7 Ways to Win AI Search When Your Competitor Has 10x Your Domain Authority

Domain authority does not decide which vendors AI engines recommend. DerivateX works on the signals that do: entity consistency, specific evidence and third-party corroboration. 80% of URLs cited by ChatGPT and Perplexity are not in Google’s top 100, so a brand with a tenth of a competitor’s authority can still be named first in a buyer’s shortlist.

One clarification before anything else: there is no such thing as an AI search domain authority score. No checker tool returns one, no engine publishes one, and nothing you can look up today will tell you how likely a model is to recommend you. The phrase describes a misconception, that the metric built to predict Google rankings also predicts which brands a language model names in an answer. It does not, and the rest of this piece covers what does.

Why is my SaaS brand missing from AI recommendations when my competitor outranks me?

Because the competitor is not winning on authority alone. They are winning on being easier to retrieve, easier to resolve as an entity, and easier to corroborate from sources outside their own domain. DerivateX sees the same pattern across B2B SaaS accounts at $5M to $50M ARR: strong Google rankings on the money terms, and near-zero presence when the same buyer types the question into ChatGPT or Perplexity instead.

Here is the mechanism. When a buyer asks an engine for the best tool in a category, the model does not rank the web. It assembles a small candidate set, usually 3 to 4 brands recommended per AI category query, then writes a justification for each one. Getting into that set depends on whether your brand resolves cleanly as an entity in the category, whether retrieval surfaces passages about you that answer the specific question, and whether independent sources say roughly the same thing your site says.

A competitor with ten times your backlink profile has usually accumulated something else along the way: hundreds of third-party mentions, review profiles, podcast transcripts, comparison posts, Reddit threads and analyst write-ups. That corroboration layer is what the model is reading. The links were a byproduct. Chasing the links and ignoring the corroboration is how software companies spend two years publishing and stay invisible.

---

Does domain authority actually influence AI search recommendations?

Indirectly and weakly. Domain Authority is a Moz metric built to predict the ranking power of a site’s link signals in Google search results. It was designed to forecast organic rankings, and it does that job. It was never designed to predict whether a language model names you in an answer, and the evidence suggests it does not predict that well.

Three approved data points make the gap concrete. 28% of ChatGPT-cited pages have zero organic Google visibility. 80% of URLs cited by ChatGPT and Perplexity are not in Google’s top 100. Only 11% of domains are cited by both ChatGPT and Perplexity, which means even two AI engines disagree with each other more than they agree. If link equity were the dominant input, those three numbers would all be much higher.

Worth noting that the SEO tooling market has moved in the same direction, with several vendors now publishing brand-level measurement alongside their link metrics. That split is the right instinct. The signals that make a model recommend you sit closer to brand corroboration than to link volume.

Input	Weight in Google organic ranking	Weight in AI recommendation
Backlink profile and domain authority	High	Low to moderate, mostly as a trust proxy
Entity clarity (consistent name, category, facts)	Moderate	High, decides whether you enter the candidate set
Specific extractable evidence on page	Moderate	High, decides whether you get quoted
Third-party corroboration (reviews, communities, comparisons)	Moderate	High, decides whether the model trusts the claim
Content volume	Moderate	Low, more pages on the same thin claim change nothing
Category fit and answer relevance	Moderate	High, a narrow precise fit beats a broad vague one

This is not an argument against SEO. DerivateX runs SEO and GEO together, because the organic surface still feeds retrieval and still converts. It is an argument against assuming that the thing which moved rankings will move recommendations. With 40% of Google queries now showing Google AI Overviews and a 61% CTR drop when AI Overviews appear, the organic position you already hold is worth less traffic than it was, and the citation you do not hold is worth more.

---

What is the five-minute diagnostic for finding the real bottleneck?

Open a clean session in ChatGPT, Gemini and Perplexity. Run the same three prompts in each: the category prompt (“best [category] for [your ICP]”), the comparison prompt (“[competitor] vs [your brand]”), and the direct prompt (“what is [your brand] and who is it for”). Write down what comes back. DerivateX runs a wider version of this as the first hour of any engagement, because the fix is completely different depending on what you see.

What you observe	Failure mode	Root cause	First fix
Brand never appears, even on the direct prompt	Not retrievable	No indexed passages that match buyer phrasing, no third-party mentions to pull from	Build the corroboration layer and question-shaped pages
Direct prompt returns wrong category, wrong pricing or outdated facts	Misresolved entity	Inconsistent descriptions across your site, review profiles, directories and press	Standardize entity facts everywhere before publishing anything new
Named in passing, never in the shortlist	Mentioned, not recommended	No evidence the model can use to justify picking you over the default three	Publish specific, checkable proof tied to a narrow use case
Recommended, but the sources cited are competitors and review sites	Recommended, not cited	Your own pages are not the most quotable version of the claim	Rewrite key pages for extraction, add first-party data
Appears in ChatGPT, absent in Perplexity (or the reverse)	Surface gap	Each engine draws on a different source mix	Target the source types that specific engine favors

Three terms get used interchangeably and should not be. A brand mention means the model said your name. A recommendation means the model put you in the shortlist and gave a reason. A citation means the model linked your page as the source. These move independently. We have seen accounts where mentions tripled and recommendations did not move at all, because the mentions came from a listicle that described the brand wrongly. If you track one number, track recommendation rate on the category prompt, and understand which signals push a vendor onto an LLM shortlist before you decide what to fix.

---

What are seven ways to beat a higher-authority competitor?

Each of these maps to one of the failure modes above. DerivateX does not run all seven at once on any account, because the sequence matters more than the count. Pick the ones that match your diagnostic.

Narrow the category until you are the obvious answer

You will not win “best CRM”. You can win “best CRM for real estate investors doing off-market acquisitions”. Models reward precise category fit because precision reduces their risk of giving a bad answer. REsimpli became the most cited and recommended real estate CRM for investors in ChatGPT within 90 days, and the narrowness of that category is the reason it was possible at all. Write the category you can defend into your homepage, your review profiles and your comparison content, in the same words every time.

Publish evidence that can be quoted without rewriting

A sentence that says “scalable video infrastructure” gives a model nothing to work with. A sentence that says “encodes a 60-minute 4K source in under 9 minutes on the standard tier, priced per minute of output” gives it a reason to pick you and a string to quote. Specificity is the single cheapest lever available to a smaller brand, and higher-authority competitors are frequently worse at it because their marketing has been sanded down by committee.

Make your entity facts identical in every place a crawler can find them

Pull your description from your homepage, your G2 and Capterra profiles, your Crunchbase entry, your LinkedIn company page and your last three press mentions. If the category noun differs, the pricing differs or the ICP differs, the model has conflicting evidence and will hedge by recommending someone clearer. DerivateX treats this as housekeeping work, unglamorous and fast, and it often produces the first visible movement because it removes ambiguity rather than adding content.

Build corroboration on the surfaces each engine actually reads

46.7% of Perplexity top sources come from Reddit. Community threads, review platforms, roundups by other operators and podcast transcripts are not promotional channels in this context, they are evidence channels. DerivateX starts this work by listing which third-party pages each engine currently pulls from for your category prompts, because that list is usually shorter and stranger than a marketing team expects. One accurate, detailed thread where a real user explains why they chose you over the market leader does more for your candidate-set odds than ten more blog posts on your own domain. This is slow work and the honest timeline is months, not weeks.

Answer the comparison prompt yourself, including where you lose

Buyers type “X vs Y” constantly, and the page that answers it best often comes from a third party with no stake in the outcome. Write your own version, name the competitor directly, describe what they are genuinely better at, and state the specific scenario where they are the right pick. Models reward the balanced version because it reads as a source rather than a pitch. If your comparison page has no losing scenario in it, it will be discounted. When the engine is already favoring a rival, the diagnosis of why ChatGPT recommends a competitor usually comes back to this gap.

Shape pages for retrieval, not just for reading

Retrieval works on passages, not pages. A section that only makes sense after reading the previous four sections will not survive extraction. DerivateX rewrites key pages so that each section answers one buyer question on its own, with the answer in the first two sentences under the heading. Use the buyer’s question as the heading, keep definitions in plain “X is Y” form, and name your brand inside the passage rather than relying on “we”. This costs nothing and changes what a model can lift.

Create first-party data nobody else can report

Original numbers are the one asset a far more authoritative competitor cannot copy. Benchmark data from your own product, a survey of your user base, pricing transparency in a category that hides pricing, failure rates, implementation timelines. These get cited because there is no alternative source for them, which is also why proprietary data assets earn LLM citations at a rate ordinary thought-leadership content does not approach. Gumlet attributes more than 20% of monthly inbound revenue to AI discovery.

What should a software company change first?

Entity consistency first, then evidence, then corroboration, then volume. DerivateX sequences it that way because each stage is a prerequisite for the next one being measurable. Publishing a brilliant comparison page while three directories still describe you as a different category wastes the page, since the model has to reconcile conflicting signals before it can use what you wrote.

A realistic first 90 days for a B2B SaaS brand in the $5M to $50M ARR range looks like this. Weeks one and two: fix the entity record everywhere, build the prompt panel, establish the baseline. Weeks three to six: rewrite the eight to twelve pages that map to real buyer prompts, with specific evidence added to each. Weeks six to twelve: corroboration work on third-party surfaces, plus the first original data asset. Verito moved from an average position of 40 on Google to the first page, and is cited and recommended on ChatGPT and Google AI Overviews for 40 of their commercial hosting queries.

What DerivateX does not do is start with a content calendar. Volume is the last lever, not the first, and for most software firms arriving with strong rankings and zero AI visibility, volume is already not the constraint.

---

How do I measure whether recommendation behavior actually changed?

Fix a prompt panel, run it on a schedule, and score three things separately. DerivateX tracks mention rate, recommendation rate and citation rate across ChatGPT, Gemini, Perplexity, Claude and Google AI Overviews, on a panel of 50 to 200 buyer prompts chosen to match real purchase questions rather than keyword lists. 64% of marketing leaders are unsure how to measure AI search, and most of that uncertainty disappears once the panel is fixed and the baseline exists.

AI share of voice is a related but different number: it describes what proportion of all brand mentions in your category belong to you, while recommendation rate describes how often an engine actually places you in the shortlist for a specific buyer prompt. DerivateX reports recommendation rate as the primary metric, because a brand can hold a healthy share of voice through passing mentions and still lose every shortlist that produces a demo request.

The AI Visibility Score, or AVS, is the composite number DerivateX reports to a board: weighted recommendation rate across the engines and prompts that matter for your category, tracked against a named competitor set. The Citation Surface Map is the inventory DerivateX maintains of every third-party page an engine currently pulls from when answering your category prompts, which is what turns corroboration work from guesswork into a list. Citation Engineering is the methodology DerivateX uses to make the evidence on and around a brand quotable enough that models recommend it deliberately rather than by accident.

Two honesty notes on measurement. First, engine outputs vary between sessions, so a single run proves nothing and you need repeated sampling before you call a change real. Second, correlation is not causation here. We can show that recommendation rate rose after the entity, evidence and corroboration work shipped. No engine exposes its selection logic, so attributing the lift to one specific change requires holding other variables still, which DerivateX does where account timelines allow and flags where they do not. Anyone promising certainty on this is selling something.

On the revenue side, tag AI-sourced sessions and follow them through to pipeline. AI-sourced visitors convert at 4.4x the rate of other channels, which means a modest traffic number can carry a disproportionate share of demos. If you want to compare the tooling options before building this in-house, the breakdown of LLM visibility trackers for B2B SaaS covers what each platform can and cannot see.

---

What does this cost, and when is DerivateX the wrong choice?

DerivateX engagements start at $6,000 to $6,200 all in per month for Rank & Get Found, which is a $5,000 retainer plus $1,000 to $1,200 of off-site budget, on a 90-day pilot with no lock-in afterwards. Own Your Category runs $9,500 to $10,000 all in. Market Leader runs $14,000 to $15,000 all in with a six-month minimum. A standalone Diagnostic is $3,500, delivered in two weeks, and credits in full against month one if you convert within 30 days.

Now the part most agency pages skip. If what you actually need is visibility tracking rather than execution, buy a tool. Moz Pro’s published pricing lists $99 per month on the Standard plan and $299 per month on the Large plan as of 2026, with 150 tracked prompts on Standard and 600 on Large across GPT, Gemini, Google AI Mode and Perplexity, alongside the link and rank tracking Moz has built over 20-plus years. For a software company that has an in-house SEO team capable of doing the work and just needs the measurement layer, that is a far better use of $1,188 a year than $72,000 a year with us.

DerivateX is also the wrong choice in three other situations. If you are under roughly $2M ARR, the floor price is a disproportionate share of your marketing budget and you should do the entity and evidence work yourself from the diagnostic above. If your category genuinely does not get asked about in AI engines yet, measure first and spend later. And if your product has an unresolved retention problem, more shortlist appearances will accelerate churn rather than growth. We have turned down work on all three grounds.

Where DerivateX is the right fit: a B2B SaaS company between $5M and $50M ARR, with a defensible category claim, an existing SEO base that is no longer converting the way it did, and leadership that wants a number for AI visibility they can take to a board. Teams comparing providers can read the current survey of GEO agencies working with B2B SaaS rather than taking our word for the shape of the market.

---

Frequently asked questions

How do I improve AI search domain authority?

There is no AI search domain authority metric to improve, so treat it as three jobs. Make your entity facts identical across your site, review profiles and directories. Add specific, checkable evidence such as numbers, limits and pricing to pages that match buyer prompts. Then build third-party corroboration on review platforms and communities.

Can a low-authority site get recommended by ChatGPT?

It can, and this happens often. 28% of ChatGPT-cited pages have zero organic Google visibility, and 80% of URLs cited by ChatGPT and Perplexity are not in Google’s top 100. Retrieval rewards precise category fit, quotable evidence and independent corroboration, so a narrow, well-evidenced claim from a small domain routinely beats a vague one from a large domain.

What is the difference between a brand mention and a citation in AI search?

A brand mention means the model said your name anywhere in the answer. A citation means the model linked your page as a source. A recommendation means the model placed you in the shortlist with a stated reason. DerivateX scores all three separately because they move independently and require different fixes.

What does AI share of voice mean?

AI share of voice is the proportion of all brand mentions in a category that belong to you across AI engines. It differs from recommendation rate, which counts how often an engine puts you in the shortlist for a specific buyer prompt. DerivateX reports both, and treats recommendation rate as the number tied to pipeline.

How long does it take to change AI recommendation behavior?

Entity corrections can show up within two to four weeks because they remove ambiguity rather than adding content. Evidence and corroboration work typically needs 60 to 120 days before recommendation rate shifts measurably. DerivateX runs 90-day pilots for that reason and reports targets against a baseline prompt panel, not guaranteed outcomes.

Should I buy an AI visibility tool or hire a GEO agency?

Buy the tool if you have an in-house SEO team that can execute and only need measurement, since Moz Pro’s published Standard plan is $99 per month as of 2026. Hire an agency when the constraint is capability and capacity rather than data. DerivateX engagements start at $6,000 to $6,200 all in per month.

---

The mental model worth keeping

Think of AI search as a hiring decision rather than a ranking. The engine is filling three or four slots on a shortlist and has to justify each pick to a buyer who will hold it responsible for a bad one. Domain authority is a reference from one former employer. What actually gets you hired is a clear statement of what you do, specific proof you have done it, and several independent people saying the same thing. A competitor with ten times your authority has usually accumulated all three over a decade. You cannot out-publish that. You can out-specify it, in a category narrow enough that the engine has no better answer than you, which is the part of this that genuinely favors smaller software companies.

If you want to see which of the four failure modes applies to your brand before committing budget, DerivateX runs a free AI visibility audit with a 48-hour turnaround that returns your current mention, recommendation and citation rates across ChatGPT, Gemini, Perplexity and Google AI Overviews, benchmarked against three competitors: request the free AI visibility audit.