Skip to content

BlogCraft

Answering an IT services tender: what actually wins

An IT services tender is won before the writing starts: an honest go/no-go, available consultants, a price read off the grid. Why most responses lose.

9 min read

Answering an IT services tender is mostly won before the first line is written: on the decision to bid or not, on what you knew about the requirement before it was published, and on consultants who are genuinely available on the start date. The technical proposal and the price then decide between offers that have all passed that first filter. Most responses lose because they are sent without having passed it, and because nobody tries to find out why afterwards.

There are numbers behind this, even if they do not come from French IT services firms. In Loopio’s 2025 survey of 1,544 people who answer tenders, the average reported win rate is 45%. Teams put their losses down first to price (61%) and to a competitor or the incumbent (61%), and only 23% to the quality of their own response. That last figure says less about the quality of responses than about the way a team explains a loss to itself.

Should you answer every tender?

No, and the decision not to answer is the most profitable step in the whole process, because it costs an hour where a response costs several dozen. Loopio measures 25 hours of writing per response on average, before counting the time of the consultants you call on, the reviews and the formatting. For a firm of thirty consultants whose sales director writes the proposals himself, that is a week of selling that goes into one file.

The same report says that 83% of teams use a go/no-go grid. The figure is falsely reassuring: a grid filled in after deciding to bid justifies the decision instead of making it. The one that works asks few questions, and it accepts that the answer may be no.

Did you know about the requirement before it was published? Having met the manager behind it, having a consultant on assignment at that client, having been consulted while the specification was written: each of these changes the odds. A tender that arrives unannounced, with three weeks to respond and a very precise combination of technologies, has often been drawn up by someone who knew it before you did, whether the incumbent or the competitor who did the groundwork.

Do you have the consultants, on the start date? Not in the talent pool, not “being recruited”: consultants who are identified, informed and available. If the response depends on a hire to be made in three weeks, the risk is paid for at scoring or at the start.

Does the expected price leave a margin? A contract whose implicit budget sits below the loaded cost of your consultants does not become profitable because you won it.

Not every tender is a lottery lost in advance, though. According to the European Commission’s Single Market Scoreboard, 22% of French public contracts published at EU level received only one bid in 2024. A well-chosen requirement, in a niche you know, can be played against very few competitors. The arithmetic is the same as for a recruitment role worked in open competition: what counts is not the headline value but that value multiplied by your real probability of winning.

How do you read an IT services specification?

By starting with what the buyer will score, not with what they describe. The consultation rules give the criteria, their weighting and often the price scoring formula; the specification describes the need. A firm that reads the second first writes an accurate text on the wrong subject.

The second reading separates requirements from wishes. A criterion is only eliminatory if the buyer would turn down an offer that was perfect on everything else for lacking it, the same rule we apply to a client’s brief in recruitment. Five years on a rare software package, a named certification, three days a week on site: a response is not built the same way around a constraint and around a preference.

The question phase is the third reading, and it is underused. In a public contract, the buyer’s answers are shared with every candidate, which discourages people from asking. That is a poor calculation: a precise question shows that you have understood the subject, and its answer sometimes reveals a volume or a start date different from the document’s.

What makes a technical proposal win?

A proposal wins when the evaluator can score it line by line without searching. It follows the grid in its order, reuses its headings and answers each criterion in this client’s context, with its constraints, its tools and its organisation as you know it.

What makes the difference is rarely the presentation of the firm. It is the understanding of the need restated precisely, the proposed team organisation, how the team ramps up, the governance (who reports on what, how often), and above all the two subjects buyers dread and responses skim over: what happens when a consultant leaves mid-assignment, and how the client gets the knowledge back at the end of the contract. One concrete paragraph on continuity is worth more than three pages on company values.

This work matters because price does not decide alone. The same Scoreboard shows that in France, 22% of procedures published at EU level were awarded on lowest price alone in 2024: in all the others, technical merit enters the score. Private buyers do not publish their weightings, which is one more reason to read the one in the consultation rules before writing the first page.

The most visible flaw is still copy and paste, which gives itself away through another client’s name left in a paragraph or tools this client does not use. The evaluator does not just take points off the criterion concerned: they reread everything else with suspicion.

Which consultants should a tender response put forward?

Only consultants who have been told, have agreed and are available on the start date. It is the rule most often broken in time-and-materials work, and the one whose cost arrives latest.

The temptation is familiar. The firm’s best profile is on assignment until March, the tender asks for a January start, and their dossier will win points. It goes in, on the basis that a replacement will be offered at the start. The client scored one CV and receives another, may refuse the substitution, and above all remembers that the response was not true. The scoring gain turns into a loss of trust at the precise moment the relationship begins.

The opposite temptation exists too: putting forward the benched consultant first, because they cost €5,480 a month for as long as they wait. That is legitimate if they fit the requirement and expensive if they are merely available: a response lost with the wrong profile does not bring bench time down.

Each consultant’s skills dossier then follows the rules we set out in the skills dossier is a sales document: choose, order and rephrase in the client’s words, never add. On form, Cobalt has listed the presentation mistakes that get a dossier set aside, from a generic title to out-of-date availability. A tender adds one more: a dossier that does not take the grid’s requirements in the same order makes the evaluator search, and they do not search for long.

How do you set the price of a response?

By reading it off the scoring formula before calculating it from your costs. Many consultation rules score price in proportion to the cheapest offer, so a few euros on the day rate can weigh more, or less, than a point of technical merit. Simulating your score at two or three price levels takes half an hour, and it is the most useful half hour in the file.

Going below cost to win amounts to signing a commitment rather than following a strategy. In a French public contract, an offer whose price is clearly undervalued and likely to jeopardise proper performance is an abnormally low offer (article L2152-5 of the Public Procurement Code): the buyer must ask for justification before rejecting it, and a large gap with the other offers is not enough on its own to qualify it. But the firm that passes this check has mostly locked in its margin for the full term of the framework agreement.

That is where the price of a response meets a subject we have covered separately: the margin on a long assignment is lost at renewal, as the consultant’s salary rises and the rate stays put. A revision clause read before bidding, and a starting rate that can carry it, avoid finding out in year two that you have won a loss-making contract.

Why do most responses lose?

Because they rarely have only one weakness. The most common pattern is a bid decided out of habit, on a requirement discovered at publication, with a proposal lifted from an earlier response, consultants of whom only one is really available, and a price worked out without looking at the formula. Each flaw is discreet, and together they put the response at the bottom of the ranking.

The last flaw is the most expensive, because it guarantees the others will come back: the loss nobody analyses. In a French public contract, article R2181-4 allows a bidder whose offer was not rejected to ask for the characteristics and advantages of the winning offer, and the buyer must answer within fifteen days. With a private client nothing obliges the buyer, but a ten-minute call is rarely refused to a supplier who asks without contesting the result.

We prefer to read Loopio’s figure in that light. If 61% of teams blame price and the incumbent, it is partly because they never asked what the buyer actually scored. The incumbent often wins because it knew the requirement, and that is exactly what the go/no-go should have noticed before the 25 hours were spent.

What can an agent prepare in a tender response?

Everything that is extraction and checking, and nothing that commits the firm. An agent reads the consultation rules and the specification and pulls out the list of requirements with their weighting, finds the consultants who match and checks their real availability, gathers earlier responses and comparable references, and flags the administrative documents that need renewing, the calendar we described in the context of getting onto a large client’s supplier list.

It can also lay out the facts for the go/no-go at the moment the decision is made: what is known about this client, who last worked there, which consultants are free on the requested date. That is what we have Balt do, and it saves preparation time rather than raising a win rate: we have no measurement that would let us write that an agent wins more tenders.

The response itself leaves the company and commits it to a price, deadlines and named people. An agent can draft the first version, and a person reviews and releases it, under the rule we apply to everything that goes out. The named consultants read their own dossiers before that person does, because they are the ones who will be questioned at the oral presentation.

The fact remains that the tenders most easily won are those whose requirement you knew before anyone else, and that requirement is almost always discovered at the clients you already serve and call too rarely.

Frequently asked questions

  • How do you decide whether to answer a tender?

    By checking, before opening the file, that the firm knew about the requirement before it was published, that it has the requested consultants available on the start date, and that the expected price leaves a margin. If any of these is missing, the response costs dozens of hours for a low chance of winning.

  • What should an IT services firm’s technical proposal contain?

    An answer to every criterion on the scoring grid, in its order and in its words, applied to this client’s context: understanding of the need, team organisation, ramp-up, governance, what happens when a consultant leaves, and reversibility. Text lifted from an earlier response is recognisable on first reading.

  • Can you put forward a consultant who is not yet available?

    Only by saying so, with their real availability date and their agreement. Proposing someone you know is unavailable in order to swap them at the start exposes you to a refused substitution, and what the client remembers is that the response was not true.

  • What should you do after losing a tender?

    Ask for the reasons and, in a French public contract, for the characteristics and advantages of the winning offer, which the buyer must provide within fifteen days. With a private client, a ten-minute call with the buyer often teaches more than any internal review of the response.

Sources

  1. Loopio, RFP Trends & Benchmarks Report 2025 (1,544 respondents)link.loopio.com
  2. European Commission, Single Market Scoreboard, France country data, public procurement 2024single-market-scoreboard.ec.europa.eu
  3. Marche-public.fr, Offre anormalement basse (articles L2152-5 onwards, DAJ guidance)marche-public.fr
  4. French Public Procurement Code, article R2181-4code-commande-publique.com
  5. Cobalt, Dossier de compétences : 7 erreurs qui font perdre des missions (April 2026)cobalt-ia.com

Balt does this work

Read next

Give Balt its first job.

Thirty seconds to add it to Teams and get to work.

Try for free