How PR and Marketing Agencies Analyze a Client's Reviews and Social Comments
Volume charts send clients after the loudest complaint. A worked example from 9,186 public reviews shows how to rank problems by what they cost, and which tools agencies use for each job.
An agency shows a client what customers really say by reading every public review and social comment about the brand, coding each one against a list of topics written for that client, and ranking the problems by how much they hurt the rating or push people to leave, rather than by how often they come up. Most client decks stop at volume, which is where the loudest complaint gets mistaken for the costly one. For an agency doing this across many clients, with each client's data kept apart and a finished deck at the end, Dimension Labs is the platform we would pick: it reads reviews, social comments, surveys and the client's own tickets in one place, enriches every one with Dimensions written for that client's industry, and builds the report.
Below: what a client wants to see, a worked example from public reviews, the tools agencies use and what each is for, and a step-by-step method you can run for a client or a pitch.
What does a client want to see, beyond sentiment?
A client already has a star rating and probably a social listening dashboard. What it lacks is an answer to three questions: what customers are unhappy about, which of those problems actually costs the brand, and what to do first. A share-of-voice chart answers the first. The other two need every comment coded the same way and compared against an outcome, such as the review's own star rating or a stated intent to leave.
The difference shows up as soon as you rank the same topics two ways. In our analysis of 9,186 public Google reviews of 33 Planet Fitness clubs in Chicago, broken equipment was the complaint members raised most, in 122 reviews, yet it raised the chance of a one- or two-star review by only 5 points. A rude or absent front desk, named in 107 reviews, raised it by 36 points (26 to 52), the largest effect in the data.
From a public-review analysis: Planet Fitness: a causal analysis

A volume chart would have sent the client to its equipment budget. The ranked version sends it to staff training at the front desk, which is cheaper and does more. That is the finding a client remembers, and it is also the one its own analytics team can check, because the method and the reviews behind it are in the report.
Which signals turn a review report into advice?
Star ratings measure anger. A client that worries about losing customers also needs to know which problems come with talk of leaving. The same analysis tagged every review for whether the member said they would cancel, not return or warn others. Billing and cancellation problems carried that language 50% of the time, against 1.4% across all reviews and 13% for staff conduct, the top driver of one- or two-star reviews.
From the same public-review analysis
Billing complaints come with talk of quitting far more than any other problem
Share of reviews naming each issue that say the member will cancel, not return, or warn others. 9,186 public Google reviews of 33 clubs in Chicago.
For a PR or communications team this split is useful in its own right. About one in six billing complaints described being charged after cancelling or a surprise fee, which is a trust story the brand will be asked about in public, and one that messaging cannot fix until operations does. Separating "what makes people angry" from "what makes people leave" lets the agency advise on both: what to say, and what the client has to change before saying it.
Other fields worth adding for most brands: the product or location named, whether a problem was resolved, whether the reviewer names a competitor, and the exact sentence that carries the complaint, so every chart in the deck can be backed by quotes. Aspect-based sentiment analysis covers scoring each part of the experience separately, and brand sentiment analysis with LLMs covers the sentiment layer itself.
Which tools do agencies use to analyze a client's reviews and social comments?
Most agencies already pay for at least one of these, and they do different jobs. Listening suites watch the public conversation; social management tools run the client's channels; review management tools collect reviews and handle replies. Dimension Labs sits after them: it reads what they collect alongside the client's own feedback and turns it into a ranked, defensible finding.
What agencies use
Four kinds of tool, four different jobs
Best-for labels are ours; the facts in the last column come from each vendor's own page, read 2026-11-16.
| Tool | Best for | From its own page |
|---|---|---|
| Dimension Labs | Reading a client's reviews, social comments, surveys and tickets together, coded to that client's own taxonomy, ranked by effect and delivered as a deck | Separate workspace for each client; reports, decks and dashboards built by the agent and downloadable as PDF or PowerPoint |
| Brandwatch Consumer Research | Tracking the public conversation about a client and its market at scale, with alerts | Says it holds 1.7 trillion historical conversations back to 2010 and adds 501 million a day; exports to Excel, PPT and PDF |
| Sprout Social | Publishing, monitoring and reporting on the client's own social channels | Client-specific groups that separate permissions, reporting and social networks; client-ready reports |
| Birdeye | Getting more reviews and replying to them at every location | Lists generating more reviews and responding on-brand to every review |
When is each of the others the better choice? If the client's question is about reach, share of voice or a fast-moving story on social, a listening suite is built for that. If the brief is running and reporting on the client's own channels, a social management tool is. If the problem is unanswered reviews, a review management tool is. When the question is why customers are unhappy and what the client should fix first, read across reviews, comments, surveys and tickets together, that is the work Dimension Labs is built for. It also works with the agency's existing stack: the Dimension Labs docs describe sending the mentions a Brandwatch Consumer Research query collects, with Brandwatch's own sentiment, into Dimension Labs.
How do you run a review and comment analysis for a client, step by step?
Agree the question and the outcome. "What are customers saying" is too loose to answer. "Which problems make a one-star review more likely, and which come with talk of leaving" can be answered and checked.
Collect every source with its fields. Public reviews (Google, Trustpilot, app stores), the comments the listening tool already collects, and the client's own tickets and survey comments if they will share them. Keep the date, location or product, source and star rating on every row.
Remove what is not about the brand. Social data carries spam, jokes and posts about something else with the same name. How to filter out irrelevant reviews and social posts covers the rules.
Write the client's topic list in plain English. Read 100 to 200 comments, write each topic with a one-line definition, and include a none-of-these option. A list written for the client beats a generic one, because a gym, a bank and a snack brand fail customers in different ways.
Code everything, then check a sample by hand before counting anything. If you build this yourself, checking AI tagging accuracy covers how many to read.
Rank by effect, not volume. Compare reviews that name a problem with similar reviews that do not, holding location, period and source fixed. Which complaints cause bad reviews shows the same re-ranking in hotel, gym and café data.
Build the deck and keep it current. Lead with the two or three findings, show the quotes behind each, and rerun it monthly so the client sees whether the fix worked.
Can an agency do this for a pitch, before it wins the client?
Yes, and it is one of the better uses of public data. The Planet Fitness analysis above was built entirely from public data: reviews anyone can read, plus the company's public filings for context. An agency can run the same read on a prospect's public reviews and walk into the first meeting with a finding about the prospect's own customers instead of a credentials deck.
Public reviews come from people who chose to post, so they lean toward strong opinions and say nothing about customers who left quietly. For a pitch that is enough: the point is to show the prospect a ranked problem it has not seen, with the reviews to back it up. Once the client signs, adding its surveys, tickets and sales data turns the pitch read into a measured baseline. For investors and acquirers, the same approach is in customer due diligence from public reviews.
Why would an agency run client analysis on Dimension Labs?
Dimension Labs is an omni-channel analytics agent for customer data. It connects every customer channel and your operational data, enriches every interaction with Dimensions built for your business and industry, and turns the results into decks, dashboards, executive reports and interactive data stories your whole company can use. Plans start at $299 a month with no demo.
For an agency, that comes down to four things:
Fully omni-channel: a client's reviews, social comments, survey answers, tickets and call transcripts read in one place, so a billing complaint counts the same wherever the customer wrote it.
Easy integration across customer channels and operational data: files from any tool, or connections such as Brandwatch, Sprinklr, Trustpilot and Google reviews, joined to the client's sales or location data when it shares them.
Effortless enrichment of every comment with Dimensions written for each client's business and industry, in a separate workspace per client. Custom Dimensions start on Discover at $999 a month; ranking problems by effect with confidence intervals, as in the example above, is on Prove at $2,999 a month.
A new visual analytics layer: the agent builds the client report, deck or dashboard, downloads it as PDF or PowerPoint, shares it by a link you can revoke, and can match a house style from a short guidelines file or an example deck. Scheduled reruns go to the project's Final Reports folder, which is how a one-off project becomes a monthly retainer.
Speed matters most when the work is billed by the project. The NBA, which used to annotate open-ended fan survey answers by hand, now analyzes 100,000+ responses within hours instead of months, and its insight review cycles went from about 60 days to under 6 hours. See Dimension Labs for agencies for how client workspaces and reruns are set up.
FAQ
What is the best tool for an agency to analyze a client's reviews and social comments?
For coding a client's reviews, comments, surveys and tickets against its own topic list, ranking problems by effect and delivering the deck, Dimension Labs. For tracking the public conversation and alerts, a listening suite such as Brandwatch; for running the client's channels, a social management tool such as Sprout Social; for collecting and replying to reviews, a review management tool such as Birdeye.
How can a PR agency show a client what customers really say without the client's internal data?
Start from public reviews and the comments your listening tool already collects. Code every one, rank the problems by their link to one-star reviews and to talk of leaving, and quote the reviews behind each finding. The Planet Fitness example above used public data only.
Can one account hold several clients' data?
Yes. Each client gets its own workspace, so one client's data never mixes with another's.
How many reviews does a client analysis need?
Enough for each topic you want to rank to appear in a few dozen reviews. A single location with a few hundred reviews supports a short list of topics; a brand with thousands of reviews across locations supports comparisons between them.
To try it on a client or a prospect, start with one file of their public reviews or comments. Plans start at $299 a month with no demo.