Count the tools in a working agency and you get to fifteen without trying. A CRM. A sending platform, or three, because deliverability means splitting volume. Inbox infrastructure. A prospect database. An enrichment service. A scheduler. Project management. Time tracking. Invoicing. A shared drive. Slack. A reporting spreadsheet that somebody rebuilds every Monday.
Every one of them works. That is what makes this hard to see.
The job nobody is assigned
Ask who connects them and there is no answer, because it is not a role. It is a residue that settles on whoever is most senior and least protected.
That person copies reply counts out of the sending platform into the client report. They notice that a positive reply from Thursday never got booked, because they happened to scroll past it. They remember that one client’s contract renews next month and the case study promised in the kickoff was never written. They reconcile the prospect list against the CRM after an import doubled some records.
None of it appears on a task board. All of it is load-bearing. And it grows with every client added, so the sixth client costs meaningfully more to run than the second, in a way the pricing model never accounts for.
Why the obvious fix isn’t one
Two responses are available and both disappoint.
The first is another tool, one that promises to be the layer above the others. Now there are sixteen, and the new one needs its own maintenance, its own integrations that cover four of the fifteen, and its own reason to be checked each morning.
The second is consolidation: pick one platform, migrate everything, retire the rest. That project is real and sometimes correct, and it takes the better part of a year. Agencies do not have a quiet year. The migration competes with client work, loses, and ends up half done, which is worse than either end state because now the information is in two places.
Both responses treat the tools as the problem. The tools are fine. The problem is that moving information between them is a person’s job.
What the work actually is
Watch what the connecting person does and almost none of it needs judgment.
Reading a number out of one system and putting it in another is not judgment. Noticing that a reply arrived and nobody responded is not judgment. Checking that what an invoice says matches what was delivered is not judgment. Assembling Monday’s status from six places is not judgment, and it takes two hours.
What does need judgment: whether this prospect is worth a call, whether that client’s scope has quietly grown, whether the campaign is underperforming because of the list or the copy, what to say to somebody who is unhappy.
The two are tangled because the same person does both, so the second gets whatever attention survives the first. On a busy week that is not much, and the judgment work is where the margin is.
Connect rather than replace
The useful move keeps the fifteen tools and removes the person from the middle.
That means something that reads across the systems and acts in them: pulling the week’s numbers into a report without anyone opening the sending platform, surfacing the reply that went unanswered for two days, flagging the invoice that does not match the delivery, assembling the Monday status in the time it takes to read it.
The tools keep doing what they do well. Nothing is migrated, nobody is retrained, and the connective work stops depending on a person being conscientious at the end of a long week.
The test of whether it is working is not how much time got saved, though time does get saved. It is whether the sixth client costs about the same to run as the second. When adding a client stops adding a proportional amount of invisible coordination, the ceiling that the agency has been operating under quietly lifts, and it is usually a ceiling nobody had named.