Generative engine optimization is the work of getting your company named and cited in AI answers. Eicra Soft Limited, a Dhaka-based team, measures 50 buyer prompts on four engines, rewrites five pages to be the quoted answer, and re-measures the same set at day 30, at a fixed price per step. Every number is measured three times, never estimated.
Every step can be your last; none commits you to the next. We start with the 50 questions your buyers actually type into an AI engine. Each is run on four engines, three times; who is named, who is cited and who is absent goes into one sheet you keep, and the smallest change that would move your share, usually five pages and the entity data, is priced against your own numbers.
You get a written verdict — an audit first, answer-first content, entity and schema work, third-party presence, a full pilot, or nothing yet. If nothing justifies a build, you stop here and keep the audit.
Generative engine optimization services cover the five things that decide whether an engine built on a large language model cites one company and skips another: pages that answer the question directly with sourced numbers, an entity the engine can verify, presence on the sources it already trusts, pages it can crawl, and someone who measures the result.
Fifty buying questions are agreed with you and run on four AI engines three times each. Every mention, citation and absence is recorded: your citation share, your competitors’ and the sources used. The sheet is yours to keep whatever you decide next, and the audit fee is credited to the pilot, so measuring first costs nothing extra.
Service and landing pages are rewritten so each heading asks the buyer’s question. The first paragraph answers it in 40–60 words with a statistic, a quotation or a named source. Five pages are rewritten in the pilot, each approved by you before publication, and further pages follow in batches of five with their own re-measure.
Organization, Service and FAQ schema go in, with consistent name and address data. About and author pages and sameAs links to your profiles let an engine verify who is speaking. The schema is validated after publication, lives in your own site, and is documented so your team can extend it without us.
Honest submissions go to the directories, industry lists and communities the engines cite for your prompts, tracked to publication. The publisher decides; no mention is bought or faked. The audit names which sources the engines actually cited for your 50 prompts, so submissions go where they can change the answer.
Robots rules, indexing, page speed and rendering are checked for the pages in scope, because an engine cannot cite a page it cannot read. Fixes are listed or done in the pilot. The check uses read access to Google Search Console on your account, and every finding is written down with its fix and owner.
The same 50 prompts are re-run at day 30 and monthly. Citation share, competitor share and the sources cited sit in one report, with optional monthly monitoring as engines change. The report shows movement either way, names the engine behind each change, and recommends the next batch only when the numbers support it.
Answer engine optimization (AEO) is the older, narrower layer inside generative engine optimization (GEO): the featured snippet and the AI Overview, won by one direct answer. GEO adds ChatGPT, Perplexity, Gemini, entity checks and third-party sources. AEO is the first pass of every rewrite, in the same fixed steps and never sold separately, so you never need two vendors.
Google Search Central states that “the best practices for SEO remain relevant for AI features in Google Search”, so every AEO pass starts from crawlable, well-structured pages. SEO itself is free: Google takes no money to rank a site, so what you pay for is the work. New to SEO? Start with three basics: pages Google can crawl, one buyer question answered on each page, and mentions on sites your buyers trust.
Every heading is the question or keyword a buyer types, so the engine can match the page to the prompt without guessing. Headings come from the audited prompt set, not from guesswork, so each one maps to a question buyers already ask an engine.
The first 40–60 words under each heading answer it outright, before any context, so the passage can be lifted whole. An engine quotes passages, not pages, and a passage that answers cleanly in its first sentences is the one that gets lifted.
Each answer carries a number, a quotation or a named source, so the engine has something it can check before citing it. Anything that cannot be sourced is removed rather than softened, because one unverifiable claim can cost the page its credibility.
The questions people also ask, answered in the same form and marked up with FAQ schema, so snippets and AI answers can read them. The FAQ set comes from the audit’s prompt list and from your sales calls, so it answers what buyers ask, not what we assume.
A generative engine optimization pilot takes two weeks to build after a five-day audit, then the same 50 prompts are re-measured at day 30. The example is a services company; every item traces to a prompt run, a page version or a re-measure.
The schema.org vocabulary defines sameAs as the “URL of a reference Web page that unambiguously indicates the item’s identity”, which is how an engine checks who is speaking.
| Step | Time | Price | What you receive |
|---|---|---|---|
| Scope | 30 minutes | Free | The buying questions, pages in scope and a quote |
| Diagnose | 5 days | $99, credited to the pilot | 50 prompts run on four engines; your citation share in writing |
| Pilot | 2 weeks | $399, fixed | Five pages rewritten, schema set, re-measured at day 30 |
| Production | Per batch | $800 per five pages | Further pages in approved batches, each re-measured |
| Managed Ops | Monthly | $499 a month, cancel any month | Prompts re-run monthly, pages refreshed, one report |
Pilot log · 50-prompt set
What the step does · its result
Audit
Who is named, cited or absent, before any change.
| Audit measure | Sample result |
|---|---|
| Prompts | 50 buying questions |
| Engines | 4 |
| Runs | 3 per engine |
| Named | 4 of 50 |
| Competitors | 3 mapped |
Rewrite
Each page answers the buyer’s question first.
Entity
So an engine can verify who is speaking.
Sign-off · a person signs
Nothing is published before you approve it.
Day 30
The same set, re-run on the same engines.
Choose a step, or a number below, to see its result
Managed Ops re-runs the same prompt set every month, so the number stays comparable.
Illustrative example. Yellow marks where a person signs; nothing is reported before it is measured.
An AI visibility engagement runs in five steps, each ending in a written result you can stop on: a free call, a five-day AI visibility audit that quotes the pilot, a two-week pilot that rewrites five pages and re-measures at day 30, production in batches, then Managed Ops. Our generative engine optimization services are priced one step at a time.
An SEO specialist who owns the prompt set, audit and schema, a content editor who rewrites pages in your brand voice and sources every number, and a project manager. All three work inside your CMS and your accounts, and your approver signs every page before it goes live.
The first GEO step depends on your gap; five questions show which fits. Never measured means the audit first; pages that describe but never answer, content; engines that cannot verify you, entity and schema; absence from the lists engines quote, third-party presence; all four partly in place, the full pilot. Drafts can come from a tested prompt system.
1. Has anyone checked what AI engines say about you?
2. Do your service pages answer buyer questions directly?
3. Are you on the directories and lists AI engines cite?
4. Is schema and entity data in place?
5. Who runs your SEO today?
Answer the five questions and the checker suggests the smallest useful step: the audit alone, answer-first content, entity and schema work, third-party presence, or the full pilot. The five-day audit then confirms the verdict in writing before any build is priced.
Nobody has measured whether AI engines name you today, so nothing else can be prioritized: 50 buyer prompts on four engines, three runs each, and a written verdict on where you are cited, absent or beaten. The $99 fee is credited to the pilot if you continue.
A first estimate; the audit confirms it.
How the verdict is decided
A generative engine optimization agency should show its price and its method before it asks for a retainer: our audit is priced on the cards above, and our method is written down, 50 prompts on four engines, three runs each. Personal data in analytics or author profiles stays under Article 28 of the General Data Protection Regulation (GDPR).
This is the pattern that gets domains flagged and budgets wasted. A vendor sells a retainer with no baseline measurement, buys presence the engines later discount, and reports rankings while the AI answer still names a competitor. Nothing can be verified, so nothing can be managed.
Every claim is measured before and after on the same 50 prompts, so the pilot report below can be checked line by line. The method is written down, the price is on the page, and every rewritten page carries your approval. The report is an illustrative example, not a client result.
Outsourcing generative engine optimization is safe when honesty, access and approvals are in writing, because the real risk is a vendor who fakes presence and gets your domain flagged. As a Bangladesh-based company, we work in your CMS, publish nothing without your approval, never buy reviews or seed mentions, and leave at handover.
Every agreement below is signed before we see an account, an analytics property or a prompt set. That covers a mutual NDA, processor terms for any personal data, and transfer clauses where your jurisdiction requires them. Access runs on your seats and you can revoke it any day.
You keep the veto while the work runs. Nothing publishes without your written approval, every submission sits in a log you can audit, and every statistic carries a named source. Pages, schema, the prompt set and the reports live in your accounts from day one.
Before you pay for generative engine optimization, you get evidence instead of promises: a written method, a five-day audit that gives you your own number, and a pilot re-measured on the same 50 prompts at day 30. Named client results follow once each client agrees to publication.
50 prompts
Run on four AI engines, three times each, and recorded before you pay for any rewrite.
5 days
For the AI visibility audit, ending in a written verdict on where you are cited, absent or beaten.
Day 30
The same 50 prompts are run again after the pilot, and the change is reported either way.
Twelve questions buyers ask before they commission generative engine optimization, each answered in its first sentence. They cover cost, what GEO is, how it differs from AEO and AI visibility, measurement, the engines covered, SEO, the steps after you enquire, ownership, data handling, switching vendors and what we will not promise.
More answersGenerative engine optimization services here are priced per outcome, never per hour, and the cards at the top show each price: a five-day AI visibility audit of 50 buyer prompts across four AI engines with a written verdict, credited to the pilot; a pilot that rewrites five pages and re-measures at day 30; further batches; then Managed Ops.
Generative engine optimization is the work that makes an AI answer engine (Google’s AI Overviews, ChatGPT, Perplexity, Gemini) name and cite your company when someone asks a buying question: pages that answer questions directly with sourced numbers, clear entity information the engines can verify, presence on the third-party lists they draw from, and measurement of how often you are cited.
AI visibility is the result: how often your brand is named or cited in AI answers for a fixed set of prompts. Generative engine optimization is the work that raises it across ChatGPT, Perplexity, Gemini and AI Overviews. Answer engine optimization is the older, narrower version aimed at featured snippets and AI Overviews. One service, one measurement, three names.
With a fixed prompt set. We agree 50 buying questions your customers ask, run each on four AI engines three times, and record whether your company is named, cited with a link, or absent. That gives a citation share, for example 4 of 50. The same set is re-run at day 30 and monthly, so the number is comparable.
No. AI engines still learn from crawlable, well-structured, authoritative pages, so technical SEO and real content remain the base. GEO adds what the engines reward on top: direct answers with statistics and quotations, verifiable entity data, and presence on the sources they cite. If your site has crawl problems, the audit says so first.
A free 30-minute scoping call comes first, and every later step is yours to approve. Each step ends in a written result, and if the audit shows nothing worth building, you stop there and keep it. The steps run in this order.
Four: Google AI Overviews, OpenAI’s ChatGPT, Perplexity and Gemini. Every prompt is run by hand on each engine three times per measurement, because answers vary between runs, and the engine behind every recorded citation is named in the report. New engines are added to the set when buyers actually use them, as a change you approve.
You do, from day one. Rewritten pages live in your CMS, schema sits on your own site, and the prompt set and every report are files in your accounts; nothing is hosted on our side. If you stop after any step, everything produced up to that step stays with you, including the audit sheet.
Under a mutual NDA signed before we see anything, and a data processing agreement following Article 28(3) of the GDPR, or your national equivalent, for personal data in analytics or author profiles. Access is editor-level in your CMS and read-only in Search Console, on your seats, revocable by you at any time.
Yes, and nothing needs handing back, because the audit builds your baseline from scratch: 50 prompts, four engines, your current citation share. Existing SEO work keeps its value, since crawlable, well-structured pages are the base GEO builds on. The audit also says plainly if your current setup already covers part of the work.
The first re-measure is at day 30 after the pilot, and that is the honest earliest point to judge movement. Engines refresh at their own pace, so some prompts move in weeks and others take months; the monthly re-runs under Managed Ops show the trend either way, on the same fixed prompt set.
No, and no honest vendor can, because the engines decide what they cite. What we guarantee is the work and the measurement: sourced answer-first pages, verifiable entity data, honest submissions, and the same 50 prompts re-run so you see exactly what changed. The illustrative pilot on this page moved a sample set from 4 to 11 named prompts.
Start with a free 30-minute call or the five-day AI visibility audit.
Request A QuoteBook a DiagnosticYour request is subject to our Terms and Privacy Policy.
