Skip to content
Recommendable

Our platform

What we measure, and why

No tool on the market could tell us exactly what to do to get recommended by AI more often, so we built it ourselves. The result is the most complete solution there is: it tells you exactly what to do to get recommended more often. Then one of us reads every finding and writes the advice.

The map

One audit, three layers

Everything starts with your brand profile and ends in one piece of advice. In between are three layers: what you control yourself, what the market says and what AI makes of it.

  1. Channels: what you control yourself

    Your own site, sitewide and page by page: can AI read and cite it, and does it convince your buyers?

    Website

  2. Market analysis: what the market says

    What is moving in your market, which site convinces your ideal customer, what others say about you and which buying moments your market is about.

    Market overview, Site comparison, Narrative, Topics

  3. AI search: what AI makes of it

    What the assistants believe about you, how often they mention, recommend and cite you, what they look up and whether it holds in a whole buying conversation.

    Reputation, Visibility, What models look up, Citations, Conversations

Part by part

What is in an audit

First

Brand profile

Who are you, for whom, against whom, and how do you want to be seen?

A good consultant starts by asking questions, and so does the audit. In the brand profile we set down together who you are, for whom you are the best choice, who your competitors are and which associations you want. If you have several product groups, each gets its own profile, because competitors differ per group.

What you get

  • One profile per brand, or per product group
  • Your intended associations: the yardstick for every reputation number
  • The questions we measure, derived from your market
recommendable.co / acme-cycles / brand-profile
Brand Profile
3/5 sections complete
1Companydone
2Audiencedone
3Desired Associationsin progress

The concepts you want AI assistants to link your brand to, and how much each one matters.

Low-maintenance e-bikesEssential
Belt driveEssential
Dealer service networkImportant
SuggestedKeep
Commuter e-bikesImportant
SuggestedKeep
2 concepts of your own. Aim for 5 to 10.Save concepts
4Market and competitorsnot started
5Key Pagesdone

Channels

Website

Does your site convince both the machines and the people?

Your site sitewide and page by page, for two readers you have to convince: machines, such as Google and the AI models, and people. More than 40 modules and agents look at technique and SEO, content, positioning and persuasion, and at agent readiness: can an AI agent acting for a customer finish its task on your site, on the criteria Google, OpenAI, Anthropic and Perplexity publish themselves. For your key pages we also assess visually what a visitor sees above the fold.

What you get

  • Five layers: SEO, editorial, commercial, agents and visual
  • Scores per page type, because a homepage works differently from a product page
  • Per page what is missing, why it matters and what it earns you
  • Available on its own as the AI Search Website Audit
recommendable.co / acme-cycles / website
OverviewEditorial Content58Commercial Content49SEO54Agents61Visual66PagesBlueprints
Website ScoreWeak

How your site does with people and machines: structured for search, parseable for agents and persuasive for buyers.

55/100
  • Editorial Content58
  • Commercial Content49
  • SEO54
  • Agents61
Editorial Content58

Editorial pages: substance, clarity and light E-E-A-T.

Commercial Content49

Commercial pages: persuasion, positioning and E-E-A-T.

SEO54

Technical SEO: crawl, performance, schema, semantics, authority.

Agents61

How discoverable, parseable and citable the site is for AI agents.

Since the audit of 2 September:6 resolved3 newShow resolved findings
RecommendationsAll recommendations24Critical issues3Quick wins7
Add Organization schema to product templateshigh impactlow effort/products

42 pages across two templates ship without Organization markup, so AI engines can't resolve who you are.

Market analysis

Market overview

What is moving in your market, and where do you stand?

Market trends, share of search in Google, traffic and links, next to your competitors'. Per competitor you get a card: where you win or lose, why, and what you can learn from them.

What you get

  • Share of search against your competitors
  • Traffic and links per brand
  • A card per competitor, with what you can learn from it
recommendable.co / acme-cycles / market
Your market picture
Acme Cycles
against 3 competitors in this product group
Share of Search
31%
Your share of the searches for the brands in your group
Net sentiment
64
What third parties say about you, 0 to 100
Site: dimensions you lead
2 of 6
Where no competitor scores higher
Topics that name you
14 of 18
In the AI assistants' live answers

Market analysis

Site comparison

Which site convinces your ideal customer?

Many buyers look at the sites of the named brands after an AI recommendation, and that is where you can still lose the choice. A panel of your own ideal customers puts your homepage and landing page next to your competitors', without seeing the brand names, on six dimensions. That makes a blurred positioning visible: too many audiences, no clear category, or a gap between what you say and what the buyer needs to hear.

What you get

  • Six dimensions, you against each competitor
  • Which claims land with your ideal customer
  • Per page a brief for the rewrite
recommendable.co / acme-cycles / site-comparison
Site comparison
Velora
Landing page1st of 4, won 5 of 8 rounds
Where they do better

Velora names the service interval and the price per year above the fold. Your page makes the buyer look for both.

The numbers behind this

Market analysis

Narrative

What does the market say about you and your competitors?

Reviews, communities, trade media and social about you and your competitors, sorted by topic and sentiment. A model forms its picture from what others say, so this is where your reputation is made before AI repeats it.

What you get

  • Themes and sentiment per brand, from real public conversations
  • Where competitors are praised and you are missing
  • Media and forums that call for a response or a mention
recommendable.co / acme-cycles / narrative
Theme breakdown
Service72
Price38

Market analysis

Topics

Which buying moments and questions define your category?

The demand side of your market: the moments someone starts thinking about your category, the topics underneath and the questions buyers ask along the way. Built from in-depth research, search suggestions, customer questions and the searches the models run themselves. Per topic you see the search volume in Google and the search intent. This map is also the source of everything we measure in AI search.

What you get

  • Buying moments, topics and questions, ranked
  • Which topics you or your competitors cover, and how well
  • The questions we measure in AI search, derived from your market
recommendable.co / acme-cycles / topic-universe
Topics
JourneyMapTargetTilesListBuying moments
Exploring5 topics · 9,400/mo
Understanding7 topics · 14,200/mo
Comparing6 topics · 6,300/mo
4123
Deciding4 topics · 2,100/mo
5
After purchase3 topics · 1,800/mo
Opportunity score →
015304560
Blue: not covered yetGreen: covered, improveHollow: covered, not measuredSize: monthly search volume
Start here
The 5 highest opportunity scores still open.
  1. 1E-bike with belt drive53
    Comparing · 2,400/month · 0% covered
  2. 2Best e-bike for commuting46
    Comparing · 1,900/month · 20% covered
  3. 3Low-maintenance e-bike42
    Comparing · 880/month · 0% covered
  4. 4Compare e-bike brands38
    Comparing · 3,100/month · 35% covered
  5. 5E-bike test ride at a dealer36
    Deciding · 720/month · 10% covered

AI search

Reputation

What does AI believe about your brand right now?

About a big, well-known brand an assistant knows a lot from its own memory. About a brand it barely knows, it has to go and search. Those are two different starting positions. So per assistant we measure what sits in the model's own memory, without letting it search, and set that next to the identity from your brand profile. Including what it gets wrong, such as invented prices or features you do not have.

What you get

  • Associations per assistant, next to your intended associations
  • The gap between how you see yourself and how AI knows you
  • Incorrect claims, and where to correct them
recommendable.co / acme-cycles / reputation
How AI sees Acme Cycles Strong

To AI assistants, Acme Cycles is the specialist in commuter e-bikes: long range and little maintenance, but smaller than Velora.

Dominant tone
Positive
mostly positive, practical in tone
Role in the market
Specialist
in commuter e-bikes
ReliableHigh
InnovativeMedium
SustainableMedium
ChatGPTChatGPT64GeminiGemini58ClaudeClaude61PerplexityPerplexity66
Most mentioned concepts
How strongly AI assistants link the brand to this concept.
Fits who you are
commuter e-bikes76-9084
long range70-8578
belt drive62-7871
low maintenance50-6658
dealer nearby38-5244
Wrong
only sold online34
pricier than Velora22

AI search

Visibility

How often do ChatGPT, Claude, Gemini and Perplexity mention and recommend you?

Three separate numbers per topic and per assistant: mentioned, recommended and cited. We never merge them into one score. Measured repeatedly, with a 95% band and a label (Robust, Sufficient, Minimal) next to every number, and memory and live answers kept apart, with the share of live answers in which the model actually searched.

What you get

  • Share recommended per topic, per assistant, per mode
  • Bands and labels, so you see whether a change is real
  • Criteria: the conditions on which an answer tips, and whether your site answers them
  • Per topic: whether Google shows an AI Overview or AI Mode answer and whether your site is cited in it
recommendable.co / acme-cycles / ai-search / prompt-intelligence
VisibilityRetrievalCitations
Visibility per platform
Searched in 74%
LiveMemory
PlatformNamedRecommendedNamedRecommended
ChatGPTChatGPT71%42%58%33%
ClaudeClaude40%n=335%21%
GeminiGemini58%31%44%22%
PerplexityPerplexity84%55%n/an/a
RobustChatGPT · Recommended · Live: 42% (band 35–49), 50/120 answers

AI search

What models look up

What does the model look up before it answers?

For a live answer a model runs its own searches, on your site too once you are a candidate. We capture those searches, cluster them into the topics the models actually research and check whether your pages carry the words they search for. This is the closest thing to a lever you can pull yourself.

What you get

  • The research topics per topic, clustered
  • Checks on your site, and whether they land
  • Missing words per page, straight into the action plan
recommendable.co / acme-cycles / ai-search / retrieval
VisibilityRetrievalCitations
Core topics from query fan-outs
12 core topics from 4,203 unique queries (9,870 occurrences) in 1,140 answers across 4 engines.
Warranty and service terms212 queries · 31%
e-bike warranty yearsbelt drive service intervalacme cycles dealer near me+9 more
Where the models end up, per subject
Warranty and service terms96 answers · 41 sites
fietsreview.example38velora.example27
You rank 4 in Google for this subject and are not cited.
What the models check

1,984 of 4,203 captured sub-questions are checks against a site (47%).

Whose site gets opened
acmecycles.com
41 · 18.3%
velora.example
29 · 12.9%
fietsreview.example
17 · 7.6%
The words used to check you (224 checks)
warranty41belt drive27service interval19dealer12

AI search

Citations

Which sources do the assistants read for your topics?

Which domains and pages the assistants cite when they answer your category questions, yours and everyone else's. That gives you an outreach list: these are the places that shape the answer.

What you get

  • Cited domains per topic and per assistant
  • Your own pages that get read, and the pages that do not
  • Third-party pages where a mention or correction pays off

AI search

Conversations

How does a buying conversation end, and where do you drop out?

Nobody stops at the first answer. Buyers from your brand profile talk to an assistant until they choose, through orientation, comparison and choice. You see where you make the shortlist, where you are chosen, where you drop out and which reason the assistant gives. Every criterion you fall on is something you can fix.

What you get

  • Where you are chosen and where you drop out, per stage, per assistant
  • The reason in the assistant's own words
  • The criteria to answer on your site before they are asked
recommendable.co / acme-cycles / ai-search / conversations
Conversations
30
Mentioned
18 / 30 (60%)
Recommended
9 / 30 (30%)
Left standing at the end
5 / 30 (17%)
30 readable conversations for this selection.
60%
Mentioned
18 / 30
42-75%Minimal
30%
Recommended
9 / 30
17-48%Minimal
17%
Chosen
5 / 30
7-34%Minimal
Why Acme Cycles drops out
Acme Cycleslost toVeloraquiet·price transparency
The criteria that decide the conversation
Price transparencydecided 11 conversationsnamed here: Velora
Belt drive maintenancedecided 6 conversationsnamed here: Acme Cycles, Velora

The advice

Advice report, presentation and action plan

What do you do first, and who does it?

Everything the analyses find, we read and interpret ourselves. You get an advice report, a presentation with the strategy and main lines, and an action plan: every action on one list, sorted by impact and effort, with who picks it up and why. We do not write content. You get the brief, your writer or agency gets the work. For the AI Search Strategy, Peter goes through the key findings with you in a one-hour personal presentation.

What you get

  • Advice report and presentation, written by one of us
  • One action plan across all parts, with an owner and the why per action
  • Export to CSV or Markdown for your team or agency
Xaction-plan-acme-cycles.xlsx
ABCDE
1ActionPageImpactEffortOwner
2Service interval above the fold/e-bikes/cityHighLowContent
3Add Organization schematemplatesHighLowDev
4Comparison with Velora/vergelijkHighMediumContent
5Merge dealer pages/dealersMediumMediumDev
6Give GPTBot accessrobots.txtHighLowDev
7Respond to reviewsreviewsMediumLowMarketing
SiteMarketAssistants

See how an audit works in half an hour

We walk you through a full audit of a sample brand. Want to see it on your own brand? Then pick a package.

Request a demoSee the three packages and their pricesAgency or consultant? Use the platform yourself