Reputation Management
One reputation. One benchmark. One trajectory, decomposed by signal, owner and impact.
Reputation is read by people and, increasingly, by machines, across reviews, search, media and AI answers. We run it as one programme and read it on one Online Global Reputation index: a single 0 to 100 score, benchmarked against the competitors you name, that breaks down to the signal that moved, the team that owns it and what it is worth to the business. It starts with a four to six week baseline.
In short
Reputation compounds when you run it as a programme, not a campaign.
Reputation decides whether you make the shortlist, whether a candidate applies, whether a regulator or a partner gives you the benefit of the doubt, and increasingly what an AI assistant says when someone asks about you. It is built in the same places every month: the reviews you earn, the answers you give, the pages and press that get cited, the way your leadership shows up. So we run it as a programme with three repeating jobs, build, protect and measure, and we read the whole thing on one index, so the effort is visible as a trajectory and every point of movement has an owner. The rest of this page explains each job in turn.
Why now
Reputation is now shaped across reviews, search and AI answers.
For a decade, reputation online meant the first page of Google and a star rating. In 2026 the same reputation is written in more places at once: the reviews people read, the results and summaries search returns, the answers assistants give, and what your own people say. Three shifts define the job now.
AI answers joined the checklist
The share of consumers who have used ChatGPT or another assistant to research a local business jumped from 6% to 45% between the 2025 and 2026 surveys, and 82% read the AI review summaries. For a growing share of buyers, part of the first impression is now assembled by a machine, from sources you can influence.
The star floor rose, and reviews expire
68% of consumers now need at least 4 stars to consider a business, up from 55%, and 31% want 4.5 or better. 74% only trust reviews from the last three months. A good rating from 2023 is no longer proof.
Trust moved closer to home
People trust their own employer (78%) and their own CEO (66%) far more than institutions in general, while nearly 7 in 10 fear leaders are deliberately misleading them. That makes your leadership and your people two of your most credible channels.
Reputation used to be a communications metric. It is now a commercial asset with a first impression written by machines, a floor set by consumers and a trust base that sits inside your own company, and it needs the same operating discipline as your media or your CRM.
The index
One index. One benchmark. One trajectory.
This is the centre of the offer. Most reputation reporting is a pile of platform dashboards nobody can add up. The Online Global Reputation index replaces it with a single 0 to 100 score, built from six signal families with transparent weights, benchmarked against the competitors you name, and decomposable to the signal that moved, the team that owns it and the business impact it carries.
Six signal families, agreed weights
Ratings and reviews, response and recovery, search reputation, AI-answer reputation, social and media sentiment, and employer and leadership. Weights flex by sector: a multi-location retailer leans on reviews, a bank leans on media and leadership. We agree them before the baseline and do not move them afterwards.
Always against a benchmark
A 71 means nothing on its own. A 71 against a competitor average of 66, rising three points a quarter, is a strategy. Every client sees its own trajectory against its own benchmark.
Social and media sentiment, scored like the rest
Press and social coverage is scored on tone, tier and reach, so a critical piece in a national daily weighs more than a neutral trade mention, and an employee's viral thread counts. This family measures the standing tone over 90 days; the sibling listening service watches its hour-by-hour shifts.
Decomposable to an owner
Every point of movement traces to a signal, a location, a product or a person, so the monthly read is a list of actions with names next to them.
- Rolling 90-day windows with smoothing, so one bad week never swings the board deck.
- Runs on our own BigQuery foundation, alongside whatever listening platform you already use.
- Quarterly board view that links movement to actions and to your commercial data.
Weights shown are illustrative. Yours are agreed at the baseline and stated in every report.
In our retail programmes the gap to the competitor average has typically sat in two places: a cluster of stores under the four-star floor pulling the reviews family down, and a search family dragged by an old press story still sitting in the knowledge panel and in two assistants' answers. The plan is written against exactly those two items. Within two quarters the store cluster clears the floor, the story is displaced by fresh, accurate sources, and the index moves by several points.
Individual client trajectories are shared, with permission, in the baseline conversation.
Share of AI answers is the share of a fixed set of buyer questions, in your markets and languages, on which each assistant mentions you, describes you accurately, recommends you, and cites a source you can influence. A standalone share-of-model dashboard stops at the visibility number. Here the same signal is decomposed to the sources behind it and weighted alongside reviews and search, so it becomes an action list and one number rather than another tool.
What it covers
Four jobs, one programme, one index.
Reputation management is broader than reviews and narrower than communications. Everything below feeds the index, and every capability has a place in one of four jobs.
Build reputation
Earn the proof people and machines find. Compliant review generation on Google Business Profile, Trustpilot, marketplaces, app stores and the vertical sites of your category; owned content that answers buyer questions; digital PR toward the outlets search engines and assistants cite; a policy-compliant approach to Wikipedia; executive and employer presence, including the Glassdoor and Indeed picture.
Protect reputation
Answer, correct and govern. Response lanes with service levels by severity, an escalation matrix, AI-drafted and human-approved responses in your tone; evidence-based flagging and takedowns; GDPR erasure and delisting where the basis exists; coordination with legal on defamation; brand-suitability controls in the media you buy; review themes sent back to product, operations and CX.
Shape discovery
Decide what appears when someone looks you up. Branded results, autocomplete, the knowledge panel, People Also Ask and AI Overviews; what ChatGPT, Gemini, Claude, Perplexity and Copilot say when a buyer asks about you, which sources they lean on, and a monthly audit that turns each answer into a fixable list.
Measure reputation
One number the board can steer by. The Online Global Reputation index, competitor-benchmarked and decomposed to signal, owner and business impact; exception reports for hundreds of locations, listings and product pages instead of dashboards nobody opens; the quarterly view that joins the index to branded demand, conversion, footfall and acquisition cost.
Live monitoring, spike alerts and crisis response are the other half of the discipline, and they get their own page. See how the two connect.
AI-answer reputation
What the machines say about you.
When a buyer asks an assistant whether you are trustworthy, the answer is assembled from sources: review sites, Reddit and forums, Wikipedia, press and your own pages. Only about 5% of citations overlap between assistants, so ChatGPT, Gemini, Claude, Perplexity and Copilot are separate channels with separate opinions, and the executive searches a bank cares about often run inside Microsoft 365, so through Copilot and Bing.
Measure it like a channel
A fixed set of buyer questions, in Portuguese, Spanish, English and whatever else your markets speak, run every month across the assistants and Google AI Overviews. We record mention, description, sentiment, recommendation and the sources behind each answer, and put the result into the index as its own signal family. Monthly matters: in one documented case Reddit's share of ChatGPT citations fell from roughly 60% to 10% in two weeks, and answers everywhere changed with it.
Change what the model reads
You cannot argue with a model. You can change what it reads. Accurate listings everywhere, fresh reviews, pages that answer the exact questions buyers ask, and digital PR toward the domains each assistant cites. Around 86% of the sources in AI answers are properties a brand can manage or influence.
Then re-measure
Movement shows within two to three monthly cycles for most brands. Deeper narrative repair, including the residue a crisis or a manipulated video leaves in answers long after the news cycle has moved on, takes longer, and we say so at the baseline. The stakes keep rising: AI agents now build shortlists on the buyer's behalf, so a brand described poorly is excluded before a human reads a single review.
- 45% of consumers have used an assistant to research a local business, up from 6%.
- 45% of marketing leaders cannot accurately measure their brand's visibility in AI answers (Semrush, 2026).
- Powered by our GEO practice and Wise GEO Compass, our answer-engine framework.
Sources: BrightLocal 2026 (US sample); Yext research on AI citations; Semrush AI Visibility Index 2026; 5W citation audit; Forrester on AI agents and shopping. Assistants are named as product families.
Ratings, reviews & response
The most tangible lever you have.
Reviews decide whether you make the shortlist, feed the AI summaries above the results, and tell you, for free, what to fix in the business. We run them as an operation: generation, freshness, response and root cause, to service levels.
Generate the right way
Ask every customer, at the right moment, on the platforms that matter in your category and country: Google Business Profile and Trustpilot for almost everyone, then Booking or TripAdvisor, Amazon or the app stores, G2 or Doctoralia depending on what you sell. Never gate by sentiment, never incentivise against the rules, never buy. Volume, recency and distribution all move the index and the AI summaries.
Respond to a standard
Every review is triaged into a lane with a service level: low stars, legal and safety within four business hours; complaints within a day; the rest within two. AI drafts in your tone with the facts of the case, a person approves, and sensitive cases are rewritten by hand. Speed and quality both count: 80% of consumers favour businesses that answer every review, and half are put off by generic replies.
Close the loop
Themes are mined per location and per product and sent to operations, product and CX every month, because most reputation problems are operations problems with a public face.
- +0.12 stars and 12% more reviews when businesses start responding, in peer-reviewed research.
- One star is linked to a 5% to 9% revenue difference for local businesses.
- 97% of consumers read reviews for local businesses; nearly half trust them like a personal recommendation.
Sources: Proserpio and Zervas, Marketing Science; Luca, Harvard Business School; BrightLocal 2026 (US sample).
Where reputation is built
The platforms differ by industry. The discipline does not.
Your reputation is written on a handful of platforms your buyers actually check, and which ones depends on what you sell. We map yours at the baseline, then run them all to one standard.
Google Business Profile, Trustpilot, Glassdoor
The three platforms almost every brand lives on. Google decides local search, the Maps shortlist and feeds AI Overviews; Trustpilot carries brand-level trust into branded search and assistant answers; Glassdoor and Indeed shape who applies. Each gets a response policy, a freshness target and an owner.
Store, product and marketplace ratings
Hundreds of Google and Apple Business Connect store listings, product ratings on Amazon, Worten and the marketplaces you sell through, app-store ratings for the loyalty app, and the retailer sites where CPG products live. Themes go back to store operations and R&D.
Booking, TripAdvisor, GetYourGuide, Viator
Guests rarely reach your site before Booking.com, Expedia, Airbnb, TripAdvisor and Google have ranked you; tours live on GetYourGuide and Viator, restaurants on TheFork and Google. Rating and response velocity move revenue directly, and the OTA algorithms reward both.
Branch reviews, complaint portals, app stores
Google reviews for branches and agencies, Trustpilot for the brand, the complaint portals your market watches (in Portugal, Portal da Queixa and DECO), app-store ratings for the banking app, and the coverage regulators read. Complaint handling is supervised, so response lanes align with it.
G2, Capterra, Clutch, LinkedIn
Buyers shortlist software and agencies on G2, Capterra and Clutch, verify the people on LinkedIn, and ask an assistant to compare vendors. Fewer reviews, higher stakes per review, and the executive layer weighs more.
Doctoralia, Google, the press
Patients check Doctoralia and Google before a clinic; citizens judge a public service on Google, in the press and increasingly on what an assistant says. Accuracy and response discipline under public attention matter more than volume.
The platform map is one of the first outputs of the baseline. Once these are running to standard, the search and AI-answer layers have something solid to point at.
The quieter levers
Press, Wikipedia, legal, brand safety, and your people.
Less visible than a star rating, and often where a bank, an insurer or a public institution wins or loses the argument. Here is what we actually do in each.
Press and digital PR
We map which outlets and domains search engines and each assistant cite for your category, then earn presence there: data stories, expert commentary, corrections and right of reply where a story is wrong. In Portugal that includes the press regulator route (ERC) and the editorial statutes that oblige a published correction.
Wikipedia and knowledge sources
No covert edits. We work through disclosed talk-page requests with reliable secondary sources, fix factual errors, and keep Wikidata and your knowledge panel consistent, because Wikipedia is consistently among the most-cited sources in ChatGPT answers.
Legal removals and authenticity
Evidence-based flags on reviews that break platform policy, GDPR erasure and delisting requests where the basis exists, and coordination with your counsel on defamation. We are honest about the odds: platforms remove a meaningful minority of flagged reviews, and delisting applies to personal data rather than to criticism you dislike.
Executive and employer reputation
A monthly read of what search and assistants say about your leadership, a Glassdoor and Indeed response policy shared between HR and communications, LinkedIn presence for the executives who want it, and your people equipped as credible voices. Ownership is agreed up front, so nobody is surprised by who answers a former employee.
Brand safety in your media is the fifth quiet lever: inclusion and exclusion lists, pre-bid verification and made-for-advertising site avoidance, run by our programmatic team so the brand is never funding the pages that damage it.
How we run it
Six steps, then a loop that never stops.
A baseline in weeks, a governed programme in months, a trajectory the board can read every quarter.
Audit
Every listing, star rating and review theme; branded search, autocomplete and the knowledge panel; a fixed-prompt audit across the AI assistants; media and Wikipedia; Glassdoor; competitors; legal and authenticity risk.
Baseline
Weights agreed by sector, the index computed for you and for the competitors you name, and the first 90-day plan written against the biggest movers.
Governance
Who owns what, response service levels by severity, an escalation matrix with named triggers, a compliant review-generation policy and the human-approval rules for AI drafting.
Build
Compliant review generation, listings and structured owned pages that answer buyer questions, digital PR toward the sources search engines and assistants cite, executive and employer presence where wanted.
Protect
Responses to standard, evidence-based flagging and takedowns, GDPR erasure and delisting where the basis exists, monthly AI-drift re-runs, and brand-suitability controls in your media.
Measure
The monthly index read, decomposed to owners, plus a quarterly board view that joins the index to your commercial data: branded search demand, conversion, footfall, acquisition cost. Where volumes allow we run holdouts, for example pausing review generation in a matched set of locations, to measure what the programme adds.
Steps four to six then run as a loop for as long as the programme lasts. The human-in-the-loop discipline for AI drafting and approvals follows the same governance we apply in our Wise Agentic Helm and Agentic Security frameworks.
Curious what the machines say about your brand today? Ask for a 20-minute AI-answer read-out ›
Governance & compliance
What we do, and what we refuse to do.
Reputation work attracts shortcuts, and regulators have caught up with them. The fastest way to lose a reputation in 2026 is to be caught managing it badly. We are transparent about the lines.
- Ask every customer for a review, on compliant terms, at the right moment.
- Flag policy-breaking reviews with evidence, and file GDPR erasure or delisting requests where the legal basis exists.
- Draft responses with AI and have a person approve them, with mandatory human rewrite on low stars, legal, safety and executive cases.
- Build owned and earned content that answers the questions buyers and assistants actually ask.
- Keep your data inside your platforms and our controlled data layer under a data-processing agreement, never used to train public models.
- Buying, faking or gating reviews, staff quotas, or asking for specific wording. Google blocked more than 290 million such reviews in 2025 and penalises the businesses behind them.
- Promising to delete legitimate negative reviews or articles.
- Covert Wikipedia editing or undisclosed advocacy.
- Astroturfing, sock-puppet accounts or manufactured praise in forums and communities.
- Anything that would fail the EU Digital Services Act, the UK DMCC fake-review ban or the US FTC rule.
Tool-independent, by design. We work inside the platforms you already pay for and add our own data layer for the index and the AI audits, so there are no licences for you to manage and no dependency on us to keep the data.
Two pages, one discipline
Where reputation management ends and listening & crisis begins.
The two services share a team and a data foundation, and they overlap on purpose. Here is how we draw the line, so you know which one you are buying.
Reputation Management this page
- Horizon
- Always on; 12 to 36 month trajectories
- The question
- What do people and machines find and believe about us, and how do we improve it?
- Outputs
- The index, review and response operations, search and AI-answer shaping, executive and employer work, governance
- Metrics
- Composite score, stars and velocity, response service levels, share of search, share of AI answers
- Owners
- Brand, CMO or CCO, with CX, HR and legal
Social Listening & Crisis Management sibling page
- Horizon
- Live; hours to weeks
- The question
- What is being said right now, is it abnormal, and what do we do in the next hour?
- Outputs
- Always-on monitoring, anomaly detection, escalation, holding statements, war-room, simulations, post-mortems
- Metrics
- Volume spikes, sentiment shifts, virality, time to detect, time to respond
- Owners
- Communications and PR, with legal and the C-suite
Where they meet. Monitoring feeds both services. A cluster of angry reviews or a hostile thread can become a crisis, and the crisis team picks it up. Every crisis then leaves residue in reviews, search results, Wikipedia and AI answers, and the reputation programme repairs it. Same team, same data layer, two rhythms. Explore Social Listening & Crisis Management ›
Proof
Eight years of reputation programmes, across the sectors where it matters most.
More than eight years of programmes, for more than thirty clients, from multi-location retail and consumer goods to banks, insurers and public institutions, where a single review, article or AI answer carries regulatory weight.
Platform-neutral
No listening vendor, review platform or media network owns our recommendation. We pick what serves your index and hand it back if we part ways.
Our own data layer
The index, the AI-answer audits and the benchmark run on our BigQuery stack, built by the same team that runs analytics and measurement for our clients.
Search, GEO, social, CX in one house
The levers that move reputation, from SEO and answer-engine work to social, content and customer experience, sit inside the same agency.
The lift figure is an average across our client programmes; individual trajectories depend on starting point and sector and are shared, with the client's permission, in the baseline conversation.
Start with a baseline
Four to six weeks. Your reputation as people and machines see it, in one number.
Ways to work with us
Plug it in where you need it.
Reputation baseline audit
A fixed-scope, four to six week engagement: the full audit, the index against your competitors, and the ranked 90-day plan. Stands on its own or starts the programme.
Always-on reputation management
The full loop, run to service levels, with the monthly index read and the quarterly board view. Sized by locations, markets and the signal families in scope.
With your comms, CX or PR partner
We bring the measurement, the AI-answer practice and the review operation next to the communications or CX team you already have, with one index everyone reports against.
Connected
Six other teams move the index with us.
The levers that move reputation live across the agency, and each one is wired to the index.
FAQ
The questions buyers actually ask.
What do ChatGPT, Gemini, Claude, Perplexity and Copilot say about our brand, and can you change it?
How long does it take to improve a Google rating or an online reputation?
How much does reputation management cost?
What is a good Online Global Reputation score?
What is the difference between reputation management and social listening or crisis management?
Can you remove negative reviews, articles or search results?
Do you use AI to write our review responses?
Does reputation management work for banks, insurers and other regulated companies?
How do you stay compliant with Google's review policies and the fake-review rules in the EU, UK and US?
How does this scale to 50 to 500 locations, or across marketplaces and app stores?
Do you cover executive and employer reputation too?
Which tools do you use, and do we need to buy licences?
How do you measure share of AI answers?
How do you link reputation to business results?
Where do the numbers on this page come from?
Reputation Management
Find out what people and machines say about you, then move it.
In four to six weeks you will know what reviews, search and AI assistants say about you, where you stand against your competitors, and what to fix first.
Book your baseline →
