Can I buy only the AI search layer and keep my current SEO agency?

Yes. The AI search layer runs as its own scope, with its own deliverables and its own reporting. Your current agency keeps rankings, backlinks, technical work and the site build. The AI vendor works on the citation footprint, question-shaped schema, decision-stage pages and lead attribution. Write the split into both contracts before anyone touches the site, so two teams never edit the same title tag in the same week. Of the growth calls we ran between March and August 2026, 56 percent already had a vendor in place. Almost none of them fired that vendor to get started.

What does an AI-only add-on cost next to an existing SEO retainer?

For a single location with a working site, $800 to $2,500 a month. Setup is $500 to $1,500 one time, covering the audit, the schema build and the tracking wiring. Multi-location scopes and franchise groups run higher, up to roughly $3,200 a month, because every location needs its own entity work. Under $800 usually means someone is reselling a citation tool with a dashboard bolted on. Over $3,200 usually means you are buying a full SEO retainer with the letters AEO printed on the invoice. Ask which of the four deliverables the number covers.

How much of the work overlaps with what my SEO agency already does?

Roughly a quarter, and the crossover sits at the foundation layer. Both vendors care about schema, business listings and clean page structure. The difference shows up in what happens next. Traditional SEO shapes a page to rank for a phrase. The AI layer builds an answer a model can lift and attribute to you. Same page, different job. Handle the crossover with a written split: one team owns title tags and headers, the other owns schema blocks and answer sections. Overlap turns into damage only when nobody has agreed who edits what.

Will two vendors working on the same site break something?

Only if the split is verbal. Two teams editing the same WordPress install without a boundary will overwrite each other, usually on title tags and schema. The fix takes ten minutes. Name the owner of each surface in writing: pages, headers, schema, blog, listings, tracking. Give each vendor its own login so the revision history shows who changed what and when. Then run a monthly check on the handful of pages both teams touch. In practice the risk is smaller than owners expect, because the two scopes barely meet outside of schema and listings.

How do I tell which vendor produced a given lead?

A lead log with a source column, filled in weekly. Every call, form and text gets a row: date, the first line of what they said, and where they came from. Tracking numbers separate the channels, and an AI-sourced lead often announces itself, because the caller repeats the phrasing a model handed them. Ask both vendors to log leads the same way, in the same sheet. When one of them will not, that refusal tells you something on its own. A shared lead log settles the two-vendor argument faster than any ranking report ever will.

Is there a setup fee, and should the add-on be month to month?

Setup runs $500 to $1,500 and covers the audit, the schema build and the tracking. Ask for the retainer month to month with 30 days notice. An annual lock-in on an add-on is a warning sign, because the whole point of an add-on is that you can drop it without disturbing your main SEO contract. Watch two extras that get billed separately: content volume above the agreed page count, and entity work for additional locations. Get both written into the scope with a number attached, never a phrase like as needed.

How long before the add-on shows anything?

Movement inside 30 days, a revenue signal by 60 to 90. In the first month you should see schema live on the money pages, the citation footprint filed, the first decision-stage pages published, and a baseline showing where you appear today across ChatGPT, Perplexity and Google AI Overviews. Citations start landing in weeks four through eight for most single-location businesses. Anyone promising cited answers in week one is either working on a site that already carries strong authority, or telling you what you want to hear.

When does a full replacement make more sense than an add-on?

Replace when your current agency cannot show you what it did last month. An add-on works when the incumbent is competent and slow to adapt. It fails when the incumbent is absent, because you end up paying two retainers and receiving one set of deliverables. Test it before you decide: ask for last month's work log and the leads it produced. A clear answer means keep them and add the layer. Silence, a 30 page report, or rankings with no lead data means the money belongs with one vendor who reports honestly.

What an AI-Only Add-On Costs Next to Your SEO Retainer

TL;DR

You can buy the AI search layer on its own and keep your existing SEO agency. The two scopes sit side by side.
Most single-location add-ons run $800 to $2,500 a month, plus a one-time setup of $500 to $1,500. Multi-location scopes reach roughly $3,200.
Four things sit inside the scope: citation footprint, question-shaped schema, decision-stage content, and a lead log that names the source.
Your current agency keeps rankings, backlinks, technical crawl health and the site itself. Nothing gets taken off their plate.
The crossover is real and sits at the foundation layer, around a quarter of the work. Put the split in writing on day one.

Does It Overlap With Your Current Retainer

Short answer: yes, and the overlap is smaller than either vendor will tell you.

Most owners ask this in the wrong order. They ask what AI search costs, hear a number, then try to work out whether they are paying twice for the same hour of work. Run it the other way round. Settle the scope first, look at what actually sits on top of your current retainer, and the price question answers itself.

Your SEO agency works on rank position. Backlinks, crawl health, title tags, internal links, blog volume aimed at keywords with measured search volume. That work still matters. Traditional search feeds the AI platforms, so a site that ranks poorly in Google tends to get quoted rarely in ChatGPT.

The AI layer sits above it. The question there is whether a model can find your business, understand what you sell, and repeat it accurately when a stranger asks for a recommendation. Different mechanism, different deliverables, same website. If your incumbent has told you they already cover this, run the checks in five things to check before you believe them before you buy anything from anyone.

What Sits Inside the Add-On

Four deliverables. Ask any vendor to price against these four, line by line, and the vague quotes fall apart quickly.

1. Citation footprint

Models read a narrow set of sources: Foursquare, Bing Places, Apple Maps, industry directories, association pages, review platforms. Your SEO agency probably filed the big three years ago and has not looked since. This is claim, correct, expand, and then maintain.

2. Question-shaped schema

LocalBusiness, Service and FAQPage markup written the way a person actually asks, so a model can lift a sentence and attribute it to you. Basic schema for rich snippets is a different job with a different shape.

3. Decision-stage content

Pages that answer the comparison a buyer types in full sentences. Cost, timeline, who it suits, who it does not. Keyword pages chase volume. These chase the moment someone is choosing.

4. Lead log with attribution

Every call and form logged with a source, weekly. This is the deliverable that ends the vendor argument, and it is covered in detail in how to tell which agency a lead came from.

Concept illustration for the section What Sits Inside the Add-On — Concept illustration: What Sits Inside the Add-On.

What Stays With Your Current Agency

Rankings, backlinks, technical crawl health, site speed, the CMS, and the blog they already publish. If they build your pages, they keep building your pages. If they run Google Business Profile posting, that stays too.

The reason to keep them is simple. Ranking work compounds, and tearing it up to start again costs you six months. An add-on removes nothing from their scope, which also means the conversation with them is short. You are adding a layer, not auditing their performance. For the mechanics of splitting surfaces cleanly, read how two vendors split the work without breaking the site.

What the Add-On Costs

Now the number, with the scope already settled above it.

A single-location business with a working site runs $800 to $2,500 a month. Setup is $500 to $1,500 one time, covering the audit, the schema build and the tracking wiring. Multi-location and franchise groups reach roughly $3,200 a month, because each location needs its own entity work and its own decision pages.

Three things move the number: how many locations you run, how much decision-stage content gets produced each month, and whether your site can accept schema without a rebuild. A dental practice in Phoenix on a clean modern site lands near the bottom of the band. A restoration company in Wisconsin with eight service areas and a 2018 template lands near the top.

Under $800 usually means a citation tool with a dashboard on top. Over $3,200 usually means a full SEO retainer wearing an AEO label. Ask for month to month with 30 days notice, and get content volume written in as a number. The retainer terms breakdown lists the clauses worth arguing over.

Three Routes Compared

Route	Monthly cost	Main risk	Best when
AI add-on beside incumbent	$800 to $3,200	Two teams editing the same surfaces	Incumbent performs and reports clearly
Move everything to one AI-first agency	$2,500 to $6,000	Losing ranking momentum in the handover	Incumbent cannot show last month of work
Ask your incumbent to add AEO	$0 to $600 uplift	Schema renamed, nothing new shipped	They can name their four deliverables

Who Should Buy It, Who Should Not

Buy the add-on if your SEO agency produces leads you can trace, answers questions in ten minutes, and simply has no AI deliverable. You are filling a gap, not replacing a team. This is the most common case we see.

Buy a full replacement if your incumbent cannot produce last month's work log, sends 30 page reports with no lead data, or answers the AI question with a slide about rankings. Paying two vendors when one of them is dormant costs you twice for the same silence. The three-question test takes ten minutes and settles it.

Buy neither yet if your site cannot accept schema without a rebuild, or you have no way to log where a lead came from. Fix tracking first. An add-on with no attribution produces the exact problem you already have: spend going out, no evidence coming back. Sort the measurement, then buy the layer.

What You Should See in 30 Days

Schema live on your money pages. The citation footprint filed and verified. The first two or three decision-stage pages published. A baseline showing where you appear today when someone asks ChatGPT, Perplexity or Google AI Overviews for a recommendation in your trade. And a lead log with a source column, already filling up.

That is the check at day 30, whoever you hire. Three numbers, reported weekly: the phone rang this many times, this many turned into jobs, this much cleared. Everything else is decoration, which is the argument in what a marketing report should actually show you.

Want the number for your business?

We will scope the add-on against your current retainer and tell you the honest overlap before you pay anything.

Book a 20 minute scope call See where you appear today

About the author

Matthew Johnson is the founder of Pleiades Consultancy. He previously scaled his own marketing agency to multiple six figures before serving as CMO of an Amazon agency, where the client base tripled from 15 to 45 active clients during his tenure. He worked with some of the largest names in e-commerce, including Ridge Wallet, HexClad, BK Beauty, The Woobles, Walkize, Lonely Planet, and Obvi. He now works with local businesses to maximize their client acquisition and visibility through AI search with ChatGPT, Claude, Gemini, Perplexity, and Bing Copilot.

YouTube LinkedIn X / Twitter About Pleiades