The PR Agency's Institutional Memory Problem: What Happens When Your Best Publicist Takes 11 Years of Media Relationships With Her

Every public relations agency runs on an asset it cannot put on a balance sheet, cannot insure, and — this is the uncomfortable part — does not actually own.

That asset is relationships. Not the client relationships, which are contractual and reasonably well documented. The other ones. The features editor at the trade title who takes your calls. The producer who owes you a favour. The freelancer who has moved outlets three times in six years and whose personal address you have because your senior publicist got it at a conference in 2019.

Those relationships are the reason a client pays you a retainer instead of hiring a junior in-house. They are your entire competitive moat. And in the overwhelming majority of PR agencies, they exist in exactly one place: inside an individual employee's email account.

This post is about what happens when that employee resigns, and about the second problem that always travels with the first — the inability to prove that any of this earned media produced commercial results. The two are more connected than they look, and they have a common fix.

Why is a PR agency's most valuable asset the one it doesn't own?

Because relationships are held by people, and the systems most agencies use to manage those relationships are personal rather than institutional. A publicist's media relationships live in their sent folder, their phone contacts, their memory, and a spreadsheet they maintain privately. None of that is agency property in any practical sense, even if it is in the employment contract.

Consider what a mature media relationship actually contains as information. It contains the journalist's name and current outlet, which is the easy part and available in any commercial database. It also contains: their beat and how it has shifted; whether they prefer email, phone or a direct message; that they hate embargoes; that they responded well to a data-led angle in March but ignored two product-launch pitches; that they moved from one title to another and took their old address with them; that they once killed a story because a client's spokesperson was unavailable for eleven days.

Almost none of that second category is written down anywhere.

It is, however, the entire difference between a pitch that lands and a pitch that gets ignored. A junior handed the same contact list will get a fraction of the hit rate, not because they're worse at writing, but because they're operating without context that was never recorded.

The rented-moat problem

This creates a structural fragility that most agency principals underestimate until it bites. You are selling clients access to relationships. Your ability to deliver on that promise is contingent on specific individuals continuing to work for you. Your pricing assumes an asset your business does not control.

Agencies notice this in three situations. A senior person resigns. A senior person goes on extended leave. A senior person's accounts get reassigned and the handover produces a document listing names and outlets but nothing about how any of those people actually work.

The first is the catastrophic version. The other two are the everyday version, and they cost more in aggregate.

Why it persists

Nobody in PR set out to build a business this way. It persists for reasons that are individually reasonable.

Relationship work resists documentation because it feels reductive — writing down "responds to exclusives, dislikes phone calls" seems to flatten something more nuanced. Publicists are protective of contacts because in an industry with high mobility, your contact list is your personal market value. And logging feels like admin in a business where every hour is either billable to a client or a cost, and admin is neither.

So the knowledge stays in inboxes. Until it leaves.

What actually happens when a senior publicist resigns?

The loss is immediate and larger than the handover document suggests, because what departs is not the contact list — that can be reconstructed — but the accumulated context about how to work each contact, plus the live state of every open pitch.

Here is the realistic sequence, drawn from what agencies actually report.

Week one. The resignation lands. The publicist is professional, offers a full handover, and means it. Accounts are reassigned to whoever has capacity.

Weeks two to four. The handover happens. It produces a contact spreadsheet, a set of account notes, and several long conversations. This feels adequate at the time. It is not, and won't be visible as inadequate for another two months.

Week five. The publicist's email access is revoked, per policy. At this moment several years of pitch correspondence becomes inaccessible to the agency, including every thread with every journalist, every response, and every piece of context embedded in those exchanges.

Weeks six to twelve. The failure modes surface. A colleague pitches a journalist who was already pitched six weeks ago on the same client — the journalist notices, and says so. A story that was in play quietly dies because nobody knew it was in play. A client asks why response rates have dropped on an account. Two journalists who had a genuine relationship with the departed publicist stop responding entirely, because their relationship was never with the agency.

Month four onwards. Steady-state degradation. The account still runs. Hit rates are lower. Nobody can quantify by how much, because nobody was measuring before.

The compounding version

Now run that sequence twice in one quarter, which is what happened at Meridian Communications, a 15-person PR agency working mainly across B2B technology and professional services.

Two senior publicists resigned within seven weeks of each other — one to an in-house role, one to a competitor. Between them they had been with the agency for six and five years respectively, and had worked in the sector for considerably longer. The relationships they had built on the agency's behalf spanned roughly eleven years of accumulated work.

Meridian had no shared media CRM. Contacts lived in two Outlook accounts, one personal spreadsheet, and a Cision seat that told them who journalists were but nothing about what the agency's history with them had been.

What Meridian lost was not eleven years of names. It was eleven years of knowing which of those names mattered.

Seven of nine categories were substantially unrecoverable. The two that were recoverable were the two available commercially from any media database — which is to say, the two with no competitive value.

Meridian's managing director described the position afterwards in a way I'd encourage any agency principal to sit with: the agency had been charging clients for a capability that it turned out to be renting from two individuals, month to month, with no notice period on the underlying asset.

How do PR agencies prove commercial impact to a CFO?

Most can't, and that is the second structural problem. The standard PR reporting metrics — coverage volume, circulation, impressions, advertising value equivalency — describe media activity rather than business outcomes, and a finance director evaluating spend does not recognise any of them as results.

Let me define terms, because they get used loosely.

Earned media is coverage obtained through editorial merit rather than paid placement — a feature, a quote in a news piece, a broadcast segment. Attribution is the practice of connecting a specific marketing activity to a specific commercial outcome, such as an enquiry, a qualified opportunity or closed revenue. Impressions is an estimate of how many people had the opportunity to see a piece of coverage, usually derived from outlet circulation or traffic figures.

The problem with the standard reporting set is straightforward. Impressions estimate exposure, not attention, and certainly not action. Circulation is an outlet-level number that has nothing to do with your client's placement within it. Advertising value equivalency — pricing coverage as if you'd bought the equivalent ad space — has been formally rejected by the industry's own measurement bodies for years, and any CFO who has bought advertising knows the comparison is not sound.

The renewal conversation this produces

Here is the conversation that follows, and every agency principal reading this has had some version of it.

The client's marketing lead likes you. The work has been good. But the CFO has asked what the $6,500 a month is producing, and the marketing lead has a deck of clippings and a number reading 4.2 million impressions. The CFO asks how many customers that produced. Nobody knows. The retainer gets renegotiated to $5,000, or moved to a project basis, or held flat for a third year while costs rise.

The agency did nothing wrong. It simply had no answer to a reasonable question.

What the alternative sounds like

Now the same meeting with attribution in place. The feature in the trade title ran on the 14th. It drove 340 clicks to a tracked landing page. Of those, 47 submitted an enquiry, which the client's own CRM can confirm. Twelve were qualified into the sales pipeline. Three closed, at an average contract value the client already knows.

Nobody is renegotiating that retainer downward. The conversation moves from cost justification to how to get more of it.

The gap between those two meetings is not the quality of the PR work. In both cases the coverage was the same. The gap is entirely infrastructure — whether anybody built the plumbing to count what happened next.

What does a shared media-contact CRM actually look like?

It is a structured database where every journalist, outlet and pitch is a record owned by the agency rather than a thread in someone's inbox, with enough fields to capture the working context that currently lives only in people's heads.

The design matters enormously. Too few fields and it holds nothing worth having. Too many and nobody fills it in, which is how most PR CRM projects fail. The version below is what I build on GoHighLevel, and it is deliberately lean.

The media contact record

Each journalist is a contact record with custom fields grouped into four layers.

Identity and reach. Name, current outlet, job title, beat, secondary beats, email, phone, social handles, and — this one matters — previous outlets with dates. Journalists move constantly, and an agency that tracks the move keeps the relationship. One that doesn't loses the contact when the address bounces.

Working preferences. Preferred contact method. Preferred pitch format, such as short-and-data-led versus full-narrative. Embargo tolerance. Typical lead time. Deadline days. Whether they accept exclusives. Whether they've asked not to be contacted about certain categories.

Relationship state. A relationship-strength rating from cold through to genuinely warm. The internal relationship owner — which of your people actually knows them. Date of last contact. Date of last response. Total pitches sent, total responses, total placements. Those last three generate a response rate automatically, which turns out to be the single most useful number in the whole system.

Commercial history. Which of your clients they have covered, how many times, and what those placements produced. This is the layer almost nobody builds and the one that makes the database an asset rather than an address book.

The outlet layer

Outlets get their own records, separate from the individuals, because outlet relationships outlive individual journalists.

An outlet record holds the publication or programme, its tier, audience profile, editorial calendar and deadlines, submission process, the journalists you know there, total placements the agency has secured historically, and any standing relationship such as a contributed-column slot.

The practical value is continuity. When your contact at a title leaves for a competitor, you have both a warm contact at the new title and a documented relationship with the old one to rebuild against.

The pitch record

This is the object that most directly replaces the missing sent folder, and it's the one that has to be fast to create or it won't get created.

Each pitch record captures the journalist, the outlet, the client, the angle in one line, the date sent, the format, whether it was exclusive or embargoed, and the assets attached. Then a status that moves through a defined set of stages, a response field with date and substance, an outcome, and — critically — a linked coverage record where one results.

Thirty seconds to log, if the fields are dropdowns and the record is created from a template. That thirty-second discipline is the difference between an agency that owns its history and one that rents it.

How do you log pitch history so it survives staff turnover?

By making pitch logging a step in the pitching workflow rather than an administrative task performed afterwards, and by structuring the record so the logging takes seconds rather than minutes.

The failure pattern is predictable. An agency buys a CRM, mandates that everything be logged, discovers three months later that compliance is around 20%, and quietly abandons the system. Every agency that has tried this recognises the arc.

Three design decisions prevent it.

Log at the point of sending, not after. The pitch record is created as part of sending the pitch, from a template pre-populated with client, campaign and angle. The publicist adds the journalist and hits save. If logging is a separate end-of-day task, it competes with client work and loses every time.

Give something back immediately. The moment a publicist opens a journalist's record and sees that a colleague pitched the same person eleven days ago on an adjacent story, the system has paid for itself in that publicist's eyes. Duplicate pitching is the single most relationship-damaging error in PR, and it is entirely a coordination failure. Once people have been saved from it once, resistance to logging drops sharply.

Automate the parts that can be automated. Response tracking updates when a reply is logged. Response rates calculate themselves. Follow-up tasks generate on schedule. Coverage links back to the originating pitch. The human is only ever asked to supply information a machine cannot infer.

The structured pitch pipeline

Pitches move through a visible pipeline, which is where GoHighLevel's opportunity structure does the work. A typical PR pitch pipeline runs through stages such as angle drafted, pitch sent, follow-up one sent, follow-up two sent, journalist engaged, information requested, interview scheduled, confirmed for publication, published, and declined or lapsed.

The stage names matter less than the fact that every open pitch across the agency is visible on one board. When someone resigns, their pitches don't vanish — they're sitting in a pipeline stage with a date, an angle and a response history, ready to be picked up.

How Meridian rebuilt

Meridian's reconstruction took about eleven weeks and was mostly manual, because there was no automated way to recover what had been lost.

They started with the departed publicists' client accounts and worked backwards from every placement in the previous three years, building outlet and contact records from the coverage itself. Where a placement existed, a relationship had existed, and the record could be reconstructed at least partially.

They interviewed the remaining team exhaustively — not "who do you know" but "walk me through the last five pitches you sent and what you knew about each recipient before you sent them." That produced far more usable intelligence than a contact-list request would have.

They exported what they could from the media database to fill identity fields, and treated everything else as unknown rather than guessing.

The reconstructed database held 412 media contacts across 180 outlets. Meridian's estimate was that this represented perhaps 60% of what the two departed publicists had actually known, and essentially none of the preference and relationship intelligence.

That 40% gap is the cost of not having built this earlier. It is not recoverable. It is, however, the last time they will pay it.

How do you track coverage back to actual enquiries?

Through layered attribution — tracked links where the outlet permits them, dedicated landing pages and campaign phone numbers where it doesn't, and a source-capture question on the client's enquiry form as the catch-all. Each method covers the others' blind spots.

No single method works for earned media, which is why PR attribution has a reputation for being impossible. It isn't impossible. It's just multi-method.

Method one — tracked destinations

Where a placement includes a link, that link points to a tracked URL unique to the placement. Click-through is countable, and any form submission that follows carries the source through to the CRM record.

Coverage-with-link is the minority of placements in most B2B PR, but it's the cleanest signal available and worth capturing properly.

Method two — dedicated landing pages

For a significant campaign, the client gets a landing page built specifically for it. Where an outlet mentions a resource, report or offer, that destination is campaign-specific.

Anyone arriving at that page arrived because of that campaign. Direct traffic to a URL that exists nowhere else is close to a clean signal.

Method three — campaign phone numbers

A dedicated tracking number, used in broadcast and print placements where a link isn't possible, routes to the client's normal line while recording that the call came from that placement. Calls attributable to a radio segment or a print feature become countable in exactly the way they never previously were.

Method four — the source question

The client's main enquiry form carries a required "how did you hear about us" field with named options, and those options include the outlets where coverage has run.

It is self-reported and imperfect. People misremember. But at volume it produces a usable signal, and it catches the large category of people who saw coverage, didn't click anything, and arrived via search three weeks later.

Method five — the correlation baseline

Before a campaign, you establish the client's baseline enquiry volume — say, 40 to 55 enquiries a month across the previous six months. When coverage runs and the following fortnight produces 96 enquiries, the variance above baseline is evidence even where individual attribution failed.

This is the weakest method and should never be presented alone. Combined with the other four, it captures the halo that direct attribution misses.

What Meridian could show, for the first time

Fourteen months after the rebuild, Meridian secured a feature for a professional-services client in a trade publication with a modest circulation — the kind of placement that would previously have appeared in a report as one line and an impressions figure.

This time it was instrumented. The feature linked to a tracked URL pointing at a campaign landing page. The client's enquiry form carried the outlet as a named source option. A tracking number ran in the client's directory listing for the same period.

The placement produced 340 tracked clicks, 47 enquiries attributed across tracked link and source question, 12 qualified into the client's sales pipeline, and 3 closed deals.

The client's average contract value was in the mid five figures. Three closed deals from one feature, against a monthly retainer of $6,500, is a return that does not require a persuasive slide.

Meridian's account director made a point worth repeating: it was almost certainly not the best piece of coverage they secured that year. It was simply the first one they could count.

How should pitch follow-up work without damaging relationships?

Follow-up should be systematically scheduled and individually written — the timing automated, the words never. The automation triggers a task and supplies context; a human decides whether and what to send.

This distinction is the entire difference between a system PR people will use and one they'll sabotage. Journalists can identify an automated follow-up instantly, and receiving one is a fast route to being filtered permanently.

The problem being solved

Pitch follow-up in most agencies is inconsistent, and the inconsistency correlates with workload rather than with the pitch's merit. A strong angle sent in a quiet week gets two thoughtful follow-ups. The same angle sent during a client crisis gets none.

Across a year, a meaningful proportion of pitches that would have converted simply never got followed up. Agencies that start measuring this are usually unpleasantly surprised.

What the automation actually does

When a pitch record is created, a follow-up cadence is triggered. A typical B2B cadence generates a task at day three, another at day eight, and a final one at day fifteen before the pitch auto-lapses.

Each task surfaces the full context — the original angle, the journalist's preferences, their response history, what else the agency has pitched them recently, and any relevant client news since. The publicist writes the follow-up, or decides not to send one, and either way the record updates.

What the automation guarantees is that no pitch is forgotten. What it never does is send anything to a journalist without a human writing it.

What the cadence is worth in practice

Agencies that instrument this discover the gap is larger than they assumed. Meridian's first full quarter of pitch data showed 61% of pitches had received exactly one follow-up or none, against an internal assumption that two follow-ups were standard practice.

Once the cadence ran automatically, follow-up compliance moved to 96% within two months. Response rate across all pitches rose from 14% to 22% over the following two quarters — not because the pitching improved, but because pitches that would previously have been abandoned after one attempt were being worked properly.

Those numbers also become a management tool. When response rate is tracked per journalist, per outlet and per publicist, you can see which angles land, which outlets have gone cold, and where a relationship needs rebuilding rather than more volume. That is a completely different conversation from the one most agencies have about pitching, which tends to be about effort rather than evidence.

The exception worth making

There is one category where genuine automation is appropriate, and it runs in the other direction: when a journalist responds, an internal alert fires immediately to the relationship owner and the account lead.

Response windows in media are short. A journalist working to a deadline who asks a question and waits six hours has moved on. Sub-fifteen-minute response times on inbound journalist queries are achievable with alerting and impossible without it.

Why do PR agencies neglect new business, and how does a pipeline fix it?

Because everyone capable of winning new business is fully occupied servicing existing accounts, and business development is the only activity in the agency with no client deadline attached to it. A structured pipeline with automated nurture fixes it by removing the requirement for anyone to remember.

The pattern is close to universal in agencies between about eight and thirty people. The founders sold well enough to build the agency, then became the senior delivery resource, and now their week is consumed by client work. Business development happens in bursts when utilisation dips, which is precisely when it's too late.

The specific leak

Most agencies of this size have a substantial dormant pipeline they've never worked. Prospects who took an introductory call eighteen months ago and weren't ready. Pitch finalists who lost by a narrow margin. Former clients whose budgets were cut. Referrals that arrived at a busy moment and were answered late.

Each of those is a warm contact with no follow-up mechanism. In a relationship business with long buying cycles, that dormant list is often worth more than any cold outreach programme.

What gets built

A new-business pipeline mirroring how PR is actually bought — enquiry received, discovery call, chemistry meeting, brief received, proposal submitted, pitch presented, decision pending, won, or lost with a reason.

Automated nurture behind it, running slowly and on substance. Prospective clients receive relevant coverage the agency secured in their sector, commentary on developments affecting them, and occasional invitations. Monthly at most. No pressure, no sales sequence.

Lost-pitch re-engagement, which is where the disproportionate return usually sits. A prospect who chose another agency eighteen months ago is a warm contact whose incumbent relationship is now statistically likely to be under strain — average agency tenure in PR sits at around two to three years. A scheduled re-approach at twelve and twenty-four months costs nothing and converts at rates that consistently surprise people.

Referral capture, so that when a client mentions a peer who might need help, that becomes a pipeline record rather than a remark in a meeting.

Meridian's experience here was instructive. Working their existing dormant list — 210 contacts, none of them cold — through a nurture sequence produced 14 conversations in the first four months and 2 new retainers, at a combined value of $9,400 a month. No new lead generation was involved. The contacts had been sitting in two inboxes and a spreadsheet for years.

How do you make retainer renewals predictable instead of hopeful?

By scheduling the commercial conversation across the whole retainer year rather than concentrating it at the renewal date, using automated reminders that force quarterly business reviews, mid-term check-ins and a structured renewal approach to actually happen.

Retainer renewals are decided long before the renewal meeting. By the time the conversation happens, the client's internal position is largely formed, based on what their champion has been able to demonstrate internally over the preceding months.

The cadence

A structured renewal year, driven by automated reminders that fire to the account lead with prep tasks attached.

Each of those is an automated reminder in GoHighLevel with a task, a checklist and links to the reporting the meeting requires. None depends on anyone remembering.

Why the mid-term health check earns its place

The month-six check-in is the highest-value item on that table, and it's the one agencies most often skip.

Client dissatisfaction in PR is rarely announced. It accumulates quietly — a report that didn't answer the real question, a slow response during a busy period, a sense that the senior person who pitched has been replaced by juniors. By the time it surfaces at renewal it has hardened.

A structured six-month conversation that explicitly asks what isn't working surfaces those issues while they can still be addressed. It is uncomfortable, and it is considerably less uncomfortable than a non-renewal.

Meridian's client retention across the twelve months following implementation ran at 88%, against 71% in the year preceding it. Two accounts renewed at higher fees on the strength of documented attribution.

What replaces the manual monthly slide deck?

An automated report that pulls coverage, pitch activity and attributed enquiries directly from the CRM, delivered on schedule, with the account team's commentary layered on top rather than the whole document being assembled by hand.

The manual monthly report is one of the largest uncompensated costs in a PR agency. An account executive spending six to ten hours a month assembling clippings, chasing circulation figures and building slides, across eight clients, is spending most of a full-time role on document production.

What automated reporting includes

Coverage secured in the period, pulled from coverage records with outlet, date, journalist, type and reach. Pitch activity — sent, responded, converted, with response rate. Attributed enquiries by source placement. Pipeline movement where the client shares that data. Media relationship development, including new contacts established and outlets newly engaged. Upcoming activity for the next period.

That data assembles itself. What the account team adds is the interpretation — why a particular angle worked, what shifted in the media landscape, what the recommendation is.

The time saving is real but secondary. The primary benefit is that the report now contains a commercial section, and the client's champion has something to take to their CFO.

The client-facing dashboard

Where a client wants it, live access to their own coverage and attribution data, on demand rather than monthly.

This is a stronger retention mechanism than most agencies expect. A client who checks a live dashboard is engaged with the work continuously rather than reviewing it in a monthly batch, and continuous engagement correlates strongly with renewal.

What is a snapshot and why does it matter per client?

A snapshot is GoHighLevel's reusable template of an entire account configuration — pipelines, workflows, custom fields, forms, calendars and reports — that deploys into a new client sub-account in minutes rather than being rebuilt each time.

For a PR agency running multiple client accounts, this is what converts a system into an operating model.

What the PR snapshot contains

The media contact and outlet record structure with all custom fields. The pitch pipeline with its stages. Follow-up task cadences. Coverage-to-enquiry tracking configuration. The client reporting template. Renewal and QBR reminder sequences. New-business pipeline and nurture, at agency level.

When Meridian onboards a new client now, the account structure deploys from the snapshot and requires customisation rather than construction — sector-specific outlet lists, client-specific pipeline stages, the reporting emphasis that client cares about. Setup time went from a fortnight of partial attention to about a day and a half.

Why it's a durable asset

Every improvement made on one account can be pushed back into the snapshot and inherited by every future one. An agency two years into this has a media-relations operating system nobody else has, refined across every account it has run.

That is also the honest answer to what the agency owns when a publicist leaves. Not just contact records — a documented way of working that a new hire can be productive inside within a week.

What did Meridian look like a year on?

Fifteen months after the rebuild began, Meridian had converted the worst quarter in the agency's history into the structural change it had needed for a decade.

The media database held 730 contacts across 260 outlets, up from the 412 reconstructed at the start, with every addition documented at the point of contact.

Pitch logging compliance sat at 94%, achieved through workflow integration rather than mandate.

Attribution was live on six of nine client accounts, with the remaining three limited by the clients' own systems rather than by the agency's.

Client retention ran at 88% against 71% the prior year, with two accounts renewed at higher fees on documented commercial impact.

New business from the dormant list produced two retainers worth $9,400 a month combined, from contacts that had existed in inboxes for years.

Reporting time fell from roughly seven hours per client per month to under two.

And a third senior publicist resigned in month eleven — an unavoidable event in any agency. The handover took four days. Every contact, every open pitch, every relationship note was already in the system. The accounts were reassigned without a measurable dip in placement rate.

That last outcome is the one worth focusing on. The point of the exercise was never to prevent departures. It was to make them survivable.

What does this cost and how long does it take to build?

Initial build is typically around $1,000 and runs two to three weeks, with ongoing management between $400 and $1,200 a month depending on account volume and how much campaign and reporting work you want handled.

Here is what the build actually involves.

Week one — structure. Media contact record design with all custom fields, the outlet layer, the pitch object and pipeline, and the new-business pipeline. This is the week that determines whether the system gets used, so the field set is designed against how your team actually pitches rather than against a generic template.

Week two — migration and automation. Getting existing contacts out of inboxes, spreadsheets and databases into the shared system. Building follow-up cadences, journalist-response alerting, renewal and QBR reminders, and the attribution plumbing.

Week three — client layer and training. Reporting templates, any client-facing dashboards, the snapshot, and training the team on the thirty-second logging habit that makes the whole thing work.

The variable that most affects timeline is not technical. It is how much undocumented history you want reconstructed from current or departing staff. That work is manual, it is slow, and it is the highest-return thing you will do — particularly if someone has already handed in their notice.

If they have, that reconstruction becomes urgent rather than important. There is a window between resignation and access revocation, and it is the last opportunity to capture what that person knows.

Is this you, and what happens next?

This is you if any of the following is true.

Your media relationships live in individual inboxes. You could not, today, produce a list of every pitch sent to a given journalist across the last two years. A senior departure would cost you client relationships as well as a headcount. Your client reports contain impressions but no enquiries. Your retainers renew on goodwill rather than evidence. Your new-business development happens when utilisation drops. There are dormant prospects in your CRM nobody has contacted in a year.

None of those individually is an emergency. Together they describe an agency whose core asset is held by individuals and whose commercial value is undemonstrable — which is a fragile position regardless of how good the work is.

How this starts

A conversation, first, about how your agency actually pitches and what your current turnover exposure looks like. Not a demo. The structure of the media record has to fit how your people work, or nobody will use it.

Then the build, in the three-week shape above, with your team involved in the field design so the system reflects their practice rather than fighting it.

Then ongoing management — running the automation, maintaining the reporting, adding attribution to new campaigns, and refining the snapshot as you learn what works.

You keep everything. The account, the data, the workflows, the snapshot. An agency that has just learned what it costs to lose institutional memory to a departing employee should not immediately hand that memory to a vendor.

The relationships stay yours. They stay human, and they stay with the publicists who built them. What changes is that the agency remembers them too — and that when a client asks what the coverage produced, you have a number.