In IT consulting, the margin is lost at renewal

The margin on a time-and-materials engagement falls every year because the consultant’s salary goes up and the day rate is renewed as it stands. Nobody decides on that fall: it happens each time a renewal date passes without anyone reopening the price, and over three years of ordinary pay rises it takes about a sixth of the daily margin with it.

What makes the problem expensive is not its size on one engagement but the fact that it hits every long engagement at once and leaves no trace. A lost consultant shows up on the dashboard, and so does a lost engagement. A margin melting by a few euros a day across forty consultants only shows up at year end, when it is too late to recover it engagement by engagement.

Why does an engagement’s margin fall without anyone deciding anything?

Because the two sides of the calculation do not run on the same calendar: the consultant’s cost moves at every pay review, the billed price only moves at signature, and a renewal is almost always a signature on identical terms.

Take a deliberately ordinary example, built on our assumptions, which you should replace with yours. A consultant paid €48,000 gross a year, an employer cost multiplier of 1.45, 205 billed days a year: the cost comes to about €340 a day. Billed at €520, the consultant leaves €180 of daily margin, roughly 35%. The engagement starts well, the client is happy, and it is renewed twice.

Over those three years the consultant gets raises at the French median, which the WTW Salary Budget Planning survey puts at 3% for 2026. The salary reaches about €52,450, the daily cost about €371, and the day rate has not moved. The margin is down to €149 a day, around 29%. Thirty-one euros less a day, over 205 days, comes to a little over €6,300 a year for this one consultant, and no line in the income statement will tell you when it happened.

The example is a conservative one. It leaves out the larger raise given to a scarce profile to keep them, the step up a grade in the collective agreement, and the third year, which is exactly when departures pick up again. We described that peak in the article on the consultant nobody has called, and it has a direct consequence here: the raise you grant to prevent a departure arrives at the very moment the day rate has not been touched since signature.

Doesn’t the Syntec index already protect the rate?

Not in 2026, and not for the reason people assume. The revised Syntec index, which most French IT services contracts use for price revision, stood at 320.4 in August 2026 against 321.1 a year earlier, a fall of 0.22% over twelve months, while median pay rises reached 3%. An indexation clause applied to the letter this year would have lowered the price, very slightly, rather than raised it.

The reason lies in what the index measures. It is built on the payroll and headcount of a panel of companies, smoothed over twelve months. It follows the average cost of labour in the sector, which is exactly its job, and it cannot follow the consultant you gave 6% because a competitor was calling. A market average protects you against everyone’s inflation, not against your own pay decisions.

Then there is a more mundane problem of use. A revision clause does not apply itself: it sets a formula and a frequency, and somebody has to compute the new price, tell the client and put it on the invoice. In many firms nobody is named as responsible for that, and the clause sleeps in the master agreement while purchase orders are renewed at the original price.

Why does nobody renegotiate at renewal?

Because a renewal is experienced as a commercial win, and you do not reopen a win. The account manager who secures an extension has locked in revenue; asking them to use the same exchange to request an increase is asking them to risk what they just won.

That reflex has its logic, and we have met it before in another form. Firms make most of their revenue from clients who come back, and that recurrence has a price that is rarely spoken: you stop challenging a client you have served for a long time. We wrote about it in the piece on the repeat clients nobody calls, and pricing at renewal is its most direct expression. The loyal account is the one whose margin falls furthest, because it is the one you least want to disturb.

The second brake is the procurement calendar. At a large client, the next purchase order is often issued by procurement on the basis of the previous one, several weeks before it expires. If the revision request arrives after that, it is no longer discussed against an upcoming renewal but against a document already approved, and the buyer has every reason to push it to next year. The window exists; it simply opens earlier than the end date everyone has in mind.

So the problem is almost never a lack of argument. The account manager knows the consultant got a raise, and knows their value to the client has grown with their knowledge of the system. What they lack is the date in front of them at the right moment, with an up-to-date margin figure, while they handle twenty other deadlines and the bench demands their attention every day.

When should the pricing conversation with the client start?

Two to three months before the current purchase order ends, and always after the pay review of the consultant concerned has been set. The order matters: you do not negotiate a price whose cost you do not yet know.

In practice the sequence comes down to three dates per engagement. The first is the consultant’s pay review, which sets next year’s cost. The second, a few weeks later, is the calculation of the real margin at a constant day rate, which tells you whether there is anything to ask for and how much. The third is the conversation with the client, placed before procurement issues the next purchase order. None of these steps takes any particular talent; each one requires having been written down somewhere before it arrives.

The request itself works better stated as fact than pleaded. A consultant who has known the client’s systems for two years costs less to keep than a replacement to train, and the buyer knows it as well as you do. What persuades them is not a generic market percentage but the combination of a date kept, a modest and justified increase, and a counterpart who is not discovering the subject on the day it falls due.

What if the client refuses the increase?

Sometimes keeping the day rate is the right call, and admitting it is what makes the rest of the approach credible. A consultant on the bench costs €5,480 a month on average, as we set out in the real cost of bench time: an engagement at a 29% margin is far better than no engagement at all, and a strategic client can justify a tight rate.

What changes everything is that the decision is taken rather than suffered. An engagement you choose to keep at a reduced margin, knowing the figure and for a reason you can state, is a commercial trade-off. The same engagement renewed by default is a loss you did not know you had, and it repeats the following year, and the one after. It is the same logic as for the true price of a contingency search: the figure does not make the decision for you, it makes the decision possible.

Cobalt has documented the pressure on the sector’s margins in its analysis of the IT services model through the 2025-2026 figures. Renewal is its quietest side, because it asks nobody to cut a price: it is enough not to raise one.

What can an agent carry, and what must it leave to the account manager?

The calendar and the calculation can be delegated entirely, the conversation not at all. It is the same line as for keeping in touch with consultants on assignment, where the only thing an agent carries is that the meeting takes place.

In concrete terms, an agent with access to the management software and contract records can hold, for each engagement, the end date of the purchase order, the date of the last price revision and the date of the next pay review. It can recompute the daily margin at a constant rate whenever a salary changes, flag the engagements whose margin has dropped by more than a threshold you set, and prepare for the account manager, three months before the deadline, a costed proposal with the engagement’s history. None of this is hard; it is just what nobody has time to do for forty engagements at once.

The proposal to the client, on the other hand, leaves the company, and everything that leaves goes through human approval. The agent drafts, the account manager reviews, adjusts and sends, then leads the discussion with the buyer, because a pricing request is a conversation between two people who will meet again. We explained why that line is not negotiable in the article on approving what an AI writes to outsiders, and it applies to a buyer exactly as it does to a candidate.

That is what we have Balt do in the firms that use it: keep the renewal dates and put the margin calculation in front of the account manager while they can still use it. The negotiation remains theirs.

One broader question remains, and it comes before price: how many of your long engagements have a consultant nobody has called in months. A renewal is easier to negotiate with a consultant who is staying, and that meeting is lost in the same way, for want of a date written down somewhere.