Designing a marketing request intake form that ends the side channel

By eleven on a Thursday morning a designer in a five person studio has nine direct messages open. A resize for the Frankfurt edition. Can you just swap the date on the exhibitor deck. Quick favour, the sponsor logo changed. None of the nine exist anywhere a manager can see, and the capacity plan pinned to the wall says the studio is running at 80 per cent, which is where it has said for two years.

A marketing request intake form fixes exactly one thing, and it is worth being clear about which. It does not make the team faster. It makes the demand countable, and until demand is countable every conversation about capacity is two people trading anecdotes.

Why the side channel exists, and why it is rational

Requesters are not being difficult. A direct message is faster to send, it reaches somebody who will probably say yes, and it skips a queue whose position they cannot see.

Every one of those is a real advantage, and any form that removes them without replacing them will lose. The message costs the requester nine seconds. A form with eleven required fields costs four minutes and produces no visible response for two days. Given that trade, a rational person messages the designer, and the designer, who is helpful and sitting three metres away, does the work.

The Asana Anatomy of Work Global Index 2023, run by GlobalWebIndex across 9,615 knowledge workers in six countries, reported that 58 per cent of the working day goes on work about work: coordination, chasing, searching and status. It also found organisations running an average of 8.8 apps for knowledge workers, rising to 10 at director level and above. Requests scattered across that many surfaces cannot be counted by anybody, and the counting is what you are buying.

What the form has to require

Four fields, all required, and a hard stop at four.

The outcome, in the requester's own words. Not the deliverable. Somebody asking for a banner wants registrations from a partner's mailing list, and knowing that occasionally changes the answer to an email instead. One free text box, two sentences.

The date it is needed by. A real date. "ASAP" is a refusal to answer, and a form that accepts it has learned nothing. Where the date is driven by the edition, the field can offer the edition milestones and let the requester pick one.

The budget owner. A person's name. This is the field that does the most work and gets the most resistance, because naming a budget owner converts a favour into a cost. Requests that cannot name one are usually requests nobody has decided to fund, which is useful information arriving early.

The edition. Which show, which year. Obvious, routinely missing, and the reason cross portfolio reporting on request volume is impossible in most organisations.

Everything else is optional: references, brand notes, prior examples, the format list. Optional fields cost nothing and required fields cost completion, so put the pressure only where it changes the decision. What the form does with the request after it lands, and how it gets sorted, is a triage question with its own logic.

Two weeks of logging, and what the arithmetic says

Before changing anything, count. Ask the studio to log every informal ask for ten working days, in a shared sheet with three columns: who asked, what for, and how it arrived. No judgement, no follow up, just the log.

A five person studio supporting eight editions produced this. Forty one informal asks across the fortnight: 26 in chat, 9 spoken at somebody's desk, 6 by email addressed to an individual. The ticket queue over the same fortnight took 12 requests. Total demand 53, of which 12 were visible, which is 23 per cent.

Now the throughput side. The studio completed 44 items in the fortnight, so it is finishing 22 a week against 26.5 arriving. The backlog grows by 4.5 items a week, and because 77 per cent of arrivals never touched the queue, the tracked backlog looked flat the whole time.

Little's law, from John Little's 1961 proof in Operations Research, gives the last number. The average quantity of work in a system equals the arrival rate multiplied by the average time each item spends in it. With 26.5 arriving weekly and an observed average turnaround of about 1.8 weeks, there are roughly 48 items in the studio at any moment. The board showed 11. A manager looking at that board is out by a factor of four, and no amount of care in reading it will help, because the missing work was never written down.

Three numbers to take from that: 23 per cent visibility, 4.5 items a week of silent backlog growth, and a work in progress figure four times the reported one. All three come from a fortnight of logging and a pocket calculator.

Does a form slow people down?

It slows the first nine seconds and speeds up everything after, which is a trade most requesters accept once they have seen it work twice.

The honest version of the cost is that a well built four field form takes about 40 seconds to complete, against 9 seconds for a message. That is a real tax of half a minute per request, roughly 26 minutes a week across the 53 requests in the example above, spread across everybody who asks.

Against that, the requester gets a position, a date and a name. The chasing disappears, and chasing is the expensive part for both sides. A request sent by message generates on average two or three follow ups, each costing a context switch for the designer, and the switch is worth far more than 40 seconds.

Where forms genuinely do slow things down is when they are used as a gate for work that needs no decision. A logo swap on an existing template does not need an outcome statement or a budget owner. Route those straight through with a two field version, and keep the full form for anything that consumes design judgement.

How short can the form be and still count?

Two fields is enough to make demand countable: who is asking and what they want. That is the version to launch with if the organisation is allergic to process, because a counted request is worth more than a well specified one that never gets submitted.

The four field version is what you want within a quarter, and the way to get there is to add one field at a time and watch the submission rate. If submissions drop when you add the budget owner field, that drop is information about how much of your demand is unfunded, and it belongs in the capacity conversation instead of being treated as a form design failure.

Resist the eleven field version permanently. Every field beyond four has to earn its place by changing what the team does, and most proposed fields change only what the team knows. A brief and review workspace of the kind described on the publishing engine page can carry the long form questions later, at the point where a request becomes a brief, which is a different moment with a different audience.

Closing the side channel without banning it

Banning direct messages fails, because the people most likely to bypass the form are the people most able to ignore a ban.

What works is a reflex. Every message that arrives outside the form gets one reply: the form link, with a sentence saying it will be picked up the same day once it is in. No work starts, no exceptions, and in particular no exceptions for senior requesters, since a single visible exception teaches the whole organisation that the queue is optional for anyone who matters.

Give it six weeks. In the first two people are annoyed. By week four most requests arrive through the form, and by week six the requests that still arrive by message are genuinely urgent ones, which is exactly the filter you wanted. Then publish the queue, because a requester who can see position and date stops asking for status, and the ranking argument moves into public where it can be settled once instead of in eight separate corridors.

Where this stops

A form counts demand. It does nothing about the size of the demand, and a portfolio whose requests outstrip its studio by 20 per cent will still be 20 per cent short after the form is running, with better documentation of the shortfall.

That sounds like faint praise and it is the point. The measurement is what makes the shortfall arguable in a budget meeting, where before it was a feeling held by five tired people.

The other limit is that a form records what somebody asked for at the moment of asking, and requesters are unreliable narrators of their own needs. The outcome field will frequently hold a deliverable, because that is how people think. Reading past it takes a person, usually in a two minute conversation, and no field design removes that. Where forms cause active harm is when they arrive with answers already in them, which is a separate and underrated failure.

Start the log this week. Ten working days, three columns, no changes to any process while it runs. The ratio between logged asks and tickets is the number you take to the capacity conversation, and it is usually worse than anybody in the room expects.