- Publishing: Can the platform publish through the CMS and deliver edge fixes without a new developer ticket for every client change?
- Evidence: Can you see the exact cited URL and export the underlying data or reach it through an API?
- Delivery: Can you serve the work under your own domain rather than hand clients another vendor login?
Ask for one number per domain, no prompt packs, no per-seat uplift, and a 30-day exit. If daily refresh costs extra or additional models are add-ons, walk away. That vendor is defending the meter, not your margin.
The scorecard: check back on December 31, 2027
Bookmark this and grade me against the market, not the sales decks:
- No leading agency tool charges per tracked prompt block as its base plan, and agencies managing five or more domains no longer pay per tracked prompt.
- Flat per domain with uncapped prompts is the default quote for 5-to-15-domain books.
- Standalone monitoring-only plans under $300 a month are gone, bundled, or free.
- CMS publishing, edge delivery, and bot logs ship in every agency-tier plan.
- Agency AEO proposals quote fixed outcomes per domain, not dashboard seats and prompt packs.
If I get three or more wrong, I will write the retraction and name what I missed. If I get four right, stop buying meters. Buy a vendor that charges per domain, tracks everything, and ships the fix under your brand. groas already prices and delivers that way for agencies, which is why I am comfortable putting dates on this. Run one client through the free trial and see whether uncapped tracking plus actual publishing changes your margin math.
This will not matter equally to everyone. If you manage one domain with 20 prompts that never change, a capped $50 plan is fine for another year. The math breaks when prompts times models times clients starts climbing. This prediction is for the agency with a 5-to-15-domain book and an invoice that climbs faster than the work gets done.
Save the scorecard somewhere you will open it. Set a reminder for December 2027, put your current AEO invoice beside it, and see whether you still pay for the count or pay for the fix.