Prospecting when you're at capacity

What you actually sell

On paper, you sell special machines, automated lines, control cabinets, retrofits, commissioning. In reality, you sell the hours of your automation engineers, your electricians and your mechanics. The rest - the studies, the machining, the wiring, the testing - is the shape those hours take.

And an hour can't be stored. An automation engineer on no project today hasn't produced an hour you'll resell next month: that hour is lost, for good, and it cost full rate.

That changes the whole nature of the commercial problem. You don't have a revenue target, you have a load constraint with two edges:

• Underload - your people are on the bench. It bleeds every day, quietly, and nobody in the company can fix it short-term because the sales cycle is too long.
• Overload - you turn away work you won't see again, or worse, you take it and deliver late. Lateness eats the margin, and it damages the one thing that gets you consulted: the reputation for delivering on time.

So your commercial job isn't to sell more. It's to smooth.

Here's why it's structurally hard, and why it isn't your fault.

Between the moment a need appears at a manufacturer and the moment you bill the first hour, six to eighteen months pass: the need is born, the budget is arbitrated at the next cycle, the consultation goes out, you quote, it drags, the order lands, and commissioning is timed to a shutdown.

Now look at when you prospect.

You prospect when the order book thins out. It's human, it's even rational: it's the moment you have time, when the worry rises, when the boss picks the phone back up. Except what you sow at that moment won't come out until twelve months later - that is, precisely when you're full again, because today's dip will have filled itself by other means.

And during the months you're full, you no longer prospect. So in twelve months, dip again.

Which has a very concrete practical consequence: if you have to prospect while loaded, you must prospect little and sharp. Thirty minutes on Monday, on five deals worth it, not three days of mailing when the bench fills up. This whole guide follows from that.

Most integrators run two businesses with a single file and a single message. Yet they have nothing in common.

Read the last row of the table: these are two opposite games. The project is won in advance, staff augmentation is won on reactivity. And the row before: staff augmentation is the only lever that acts on this quarter's load hole. The project will never save a dip, it'll save next year's.

An integrator suffering from underload who answers project tenders is treating a broken leg with a retirement plan.

At a large client - automotive, aerospace, energy, public transport - you don't sell. You're approved, or you don't exist.

Approval goes through procurement, takes months, and is won on criteria that have almost nothing to do with your engineering: certifications, quality, CSR, safety, insurance, financial capacity, headcount. That's why those lines take up so much room on integrators' websites, they aren't decorations, they're entry tickets.

Two games, then, and you have to run them separately:

• Getting into the panel - long, administrative, procurement-driven. It doesn't earn an hour the day it's signed.
• Capturing deals inside the panel - fast, technical, relational, specifier-driven. That's where the revenue is.

The classic mistake is believing the first is the second. An approval won and not worked is a right of entry into a room where nobody speaks to you. You're consulted to make up the numbers, and you quote for free to validate the price of the company the specifier already wanted.

1. Announced obsolescence

Let's start with this one, because it's the finest and the least exploited.

PLC makers publish their end-of-sale and end-of-support calendars. It's public, it's dated, it's written in black and white, and it comes years in advance.

Think about what that means: your technology partner publishes, for free, a list of non-negotiable deadlines for every manufacturer running on that hardware. A site whose PLC loses support has a dated problem: no more parts, no more fixes, a breakdown that becomes a multi-week production stop. And nobody carries that risk very long.

Retrofit is the only segment of the trade where the deadline comes from the supplier, not the client. It's also the only one where you can call and say 'you have until this date' without it being sales pressure, it's a fact.

The work is the cross-reference: which obsolescence calendar, and which sites run on it. The second half isn't in any database, it's read in job postings, which name the ranges sought, in role descriptions, and in what your teams see on site.

2. The job posting that won't fill

The other gem, and it's specific to this trade.

In the classic guides, a job posting signals a project. Here, it signals a make-or-buy call that's being lost.

A manufacturer looking for an automation engineer, a maintenance technician or a methods engineer first decided to insource. That's a choice against you. But automation engineers are scarce, and that posting has a serious chance of staying online.

So the signal isn't the posting. It's the duration. An automation-engineer posting online for four months, reposted twice, bumped back to the top: that's no longer a hire, it's a confession. The department has been short-staffed for a quarter, the manager is running up delays he has to explain, and the solution he refused in January is starting to look like the only one left.

It's the only signal in this guide that acts on your load this quarter. It needs no database: it needs you to note the first-posting date, and to come back.

3. Industrial investment

New line, new site, extension, relocation, product-range change. The historical trigger of the integration project, and the best covered: press releases, trade and regional press, subsidy announcements, building permits.

Early and merciless window: it plays out between the announcement and the writing of the spec. A line whose spec is written is a line where you're quoting someone else's ideas.

4. The scheduled shutdown

You don't modify a line in production. Retrofits, migrations, safety upgrades happen during shutdowns: summer break, year-end, the annual maintenance shutdown.

And a summer shutdown is decided in February, because you have to order hardware, prepare the programs and block the teams. In other words: your client's decision calendar is deducible from his shutdown calendar, and you already know it for everyone you've worked with before.

Many integrators call in May for the August shutdown. That's three months too late, every year.

5. The new manager

New-works, methods, maintenance, industrial management. A hundred days to show he's changing something - and a specificity of this trade: look at where he comes from. If he arrives from a site where he worked with an integrator, he'll call his. If it was you, you just won an account doing nothing. If it was a competitor, you just lost one, and you'll learn it only at the next consultation.

6. A competitor's failure or buyout

Public, and doubly useful: deals in progress to take over, a seat opening up on panels - and automation engineers looking for work. In a trade where the resource is the constraint, a competitor in trouble is as much an HR subject as a commercial one.

7. Compliance constraints

Machine-safety upgrades, safety, industrial cybersecurity. Never mind the text: what matters is that an obligation imposes a date, and a date makes deferral impossible.

The exploitable signal isn't the regulation - everyone knows it. It's the site that just ran into it: an audit passed, a gap flagged, an inspection remark, an HSE hire.

8. The consultation

The last signal, and the worst. When it arrives, the spec is written, often with someone. If your first information about a deal is the consultation, you haven't missed one opportunity: you've missed three.

Orders of magnitude, to recalibrate at home. What matters is the principle: a signal has an expiry date, and for an integrator it's the writing of the spec. After that, all you do is quote.

Too early: you call on an investment announcement, nobody has a mandate, you get filed under 'revisit'. Too late: the spec is written, the consultation is out, and you tie up an account engineer for three days to make the third quote.

The specifier and the buyer

At a manufacturer, the need is born in production, methods, new-works or maintenance. Purchasing goes through procurement. These are two worlds, two vocabularies, two calendars.

• The specifier - methods engineer, new-works project lead, maintenance manager. He has the problem, he writes the spec, he makes the shortlist. In practice, he decides.
• The buyer - he almost never chooses, but he can bar you. Without approval, the specifier can adore you, he can't order from you.

The rule: the specifier wins you the deal, the buyer lets you exist. Working one without the other produces nothing. And talk one level above the problem - the technician suffers, the department manager arbitrates, the industrial director signs and won't reply to you.

Email: find, then verify

Large-group naming patterns are very regular: one format found, and you have the whole group, all sites.

Verification isn't optional. A bounce isn't a lost email: it's your sender reputation dropping, so the next emails land in spam. And a single verifier is never enough - catch-all servers, very common at large accounts, reply 'valid' to any address; greylisting makes a good address look wrong on the first try. Cross-checking several providers is the only way to decide.

The phone: your contact is on site

As everywhere in industry, the phone works better than you'd think, for a dumb reason: your contact is physically there. There's a switchboard, and it picks up. The direct mobile goes through databases, with partial and very uneven coverage from one provider to the next - each has its sources, its regions, its trades. Cross-checking several providers changes everything: where one finds nothing, another has the number. And don't call before you've verified the email: if the domain bounces, you often don't have the right spelling of the name.

There's one thing this trade has that no other really does: your commercial market and your talent market are the same market.

The world of automation engineers is small. People move around: a methods engineer at a large client often spent five years at an integrator, and your future employees work today at your clients - or your competitors. So the same LinkedIn connection serves twice: to get consulted, and to recruit.

In a trade where the resource is constraint number one, that's not a detail. It's probably the best argument for keeping a LinkedIn network when you're an eighty-person integrator with no marketing department.

And the *new manager* corollary matters more here than anywhere: people take their integrators with them when they change sites. A specifier it went well with is an account that moves on its own, provided the link survives the change of employer. A desk number dies the day he leaves. So does an email address. The LinkedIn connection follows you - and it warns you on its own.

Acceptance is a dated signal

An invitation asks for nothing. It costs nothing to accept, nothing to ignore, it contains no sales attempt. It's the only zero-cost move in your whole prospecting, and the only one you can keep up while loaded.

When it's accepted, it tells you what nothing else will: this person just logged in. A maintenance manager opens LinkedIn a few times a year; you've got one, and the window lasts about forty-eight hours. An acceptance isn't a lead, it's a timestamp.

The real problem isn't sending, it's not missing

You send twenty invitations on a Monday. Three accept on Tuesday, two on Friday, one three weeks later. And then: nothing. No queue, no view of 'who just accepted me and I've never approached'. You're in a project-launch meeting, the notification goes by, and it's over.

The result: most acceptances die on the spot. And a note on cadence: LinkedIn caps invitations, the cap isn't public and it moves. An account sending everything at once gets restricted, and a restricted account is no longer a channel. Spread them out, it's the only rhythm compatible with a boss who has an order book to steer anyway.

The reflex: skills. 'We work in automation, robotics, industrial electricity and mechanics, with a turnkey approach.'

It's true, it's well written, and it's the sentence of every one of your competitors.

The structure, in order:

• The fact - dated, and not coming from you. An end-of-support, an announcement, a posting online since March.
• The consequence in his calendar, not yours - this is where all the value sits, and it's the line nobody writes.
• The comparable reference - same process, same size, and close. Not 'our clients', a specific site.
• A small ask - twenty minutes, not 'talk about your projects'.

If you can't write the second line, you don't have a signal, you have a pretext.

• Prospecting when the bench fills up - it's twelve months too late, and it guarantees the next out-of-sync.
• Answering everything - every quote ties up an account engineer you don't have in duplicate. Answering less is the only saving possible.
• Selling projects to fill a dip - it'll only fill next year's. Today's dip is filled by staff augmentation, or not at all.
• Confusing approval and sale - the panel opens the room, it doesn't make people talk.
• Attacking procurement on a technical need - they don't have that need. They'll send you to a form.
• The skills mailing - 'studies, wiring, commissioning' triggers nothing, because it contains no date.

Define your contexts

Not your target - your contexts. 'Food and pharma manufacturers over 100 employees' is a target. 'Food site in my agency's radius, one of whose lines runs on a range going end-of-support within two years' is a context. Write five to ten, in the form *'when X happens, then they have Z months to decide Y'.* And separate those that feed the project from those that feed staff augmentation: they aren't the same, and they don't serve the same quarter.

Wire up the sources

By decreasing yield: job postings - noting their first-posting date -, makers' obsolescence calendars, investment announcements, trade and regional press, public grants, leadership moves, legal notices. Almost everything is free. Nothing reads in five minutes.

Filter

Three questions, three binary answers: is it one of my contexts, is it in the window, do I have a specifier identified and am I approvable there. A 'maybe' is a no, you don't have the account engineers for 'maybes'.

Close the loop

Note what you handled and what it produced. Without a loop, you'll never learn that your investment announcements always arrive six months too late while your unfilled postings convert one in four. With it, you know in a month, and you stop doing what doesn't work.

None of this requires software. A methodical boss, with a spreadsheet, five well-written contexts and thirty minutes on Monday morning, will beat the vast majority of his competitors - because his competitors prospect when they're scared.

The problem isn't the method, it's holding it *when you're loaded*. That's precisely the moment the watch drops first, and it's the only moment it's useful. Obsolescence calendars move, job postings repost without warning, windows close while you're in commissioning, and filtering - the brick that decides everything - is reading work, not a rule you program once.