An enquiry has three costs. We only measure one.

Fifteen pence. Two pounds eighty-three. Eight pounds sixty-two.

That is roughly what Socitm’s benchmarking put the average cost of a council contact at in 2011: 15p through the web, £2.83 by phone and £8.62 face to face. I spent a chunk of my career in local government working to that arithmetic, alongside top-task research we ran with Gerry McGovern under our head of web, Kevin Jump. Two ideas came out of it that I have carried ever since.

Those were not simply illustrative figures. They came from Socitm’s Channel Value Benchmarking service, with around 200 councils contributing data in 2011. The important bit was not that £2.83 was some universal truth about a phone call. Councils were measuring broadly comparable things in a broadly comparable way, which meant they could compare their costs, volumes and channel mix with everyone else.

The first lesson I took from it is that once you can price a contact, you can start to price a badly written page. An unclear deadline does not stay a content problem. It arrives as four hundred phone calls, and four hundred phone calls has a number attached to it.

The second came from the title of a deck we gave at the time, Top tasks beyond the homepage. Most people never see your homepage. They land deep, from search, and either finish the job on the page they landed on or go somewhere else. So a top task is not simply a link you feature. It is a job an individual page needs to help somebody complete. Fifteen years on, with more people getting answers from search summaries and assistants without reaching a page at all, that feels more true rather than less.

Local government also tried to measure the other side of the problem. National Indicator 14 was designed to measure avoidable contact: contact of low or no value caused by a council failing to do something, or failing to do it properly.

That is not quite the same thing as failure demand, but it is very close to the question I am interested in now. It also carries a useful warning. Agreeing that avoidable contact exists is easy. Agreeing what counts as avoidable, and measuring it consistently, is much harder.

Higher education has not ignored cost. Recruitment teams talk about cost per lead, cost per enquiry, cost per application and, where the data allows it, cost per enrolment. Those measures answer an important marketing question: how much did we spend to create this interest?

What I see far less often is the other half of the equation. What did it cost us to answer the enquiry? Why did the person need to ask it? And how much of that demand did we create ourselves?

That is the bit about to become much more important.

Why it is arriving now

Institutions did not overlook this because they were careless. In a growing market, there is less pressure to interrogate every contact. If applications are rising and recruitment targets are being met, an enquiry inbox getting busier can look like evidence of success.

That cover is disappearing.

The Office for Students reported that 35.8% of institutions in England ran a deficit in 2024-25, with 42.7% forecast to do so in 2025-26. Sector restructuring costs rose by more than a fifth in a year to £218.2 million.

International recruitment has changed too. Sponsored study visa applications from main applicants were 11% lower in the year to July 2026 than a year earlier. The structural change has been sharper for dependants, with applications remaining around 87% below the level immediately before the January 2024 restrictions took effect.

This is not a doom piece. The OfS figures show considerable variation across the sector, with medium, smaller and specialist institutions collectively holding up better than some of the larger research- and teaching-intensive parts of it.

The point is narrower. When growth becomes harder, organisations start asking what things cost. That is the moment this way of thinking becomes useful.

An enquiry has three costs, and we only really measure one

We have been using one phrase for three different numbers.

The first is cost to acquire an enquiry. How much did we spend on search, social, media, events or other activity to generate that lead? Higher education understands this one reasonably well.

The second is cost to handle an enquiry. How much staff time does it take to read, understand, research, answer and, sometimes, answer again? I see much less evidence of institutions routinely putting a figure on that.

The third is the cost of avoidable enquiry. How much of that staff capacity are we using because a page, process or system did not do its job properly?

The third one needs unpacking, because it is where the word “enquiry” does most of its hiding.

Higher education tends to treat an enquiry as a win because, in recruitment terms, it often is one. Somebody has moved from passive interest to active consideration. Nobody in a recruitment meeting has ever been congratulated for reducing the number of people who want to talk to you.

But the same number can cover very different kinds of contact.

Some enquiries are genuine interest: someone weighing up a master’s degree, comparing options or wanting to speak to a person before committing a significant amount of time and money. We should want more of that contact, and we should protect it.

Some of the rest is what John Seddon calls failure demand: demand created by a failure to do something properly in the first place.

What is the deadline? How much does the course cost? Can I study part time? Where do I apply? Is this programme available next year?

Those are different from someone asking whether an unusual overseas qualification meets the entry requirements, or wanting to understand whether a particular programme is right for their career. The aim is not to label every informational question as waste. It is to identify the contact that only exists because the information was missing, unclear or difficult to find.

That category is not really a lead. It is an invoice for a content or process problem, paid in staff time.

All three costs matter, but they lead to very different decisions. A high acquisition cost might make you change your media mix. A high handling cost might make you look at workflow, systems or staffing. A high level of avoidable contact sends you back to the content and the underlying process.

The important number downstream is still applications and enrolments. There would be little value in making enquiries cheaper to handle if we also made prospective students less likely to apply.

The order changes depending on what a page is for, too. We are also looking at the parts of the school that sell services rather than recruit students, and there the priorities invert. An enquiry with revenue behind it is not something to absorb. You want more of them, better qualified, answered quickly. Same website, opposite job, and the obvious temptation is to build one enquiry mechanism and point everything towards it.

Until recently, we could not have told any of this apart even if we had wanted to.

What the new site actually changed

The honest version is that our old web estate could not reliably tell us who was interested in what, where they were or what they did next. Since the migration, it can.

None of what follows produces a cost per enquiry on its own. What it produces is the raw material for the second and third numbers, which is considerably more than we had.

Course interest by country. Not simply institutional traffic by country, which tells you relatively little, but demand for a particular programme in a particular market. For a school whose applicant base is genuinely global, and at a point when policy and market conditions are affecting countries differently, that gets us closer to a recruitment strategy based on current behaviour rather than historic assumptions.

Enquiry volume, and where enquiries come from. Which pages produce them. Which channels produce them. Which programmes attract large numbers of enquiries but relatively few applications. That does not automatically mean the page is at fault, but it tells us there is friction somewhere between interest and action. The next question is whether that friction sits in the content, the proposition, eligibility, affordability, the market or the way we handle the enquiry.

The gap between interest and enquiry. High traffic with relatively few enquiries might mean a page answered the important questions, which is good. It might also mean it failed to give anybody a reason to act. The number itself cannot tell us which, but for the first time we have enough information to start investigating it properly.

What people ask before they arrive. Search Console query data on individual course pages gives us something close to a continuous stream of top-task signals that nobody has to fill in. It is not a substitute for proper top-task research, and it only shows us part of the picture, but it is free, current and already telling us that some of the questions people bring to Google are not the questions our course pages have traditionally been built around.

None of this is exotic. It is ordinary practice in many commercial sectors. It is simply newer to us, and I would rather say that plainly than pretend otherwise.

The chatbot, and what I am wary of

We are building a chatbot in house. The stated purpose is to limit enquiries, and I want to be careful about that phrasing, including with myself.

The objective is not less contact. It is to absorb the avoidable contact so the enquiries that remain are the ones where a person adds value.

Deadlines, published entry requirements, fees, module structure and basic application questions should be answerable from content we already hold, at any hour and in any timezone. That matters when a good proportion of the people considering studying with you are asleep while your office is open.

In that sense, the chatbot is really a top-task exercise wearing different clothes. It will only be useful if it can reliably complete a relatively small set of important jobs, and choosing those jobs should be an evidence question rather than a workshop exercise.

Three sources point us towards the build list: what people actually ask the enquiry team, what they search for on our website, and what they searched for on Google before arriving. Where those sources overlap, we have a strong candidate for something the bot needs to answer. Everything else can wait.

Three things need building in from day one rather than retrofitting later.

Grounding, first. It answers from our approved content and makes the limits of that knowledge clear. A confidently wrong answer about an entry requirement is worse than no answer, because we own the consequences.

Then a working handoff. The failure mode of almost every deflection tool is trapping somebody with a genuine question in a loop. The route to a person has to be obvious.

And measurement. Containment rate, handoff rate, repeat contact, the questions it could not answer and, ultimately, what happened next. The unanswered questions may become the most valuable output, because they give us a content backlog written by the audience.

The guardrail has to be downstream conversion. A tool designed to reduce enquiries inside a function measured on applications can quietly suppress the contact you needed. If containment rises and applications fall, containment was not a success. I have built something and then tried to break it for exactly this reason.

The number I actually want

The uncomfortable lesson from the council years is that the savings often did not land.

Contacts moved online, the benefit was booked in a business case and the phone lines stayed open with the same staffing because demand had shifted rather than disappeared and nobody had decided what the released capacity was actually for.

Avoided cost and cashable saving are not the same thing. Conflating them is one of the reasons transformation programmes earn their bad reputation.

So I am not putting a savings figure in a slide. I want a baseline.

What does it cost us, roughly, in staff time to handle an enquiry? What proportion of those enquiries could reasonably have been answered without human intervention? Can we reduce that proportion without reducing applications?

Answer those three questions this time next year and we will know considerably more than we do now.

Higher education has measured the cost of an enquiry for years. The more useful question is whether we have been measuring the right one.

Which brings me to the thing I actually want to know. Socitm’s figures existed because councils measured broadly the same thing in broadly the same way and compared the results. Local government even tried to standardise the measurement of avoidable contact.

Fifteen years on, I have not found the student recruitment equivalent.

Individual universities may well know their own number. What I cannot find is a shared benchmark that tells us whether £2, £5 or £15 to handle an enquiry is good, bad or entirely normal, or what proportion of those contacts should never have needed a person in the first place.

If it exists, I would genuinely like to know where it is.

So that is the question. Does anybody in student recruitment put a number on what it costs to answer an enquiry, and compare it with anyone else?