Reputation Management When Litigation Makes the News

Dil Phagura, Co-Founder and CEO of AI Reputation

A business owner called me on a Tuesday afternoon because a complaint filed against his company the previous day had become a headline that morning. He had known about the dispute for eight months. He had not expected to read about it in a publication his customers actually open.

His first question was who had talked to the reporter. Nobody had talked to the reporter. The filing was the tip, and it had been sitting in a public system since the clerk accepted it.

The distance between what he thought had happened and what had actually happened set the shape of the next eleven months. It also explains why so many owners in his position spend their first month working on the wrong object.

The first screen held more than the article

I ran his name and the company name the following morning, before we talked about strategy. The article was there. So were five other pages that had not existed the week before.

Two were docket pages on commercial sites that republish court filings as their own inventory, each one carrying the case number, the parties and a running list of documents. One was an aggregator profile that had bolted the company name, the filing date and a single line of summary together into a page about the business. Two were short legal news items assembled from the same filing by writers who had spoken to nobody.

He had read one of those six pages. The other five were the ones built to sit still, and none of them had a byline anyone could call.

None of the five named a person either. They named the company, the case number and a date, which is enough for a customer who already heard something to decide that the something is real. A record copy does not have to be read closely to do its work.

A filing is one document. A result set is a population of copies of that document, and the copies do not end when the case does.

Why does a court filing reach the news within a day?

Because the filing is public the moment it is docketed, and the systems watching for new filings are automated. A reporter does not need a source when the record itself is already searchable.

The federal courts run an electronic public access service that gives the public instantaneous access to more than 1 billion documents filed at all federal courts. That volume is the whole explanation. A system holding that much material has to be machine readable, and anything machine readable can be watched by a program that never sleeps, never takes a weekend and does not need a reason to be interested in your company.

The same service lets anyone conduct nationwide searches to determine whether or not a party is involved in a federal case. So the second thing a reporter can establish in about a minute is whether this is the first time. That answer shapes the story that gets written more than the contents of the filing do.

State and county systems vary, and plenty of them are slower and far clumsier than the federal one. The direction of travel is identical in all of them. The record is the primary document, and everything written about it afterward is a copy at one remove from a page that was already public.

Do court records stay online after a case ends?

The record stays, and so do the copies that commercial sites made of it while the case was active. Each copy is a separate page on a separate domain holding its own position.

This is where the arithmetic goes wrong for almost every owner I talk to. They count one problem, because they read one article, and they plan a response sized for one page. The budget, the timeline and the anger are all calibrated to a single target.

What is actually in front of them is a set, and the set was still growing while they were reading. A docket page is not published once and then left alone. The site adds each later filing in the same matter to the same page, so that page gets heavier while the case runs rather than lighter.

He asked me once how a customer would ever find a docket page. They do not go looking for one. They type the company name, the way anybody does.

Nothing I do removes a court record. The record is a public document and it is supposed to be a public document, which is the part of this field that people find hardest to hear from someone they are paying.

Whether to answer a filing in public, or at all, is a legal question before it is a search question. That decision belonged with his lawyer and I said so in the first conversation, which is the only sensible place for a reputation adviser to stop talking.

The order the work ran in, and what did not move

We did three things, in this order, and the order mattered more than any one of them. The first was the least interesting: a written inventory of the six pages, what each one actually was, and who owned it. It took an afternoon and it ended the conversation about suing anybody over a summary.

The inventory is also what told us which of the six would keep mattering. Two of the pages sat on sites that had already stopped updating much of anything, and one of those had gone quiet entirely by the fourth month. Guessing which two those would be, without writing the list down first, was not something either of us could have done.

The second was to set the target on the two docket pages rather than on the article. That is the step owners argue with, and it is the one I am most confident about. The article is what stung, and a docket page is what a referred customer meets a year later when the sting has long since stopped mattering to anyone but him.

The third was to publish accurate and current material about the company in the places that already carried its name, and to keep publishing it while the matter ran. The reasons that kind of material holds a position for a name query are a subject of their own, and I have written about them in more detail elsewhere.

The work ran eleven months. What changed over that time was the material standing beside the case, which is the only input either of us controlled. What did not change was the two docket pages, and I had told him in the first week that they probably would not.

They did not move, and the mechanism is plain enough. Each of those pages holds a copy of the primary document and gains a line every time something new is filed in the matter, so the version a reader finds in month ten is a fuller page than the one that worried him in month one.

That is the honest shape of this work when litigation is the source. The set gets managed, the record does not get edited, and anyone who tells a business owner otherwise is selling a result that the courts themselves make impossible.

He asked near the end whether he should have seen it coming. The filing, no. He could not have known the day it would land, or which outlet would decide it was worth a paragraph.

The six pages, yes. The shape of that result set was predictable from the moment a serious dispute existed, and he had eight months of notice that he spent entirely on the dispute itself, which is what anyone would have done.

I am not sure that is a failing so much as a design problem in how business owners are advised. Nobody who handles a commercial dispute is paid to think about what the filing will do to a search result, and the person who is paid to think about it usually gets the call the morning after.

Nobody gets to decide what the courts publish. Everybody gets to decide what else there is to read.

Dil Phagura is the Co-Founder and CEO of AI Reputation. Connect on LinkedIn at https://www.linkedin.com/in/dilphagura

Photos by Nhan Hoang and Bryan White on Unsplash

Sources

Public Access to Court Electronic Records
Find a Case

How to Respond to a Negative Review Without Making It Worse

By Dil Phagura, Co-Founder and CEO of AI Reputation

The reply field is open and there are two drafts sitting in it. One runs four paragraphs and corrects the three things the review got wrong. The other runs two sentences and thanks the customer for saying something.

That is the whole decision, and it usually gets made inside ten minutes by somebody who is still angry. The long draft tends to win, because it feels like the one that sets the record straight.

Both drafts are permanent in a way the review is not. The review is one voice on a page of voices. The reply is the only thing on that page with the name of the business attached to it.

Most owners want a third option, which is for the review to be gone, and by the time the reply field is open that branch has already closed. What remains is the two drafts.

A public reply has one real audience, and that audience is not the person who wrote the review. The reader who decides anything is the next customer, working down the page weeks later, choosing whether to call.

What does replying to a negative review cost?

It costs you the top of the page. A reply sits directly under the review, it is read by everyone who reads the review, and it tells a stranger how this business behaves when somebody criticizes it.

That is a cost worth paying when the reply is short and specific. It stops being worth paying the moment the reply starts arguing. An owner who answers a complaint about a delivery with a paragraph about how long the firm has been in business has told the reader nothing except that the complaint stung.

Length is the tell a reader picks up first. A reply is read in sequence with the review, immediately after it, so a long answer to a short complaint reads as a business that needed to win rather than one that fixed something. Nobody counts the words, and everybody notices them.

Four controlled experiments on hotel reviews led the researchers to conclude that businesses should prioritize responding to implicit negative reviews that involve external factors. That work was done in one language and one industry, so it is a direction rather than a rule. The direction holds up anyway, because what it describes is a review a reader cannot settle without hearing from you.

The review is a complaint about one visit, and the reply is a standing statement about the business.

The reviews worth answering are the ones carrying something checkable. A missed appointment, a charge, a date, a product that arrived in the wrong condition: a reader has no way to resolve any of that alone, which is exactly why your sentence matters there.

Is it better to leave a negative review alone?

Sometimes, and more often than most owners believe. An unanswered review stays one opinion among several, while a reply can turn it into a thread that a reader scrolls through line by line.

The silent branch carries its own price, and it is worth naming rather than waving away. A complaint that names a date, a product and a member of staff reads as checkable, and leaving it alone lets a reader conclude that nobody at the business thought it was worth the time.

The cost of silence also arrives late, which is what makes it easy to miss. A review left alone in March is still the newest thing anyone reads in September, and by then the person who could have answered it has usually forgotten the visit.

Work at Colorado State University found that an unfavorable comment can sway the experienced consumers that businesses want most. Those are the readers who already know the category, read closely and compare. They are also the readers least likely to ever ask you about it.

Most businesses answer the wrong reviews. The ones that get a reply are almost always the ones that felt unfair, rather than the ones a reader cannot decode without help, and that ordering is a preference for being right over being read.

How to tell which situation you are in

Two questions settle it, and neither of them is about how unfair the review felt when you read it.

The first is whether the review contains something a reader cannot check. If it does, you are the only source for the other half of the story, and withholding it leaves the reader with one version.

The second is whether you can describe what you did about it without contradicting what the reviewer says happened. Those two things coexist more often than people expect, because a customer can be right about the experience and wrong about the cause. If the honest answer is that nothing was done, a reply will communicate that whether you intend it to or not.

A late delivery makes the shape of it clear enough. You can say the order shipped two days after it should have and that the carrier is now different, which settles the open question. You cannot say the delivery was on time, because the person who waited for it is reading.

When both questions point the same way, write the reply. When the only thing you want to correct is the tone of the review, close the window. Fields that run on ratings feel this most sharply, and I have written separately about how rating profiles behave for physicians.

What happens when you pick wrong

Most of it is recoverable. A reply can be edited or taken down on every platform worth worrying about, and a weak one usually sinks along with the review beneath it as newer entries arrive.

What does not come back is the reader who already read it. That person made a decision and left no trace of having made it, which is the part of this work that never appears in any report.

The unrecoverable version is rarer and it looks different. It is the reply that attracts a second review from the same dispute, then a third, each one describing how the business responded rather than what it sold. At that point the page is no longer about a transaction, and nothing written in the reply field will pull it back.

One situation sits outside all of this. A review that states something false and damaging about a named person is a legal question before it is a reputation question, and it belongs with a lawyer rather than with a reply.

So treat the reply field the way you would treat a page you were publishing deliberately, because that is what it is. The person it is written for has not arrived yet, does not care who was right, and will read it once.

Dil Phagura is the Co-Founder and CEO of AI Reputation. Connect on LinkedIn at https://www.linkedin.com/in/dilphagura

Photos by Towfiqu barbhuiya and Álvaro Serrano on Unsplash

Sources

Why negative reviews could have more of an impact on some of the most important customers
Words Matter EKU Research Reveals the Best Way to Handle Negative Online Reviews

Building a Personal Brand That Outranks Bad Press

Dil Phagura

Across twelve years the same result has repeated often enough to stop being a coincidence. When a damaging article finally lost its position on a name query, the page that took the position was rarely the page built to take it.

It was the page that answered the question the article had put in a reader's head. Sometimes that page already existed and only needed finishing. Almost always it was the least interesting item on the list of things to build.

What did not take the position was the ambitious work. New personal sites, standing article programs and press pushes all ranked somewhere, settled two or three places below the article, and stayed there. The money was spent, the pages were good, and the first screen looked much as before.

What actually outranks bad press?

A damaging article is usually outranked by the single page that answers the same question a reader brought to the search. Which page that is depends on the reader rather than on the person being searched.

A name query is a question with a shape. Someone is deciding whether to refer you, hire you, sit on a board with you or sign something with you, and they want one or two reassurances before they do.

Those reassurances are narrow. A referrer wants to know the practice is steady and current. A nominating committee wants to know somebody independent will vouch for you.

A counterparty wants to know what you actually do and who is accountable for it. None of them arrived wanting your biography.

A damaging article answers one of those questions badly, and that is the slot it occupies. The asset that displaces it is the one that answers the same question properly, which is a much narrower job than building a presence.

Finding the question is less mysterious than it sounds. Read the article as a stranger would and ask what it leaves a reader believing about your work, then what a reader needed to believe instead.

The gap between those two sentences is the brief. Everything that does not serve it can be built later or not at all.

This is why pushing the article down is a separate move from answering the reader, and why the two get confused. I have written before about what suppression can and cannot do on its own, and the short version is that moving a page does nothing about the question underneath it.

Almost everybody starts from the other end. Survey work from Pew Research Center found that most adults would rather keep information about themselves out of search results than be able to find information about other people, so the instinct to begin with what you want gone is close to universal.

That instinct sets the order wrong. It starts from the thing you resent rather than the thing the reader wanted, and those are two different objects that happen to sit on the same screen.

The tell shows up in the first conversation

There is a question I ask early that predicts how the next year will go. What was the reader trying to find out when they landed on that article?

People who can answer it in a sentence tend to get their results back. They know a referrer was checking whether the practice is still active, or that a committee was looking for somebody who would speak for them.

People who answer only that the article is unfair have not started yet. The unfairness is often real and it is still not an answer, because nobody runs a search looking for a verdict on fairness.

There is a second tell, and it is faster. Ask somebody to name, without looking, the one page they would want a stranger to land on.

Anyone who can name it has an asset to strengthen. Anyone who cannot has only material to produce, and producing material without a destination is how a year gets spent without the first screen changing.

The order of the work is set by the reader, and almost nobody asks about the reader first.

Most personal brand work fails on its order rather than on its quality. It gets sequenced by what is satisfying to produce, and the satisfying thing is almost always a new property with your name on it.

The unsatisfying moves are the ones that pay. Finishing a page that already places for your name, getting one reference from an organization a reader will recognize, correcting a listing that still describes a role you left.

None of that photographs well in a proposal. All of it competes for the slot the article occupies, which is the only competition that changes what a reader sees.

Why do two people with the same coverage end up in different places?

Because one of them kept the assets current and the other treated them as finished. A position is held by upkeep, while the work that feels like progress is production.

Two situations, both described loosely enough to be several people. Two professionals, comparable coverage, comparable budgets, both building three assets across a year.

The first ordered the three by what a reader was asking and went back to each one every few months, correcting what had gone stale and adding whatever the quarter had produced. The second built three good pages, published them, and moved on to the next idea.

Inside the year the first had the article in fourth place and two pages of their own above it. The second had three pages that ranked respectably for phrases nobody was typing, and an article still answering the question at the top.

The variable was not talent and it was not spend. It was whether anybody went back.

Going back is duller than it sounds and it is most of the job. A paragraph that still names the wrong employer, a date that has passed, a list of what you handle that has not matched reality for two years.

Each correction takes twenty minutes and none feels like strategy. Together they are why one page keeps earning references while another stops being worth linking to.

Upkeep matters more than it sounds because the ground moves on its own. Pew Research Center measured this directly and found that a quarter of all webpages that existed at one point between 2013 and 2023 are no longer accessible.

Read that as it applies to your own name. Some share of what supports your best page, the references to it, the listings that point at it, the pages that quote it, will quietly stop existing.

So a page nobody revisits does not hold still. It decays in place, while the article, which belongs to an organization with staff and a publishing schedule, does not.

Where this pattern breaks

The pattern needs something a reader wants. When the article is the only substantive material about a person, there is no slot to build into and upkeep has nothing to keep, so the first job is having anything worth finding.

There are also positions that will not move. A long investigation by a serious publication, or a public record, can hold its place against everything you build, and anyone promising otherwise is selling the promise rather than the result.

The honest goal in that case is to be the second, third and fourth thing a reader finds, so the article becomes part of the picture instead of the whole of it. That is a smaller win than people want and it is still the difference between a conversation that continues and one that stops.

Where the material touches a live legal matter, that part is legal and belongs with a lawyer rather than with me.

The limitation that matters most is the one about dates. I cannot tell you which month the order changes, because what is being measured is other people deciding to reference you, and nobody controls that, including me.

This is the slowest route available, and the cost of it is real. A year of upkeep buys less visible progress than a single placement does, right up to the point where the placement stops helping.

It is also the only route that compounds, because every asset that stays current makes the next one easier to place. The advantage only arrives for people who are still there to collect it.

The work that holds a name is unglamorous and it never finishes. Most people decide they want a personal brand and start producing, which is the expensive way of arriving two years later at the same short list.

Find out what the reader wanted. Build in that order, keep the list short, and go back to it. Order first, then patience, and everything else in this work is decoration.

Dil Phagura is the Co-Founder and CEO of AI Reputation. Connect on LinkedIn at https://www.linkedin.com/in/dilphagura

Photos by AbsolutVision and Jonathan Gong on Unsplash

Sources

Most Americans support right to have some personal info removed from online searches
Link Rot and Digital Decay on Government, News and Other Webpages

How Long Online Reputation Repair Takes

By Dil Phagura

The question arrives in the same shape almost every time, usually inside the first ten minutes of a call. How long is this going to take. Often it is not really a question, because there is a board vote or a closing or a hiring decision already on the calendar and what the caller wants to know is whether the results will look different by then.

I give a range. Six to twelve months is the band I quote most often for a first page that a reader would describe as materially different, and I quote it knowing that some situations settle sooner and some are still moving past a year.

A range is an unsatisfying thing to hand a competent adult who is quoted firm dates for everything else they buy. It is also the only answer that survives contact with the way search actually behaves.

How long does online reputation repair take?

Most of this work lands between six and twelve months before the first screen of a name query looks materially different to somebody reading it fresh. A crowded name or a matter still producing coverage can push that past a year, and a thin result set can settle faster.

The band is wide because repair is not a single task with a duration. Several separate processes run at different speeds, and the finish is whichever one of them turns out to be slowest for this particular name.

Those processes have very different speeds. Writing and publishing runs on a schedule anybody can hold, usually a matter of weeks, and getting the material in front of a search engine is quick after that. The slow stretch is the one where material has to earn a place, and nothing about that stretch runs to a timetable.

Part of why a range lands badly is that the systems sitting next to this one do carry deadlines, and those deadlines are written down. Dispute an item on a credit report and the company generally has 30 days to investigate it. Nothing in search carries a clock like that, and no regulator has written one.

So people arrive expecting the shape of a records process and meet something closer to weather. That comparison is not meant to be flippant. The two really do behave differently, and the difference is most of the answer.

The clock belongs to other people

The pages that displace a damaging result almost never win on the strength of what is written on them. They win on what gathers around them over time: other people citing them, referencing them, pointing at them and treating them as the place to go when the name comes up.

None of that gathering happens on a schedule I set. It happens when a conference programme goes up, when a trade publication runs a piece, when a professional body refreshes a directory, when somebody decides on their own that a page is worth a mention.

None of it can be bought forward either. Attention from other people arrives when they have a reason of their own, and the only lever available is making sure something worth pointing at is easy to find on the day somebody goes looking. That is a weaker lever than people want it to be, and it is the one that works.

That is the real reason the answer is a range. The material can be produced in a known number of weeks, and then the part that decides whether any of it ranks is handed to people who have never heard of the engagement and are working to calendars of their own.

Every honest number in this work is a range, because the thing being measured is other people deciding to reference you.

The inventory underneath all of this moves on its own as well. A quarter of the pages one large study collected between 2013 and 2023 were no longer reachable by the end of that period. Some of what sits above a name today will leave without anybody filing anything, and some of what replaces it will be worse.

What actually sets the timeline?

Four things set it: how old and how heavily referenced the damaging page is, how many other people share the name, whether the subject already owns anything that ranks, and whether the underlying matter is still generating new material. The first and the third do most of the work.

Age is the one people underestimate. A page that has sat in the index for years has collected references from places nobody remembers, and every one of those is a reason for a search engine to go on choosing it. I have written separately about what search suppression does to a result like that.

A crowded name cuts in both directions. Sharing a name with several other people makes any single page harder to place, and it also means the damaging result has to keep beating that same crowd to stay where it is.

The four rarely line up neatly. A thin result set holding one old and heavily referenced page behaves nothing like a crowded one where the damaging page is recent and barely referenced, and the second of those usually moves faster despite looking worse on first inspection.

Owning something already is the largest shortcut available. A professional with a real site, a maintained profile and a few pages that already rank for the name begins from a position that would otherwise take the better part of a year to build.

That is also why the same engagement, run the same way, can take half the time for one person and twice as long for another. The difference is rarely in the work itself, and almost always in what each of them already had.

A matter that is still producing coverage extends all of it, and how to work during a period like that is a separate subject worth its own piece.

Can anyone promise a date?

No, and a firm date offered at the proposal stage is one of the more reliable warning signs in this field. Anyone quoting a fixed number of weeks is describing either something that will not hold or something that will make the situation worse.

What can be told to you honestly is which of those four factors is the binding one for a particular name, and even that comes after the review rather than before it. Before the review, a date is a number produced to sound confident.

There is a narrower promise that is honest. A date for the material, for the pages written and published and maintained, is a real commitment and I will give one. A date for where those pages sit in a result set is not mine to give.

Dates get quoted because they are what people ask for. A proposal carrying a month on it reads as confidence beside one offering a range, and the person asking has a board paper to write and needs a number for it. The number is produced because it is wanted, not because anybody can see that far ahead.

The hardest thing to say to somebody in a hurry is that most of the variance was settled long before they called. A name with a maintained site and a decade of genuine professional material behind it moves in months. A name with nothing of its own starts the clock at zero, whatever is spent on it.

The honest unit for this work is the month, and the honest answer to how many is a range you can plan around rather than a date you can hold somebody to.

Dil Phagura is the Co-Founder and CEO of AI Reputation. Connect on LinkedIn at https://www.linkedin.com/in/dilphagura

Photos by Paico Oficial and Swello on Unsplash

Sources

How long does it take to repair an error on a credit report
Link Rot and Digital Decay on Government, News and Other Webpages

What Is Search Suppression and When Does It Work

Photo by Nicolas Gonzalez on Unsplash

By Dil Phagura

Search suppression is the one piece of this work most people have heard of before they call anyone. The belief arrives fully formed. The article cannot be removed, so you pay somebody to push it onto page two, and once it sits there it has stopped costing you anything.

Stated at full strength that is a reasonable position, and for a narrow set of results it is exactly right. It is also the belief behind a large share of the money wasted in this field, because what gets bought is rarely what the word describes.

What search suppression actually does to a result

Suppression is the name of an outcome rather than the name of an action. Nothing in it reaches the page causing the problem, which stays published, at the same address, with the same words on it.

What moves is the order a search engine returns when somebody types the name. That order is a competition for a small number of positions, and it is settled again every time the query is run, which is how Google describes the way it decides which pages to return.

A damaging result does not sink. Other pages climb past it.

The distinction sounds academic until somebody is paying for it. One of those descriptions is a job with an end date. The other is a standing competition that carries on after the invoice is settled.

Ask what is actually being delivered and the answer is usually a list of pages. Some will be profiles on platforms that already rank, some will be writing placed on sites willing to take it, and some will be material on a site the subject owns.

That is a reasonable list to be given. It is a list of candidates for positions, and handing it over is not the same thing as holding them.

Why a suppressed result comes back

The case that breaks the belief is always the one that looked finished. A professional came to me with a result that an earlier engagement had pushed onto the second page nearly a year before, and that was sitting at position four again by the time we spoke. Nothing about that result had changed.

What had changed sat above it. Two of the pages holding those positions were on a platform that had since reorganised itself and dropped them. The others were thin pieces published once and never touched again.

A position above a damaging result is rented, and the rent is paid in upkeep.

Ranking rewards pages that go on earning their place, which is the whole burden of what Google publishes about the material it expects to reward. Stop writing, stop updating, stop being cited anywhere, and the order drifts back toward whatever has the most durable signals behind it.

Very often that is the damaging page. A news report or a public record accumulates links and citations for years with nobody tending it.

There is a way to know early which of these two outcomes you are heading for, and it has nothing to do with where the result sits this month. Look at whether anything in the plan recurs.

A set of pages published in one burst and then declared finished is a result that will come back. The recurring commitment nobody wants to pay for is the part that holds the position.

Being on page one and being eligible are different things

People new to this assume a page that exists and is indexed is therefore in the running. Google separates those stages carefully in the account of indexing and serving it publishes for developers, and the gap between them is where most suppression money disappears.

A page built to fill a position has to be the kind of page a search engine would have chosen anyway. Five of those will hold a result down for years. Fifteen pages nobody has a reason to read will hold it down until the next time the index is reshuffled.

What to ask before you pay for search suppression

The question worth asking is what will still be holding the positions above a damaging result in two years. Whether the result can be moved this quarter is an easier question and a far less useful one.

Most engagements of this kind that end well would have ended well without the suppression work, because the thing that actually moved the result was the material built beside it. That is an uncomfortable claim to make in this industry and I will stand on it.

It also changes what you are shopping for. A plan that produces durable material under your own name suppresses the result as a side effect and goes on doing it. A plan built to move one result and nothing else has to be bought again, by you, every time the order resets.

The practical version of that question is short enough to ask on a call. Who maintains these pages next year, what gets published in month nine, and which of these positions depends on a platform that could reorganise itself.

The pattern shows most clearly where the page above you is a third party profile that genuinely earns its position, which is what a physician usually finds sitting above a practice page. A profile like that is not waiting to be nudged. Something has to be worth more than it is.

When search suppression on its own is enough

There are results where the original belief holds precisely as stated, and pretending otherwise would be its own kind of dishonesty. A single page on a host with no authority, no inbound links and no reason for anybody to cite it will give up a position to almost anything credible, and it will not win it back.

Two further conditions matter as much as the weakness of the page. The name needs enough real material around it that credible candidates for those positions exist at all. The subject needs to own something already worth ranking, because a position has to be filled by something.

Where all three hold, this is cheap, fast by the standards of this work, and stable. Where any one of them fails, a quarter spent moving the result buys a few months of a cleaner screen and very little else.

One more situation is worth naming, because it comes up more than anybody admits. A page sitting sixth for a name almost nobody searches is doing less damage than its position suggests, and the honest advice is often to leave it alone and spend the money on something that compounds.

Photo by Ruijia Wang on Unsplash

One thing to be clear about before anybody spends money on this. None of it touches the page itself, which stays where it was published, with the same words on it, and a reader who goes looking will find it.

What the work moves is the order of attention, and no order is permanent. I can change what a search engine puts first and I cannot make a position stay there on its own.

Suppression is what a working plan looks like from the outside, and it is never the plan itself.

Dil Phagura is the Co-Founder and CEO of AI Reputation. Connect on LinkedIn at https://www.linkedin.com/in/dilphagura

How to Remove Personal Information From Data Broker Sites

Dil Phagura

Photo by Maksym Kaharlytskyi on Unsplash

An executive I work with found out the hard way, when a courier left a signed delivery at a house she had not owned for six years. The package was legitimate. The address had come from a people search page sitting fourth in the results for her own name, one position above the firm bio her marketing team had spent a year building.

She had seen that page before and decided it was harmless, because every field on it was technically public. What changed her mind was the birth year printed beside her name and the two first names listed under relatives, which belonged to her children.

What a data broker page actually holds about you

When I open one of these records properly, the fields look dull one at a time. A name, a birth year, a current address and two or three former ones, a mobile number that still connects, an employer, and a list of relatives and people the site calls known associates.

Some of them add a property value, a vehicle, a neighbor, and a line that hints at a court record without ever saying what it was. No single field there is interesting on its own. The assembly is what does the damage, because a stranger reading from the top down ends up with a dossier that nobody authorized and nobody checked.

That is also why this category behaves differently from most of the work I do. A critical article has an author and an argument, and both of those can be answered. A broker page has neither, so there is nothing to respond to and the only move with any value in it is getting the record taken down.

Why opting out of one data broker does almost nothing

California keeps a public register of companies that have told a state regulator they trade in personal information, and the data broker registry runs to hundreds of entries. That register covers only the firms that filed. The ones that never registered are buying and selling all the same.

Underneath the retail sites the market is far smaller than it appears. A handful of large aggregators do the collecting, then license the same file out to dozens of consumer facing brands, several of which share a parent company and differ mainly in their logo. The Privacy Rights Clearinghouse has been documenting that structure for years, and reading it is the fastest cure for the idea that there are five of these sites.

So a successful opt out on one site removes a copy. The source that supplied the copy is untouched, and the next time it licenses the file your record turns up somewhere new, with your consent nowhere in the transaction.

I would rather a client work six brokers every quarter for two years than work sixty once and call the problem closed.

What order to work a data broker removal in

The first pass is an inventory, and it is slower than anyone expects. Most opt out forms want the exact address of the record rather than a general request, so somebody has to run the name queries, open every result and write each page address down. I have written before about how to check what is being said about you, and the same query discipline carries over here.

Second come the upstream aggregators, ahead of any retail site. Working the small sites first feels productive and then undoes itself, because the aggregator that fed them is still licensing and the pages come back within a cycle or two.

Third is the search side. Google will consider removing results that expose personally identifiable information, and its process for finding and removing personal info in Search is the only step in this sequence that changes what a reader sees inside a week. The page can go on existing while the result stops appearing, which buys time for the slower work underneath.

Fourth is the long tail, worked in batches over weeks instead of in one sitting. Fifth, and this is the step almost everybody drops, is a repeating calendar entry. Without it the first four steps are a single clean afternoon that quietly expires.

What the opt out forms ask you to hand over

There is an uncomfortable trade sitting in the middle of this. A fair number of these sites will not process a removal without an email address, a confirmation click, and in some cases a photograph of a government identity document.

Send only what the form requires, use an email address kept for this purpose and nothing else, and stop at any site asking for materially more than its neighbors do. Where the record underneath is a court filing or something drawn from a credit header, that is a different process under different rules, and the person to ask is a lawyer who has read the actual document.

Photo by Jan Antonin Kolar on Unsplash

How long data broker removal takes and how often it repeats

On the upstream sources, results usually start dropping out between two and six weeks after a request lands. The long tail runs three to four months, partly because a share of the smaller sites ignore a first request entirely and have to be asked a second time, in writing, with the page address attached again.

Then the cycle starts over. Public records refresh, lists change hands again, and records that were cleanly removed tend to reappear somewhere inside six to twelve months. This is a standing chore with no finish date, and anybody describing it as a one time cleanup is selling the version of the service that is easy to sell.

Two things did not move for that executive, and both were predictable before we started. Her property record stayed exactly where it was, because a county office holding a public document is not a broker and has no opt out form to send anything to. The same is true of a licence register, a corporate filing and a court docket.

Her firm bio also did not climb simply because the broker pages fell away. Removing a page frees a position. Filling that position is separate work on a separate schedule, and it runs longer than the removals do.

Her first screen was clear of broker pages about four months in, and it has taken a morning of attention every quarter since to stay that way. I cannot make a public document private, and I cannot stop a company from licensing the same file again next year. What this work buys is the gap between the two, and the gap is worth having.

The durable version of this is a recurring reminder that outlives the alarm that created it.

Dil Phagura is the Co-Founder and CEO of AI Reputation. Connect on LinkedIn at https://www.linkedin.com/in/dilphagura

Online Reputation Management for Physicians

Dil Phagura

Photo by CDC on Unsplash

The same call has come in enough times that I can usually guess the first sentence. A physician searches their own name, often after a patient says something in an exam room, and finds a rating profile they never built sitting above the website their practice paid for. By the second or third sentence they have asked whether the profile can be taken down.

I have been doing this work for twelve years and that conversation has not changed much. What has changed is how reliably it ends the same way. The removal question takes about a minute to answer honestly, and the question underneath it takes the rest of the year.

What physicians find when they search their own name

The result set is more predictable for physicians than for almost any other profession I work with. Two or three rating profiles, a group bio written by someone in marketing, an insurance network listing, and then the physician website below all of it.

What surprises people is how little of that they have ever touched. The rating profiles were built from public licensing and claims data without anyone being asked. The group bio went through a template and a committee, and the insurance listing still carries an address the physician left in 2019.

There is a structural reason those pages sit where they do, and it is worth knowing once and then setting aside. Profile pages on large sites inherit authority from tens of thousands of sibling pages, they carry structured data a search engine can read without guessing, and many have existed longer than the physician website.

That is background. The question worth spending a year on is what a physician can put into the result set that will eventually hold a position of its own.

The tell that predicts which case a physician is in

I ask two questions in the first ten minutes, and the answers predict the next two years better than anything on the first screen.

The first is what the physician has published under their own name in the last two years. Not what the practice published, and not what a marketing agency wrote in the third person. Anything the physician actually produced, including a patient education page they wrote themselves, a talk they gave that was recorded, a comment in a trade publication, a chapter nobody read.

The second is how reviews arrive. Weekly in ones and twos, or in bursts after someone decides it is time to do something about the ratings. The difference is visible in the dates and it tells me whether review collection is part of how the practice runs or a project that gets started and abandoned.

A physician who answers the first question with nothing and the second with bursts is in a worse position than their first screen suggests. A physician with two years of small, unglamorous output and a steady trickle of reviews is in a better position than theirs, even if the page one screenshot looks identical. Advisors working inside a firm review queue hit the same shortage of their own material from the other direction, which is some comfort and no help at all.

Two physicians with the same first screen and different second years

Two situations from the same period, specialties close enough that the comparison holds. Both arrived with a rating profile above their own site and an average score they were unhappy about. Both signed up for what was effectively the same plan.

The first physician treated it as a project with an end date. The website was rebuilt properly, which took three months and looked far better afterward. Review requests went out to a list of past patients, about two hundred of them, and roughly thirty people responded over five weeks.

Eighteen months later the rating profile still held the top position and the rebuilt site had moved up one place. The score had improved and then drifted back toward where it started, because the thirty reviews aged and nothing replaced them.

A physician with a plain website and a steady review flow is in better shape than a physician with a beautiful website and none.

The second physician did something less impressive and more useful. The website stayed ugly for another year. What changed was that the front desk began asking every patient at checkout, which produced four or five reviews a week rather than thirty in a month, and the physician wrote one short page a month about a question patients actually asked in clinic.

Twelve of those pages existed by the end of the first year. None of them individually did anything. Together they gave the search engine a reason to treat the physician website as a thing about that physician rather than a brochure, and by month twenty the site was above two of the three rating profiles.

What actually separated them

Budget did not separate them, and neither did the quality of the rebuild. The variable was whether the work was a build or a supply.

A build finishes and starts aging on the day it ships. A supply keeps arriving, which is the only thing either search engines or answer engines have ever rewarded over a period measured in years.

That distinction decides almost every physician case I see, and I would rather start with a plain site and a working checkout habit than with a rebuild and a quiet front desk.

Supply is also less work than it sounds, which is the part physicians tend not to believe. One page a month is an hour of dictation and an hour of cleanup, and the questions already arrive in clinic every week. The practices that fail at this rarely fail on capacity, they fail because nobody owns the task once the first enthusiasm is spent.

Photo by Brandon Holmes on Unsplash

How much review volume actually moves

Review volume is the one lever in this field that works for physicians at scale, and it is slow enough that most practices quit before it pays. A profile carrying forty reviews from four years ago behaves very differently from one carrying forty reviews from the last eight months, and closing that gap is a matter of quarters rather than weeks.

It also has to be done inside the rules, which are narrower than most practices assume. Platforms set out what is permitted when you ask patients for reviews, and selective asking, incentives and bulk collection at a kiosk all fall outside it. Getting this wrong is worse than doing nothing, because the penalty arrives later and takes the legitimate reviews with it.

Responding is where physicians carry a risk nobody else on my client list does. A reply that confirms someone was a patient has disclosed information that was not yours to disclose, and the professional standards for a physician speaking publicly about clinical matters apply to a review box exactly as they apply anywhere else. The safe reply is short, says nothing about any individual, and invites the conversation offline.

The other thing worth knowing is that the score matters less than practices expect and the recency matters more. A profile at 4.3 with reviews from this quarter reads as a working practice, and a profile at 4.8 whose newest review is from 2023 reads as a practice that stopped paying attention.

Here is what I cannot do, and it comes up on every one of these calls. I cannot make a physician page worth ranking if there is nothing behind it. If the physician has nothing to say about their own field, the site will stay a brochure however it is built, and a brochure loses to a profile page with ten thousand siblings every time.

When this pattern does not hold

The pattern breaks for employed physicians whose group owns the web property. If the hospital owns the domain, controls the bio, and feeds the directory listings, then the physician cannot supply anything to the asset that would rank, because the asset is not theirs. I have watched people spend a year trying to improve a page they have no ability to edit.

In that case the work moves off the group domain entirely, to material the physician owns personally and keeps after they leave. It is slower, it looks less relevant for the first six months, and it is the only version that survives a change of employer. Whether a particular contract permits any of it is a question for a lawyer who has read the contract, not for me.

The physicians who come out of this well are rarely the ones who were most upset at the start. They are the ones who found something small they could keep doing after the urgency wore off.

Dil Phagura is the Co-Founder and CEO of AI Reputation. Connect on LinkedIn at https://www.linkedin.com/in/dilphagura

Financial Advisor Reputation Inside Compliance Review

Dil Phagura, Co-Founder and CEO of AI Reputation

The question arrives in nearly the same words every time. What am I actually allowed to publish, when everything I write has to clear review first. It gets asked with the expectation that the answer is almost nothing.

One advisor sent me the biography his firm had approved for him. Then he sent four more from colleagues down the hall, because he had already half noticed the thing I was about to say out loud. Change the name and the years and the five paragraphs were one paragraph.

He wanted to know whether his firm was being unreasonable. It was not. What had happened to his page was never handed down by a regulator, and the difference between those two things decides what there is to do about it.

What financial advisors are allowed to publish under compliance review

The limit is narrower than most advisors hope and considerably wider than most advisors assume. What a firm permits depends on which regulator it answers to and on policy the firm wrote for itself, so there is no single industry wide boundary to quote back to anyone.

On the broker dealer side the governing text is FINRA Rule 2210 on communications with the public, which sorts communications by who receives them, requires principal approval before a retail communication goes out, and layers filing and retention obligations on top of that. Advisers registered with the Commission work from a different rule set carrying its own advertising standards. Either way the review is procedural rather than editorial, and what it tests is whether a claim is fair, balanced, substantiated and retained.

Nothing anywhere in it says the writing has to be generic. That part is a firm level decision, and it is usually made to keep the review queue moving.

Two advisors at one firm can publish genuinely different pages and both clear review, because what review examines is the claim and the substantiation sitting behind it rather than whether a sentence is familiar. A specific statement about who an advisor serves and how is often easier to substantiate than a vague one, since the reviewer can see exactly what is being asserted. Vagueness reads as safe and is frequently the harder thing to approve on its own terms.

Why approved advisor bios all read the same way

Review is a queue with a cost attached to every item in it. New language takes longer to clear than language that cleared last month, because a reviewer who has read a sentence before can approve it in seconds. So firms build a library of approved components, and the library quietly becomes the default.

The components are recognizable once you have read a few hundred of them. Years of experience, a list of client types, a sentence about a planning process, a closing line about family and a hobby. None of it is false and none of it gives a reader anything to choose one advisor over another with.

From the firm side that is rational, and from the advisor side it produces one predictable outcome. Every advisor publishes a page assembled out of the same components, in the same order, carrying the same three verbs.

The template was never meant to be anybody's public identity, and for several hundred advisors at once that is precisely what it becomes.

Most of the advisors I talk to treat the template as a ceiling, and their marketing teams encourage the reading. I think it is the wrong reading, and I think what it costs is larger than the compliance exposure anyone believes they are avoiding.

How search engines and answer engines read identical advisor pages

Ranking anything for a person name query comes down to what makes one page the better answer about that person. When a search engine finds a dozen near identical pages describing a dozen different people, there is nothing inside any of them to prefer, so the decision gets made on the signals sitting outside the text: the domain, the age of the page, what links to it.

On a firm domain those outside signals belong to the firm. The firm page wins, the advisor page settles underneath it, and third party directories carrying structured detail about the individual slot in between. Google's documentation on how it consolidates duplicate pages is written about one piece of substance living at several addresses rather than about this situation, but the behavior underneath is the one that matters here, which is that text carrying nothing distinguishing hands the ranking decision somewhere else.

None of this is a penalty. Nobody is demoted for using approved language and there is no filter hunting for templates. The page just holds nothing a ranking system or a retrieval system can act on, so both of them act on something else.

Answer engines work differently and land somewhere similar. Retrieval reaches for passages that answer the question being asked, and a passage has to say something particular to be worth reaching for. Four hundred words of approved component text contain no sentence that answers what this person is actually good at, so nothing from the page gets quoted and the summary gets built out of whatever else exists.

What gets quoted instead

Usually a directory entry, a conference program, a licensing record, or one old article. Those sources are thin and they are specific, and specific is the quality that earns a citation. An approved biography loses to a two line listing nobody at the firm would have considered competition.

That is the part advisors find hardest to accept. The thin source wins not because it is better written but because it commits to a claim, and commitment is the thing both systems are reaching for.

Before changing any of it, it is worth knowing what the systems say right now, and I have written out the order I run those checks in on this blog already.

What compliance review will not bend on

The rules governing advisor communications belong to the firm to interpret and to its compliance team to enforce, and nothing about how I work shortens either. Whether a particular sentence may publish, what substantiation it needs, how many weeks the queue runs: those answers come from inside the firm. Where the underlying question is a legal one it belongs with the firm's counsel and not with me.

FINRA's guidance on social networking websites and business communications gives a fair picture of how closely this gets watched, down to who owns a share, a comment or a correction to a listing somebody else published. Read it and the scale of the supervision is obvious. What is also obvious is that none of it prescribes the words.

So the binding constraint is time rather than permission. Getting distinct material through review means writing it, substantiating it, queueing it and waiting, and at the firms I have worked with that runs in months. An advisor who wants a different page one by the end of the quarter is going to be disappointed, and I would rather say so before anyone pays for the attempt.

The second limit is quieter and it is harder to fix. Material distinct enough to rank is material the advisor has to actually possess: a defensible opinion, a real specialty, a client population they understand better than the colleagues on the same template do. Review cannot supply that and neither can I.

There is a narrow case where the template is the right answer. An advisor two years from retirement with a closed book and no intention of taking anyone new on has no reason to spend six months pushing distinctive material through a queue. This work pays for the people whose name is about to be typed by somebody deciding something.

Which is why I never open this work on the biography. I start by asking what an advisor knows that the four hundred people sharing their template do not, and whether they can defend it in writing for a reviewer who will ask.

Everything after that is paperwork and patience.

Dil Phagura is the Co-Founder and CEO of AI Reputation. Connect on LinkedIn at https://www.linkedin.com/in/dilphagura