Anyone looking into software for tenders quickly runs into a number. Eight to fifteen percent savings on purchase prices, depending on the source also five to fifteen. The figure appears on vendor websites, in trade articles and in the materials of procurement consultancies. It is so widespread that in conversation it is no longer justified, only quoted.
For a central office proposing a shared platform to its members, that is a problem. Member companies are independent entrepreneurs. They ask questions.
In August 2026 we collected eight publicly available publications that state such a figure. Two procurement consultancies, four software vendors, two trade media. The result in one sentence. The range is quotable, but it is not verifiable.
None of the eight publications states which reference price the calculation was made against. Not even implicitly.
None gives a sample size. Every figure is phrased as “up to” or as “on average”. How many projects sit behind an average is not apparent from any of the sources.
The most frequently repeated single figure says that organizations using electronic tendering achieve average savings of eight percent. It is attributed to an internationally active consulting and benchmarking house, without a year and without a reference. A targeted search for the underlying publication produced no primary document. What can be found is the attribution, not the study.
Across all eight sources the stated values range between two and sixty percent. That is a factor of thirty.
One could take this for carelessness in communication. It is not, and the reason lies in the figure itself.
A saving is a difference. It presupposes a price to calculate against. Which price that is does not change the result by decimals, it changes it by a multiple. The same procedure can be presented as seven percent or as thirty-four percent savings, without either figure being untrue.
A percentage without a reference basis is therefore not merely incomplete. It has no informational value. It can neither be confirmed nor refuted, and it cannot be transferred to your own case, because what was measured remains unknown.
For a buying group this has a practical consequence. A number that gets questioned in the advisory board or the members' meeting and then has no derivation will not carry a decision. It costs trust, precisely among the member companies whose participation the project depends on.
If the widespread figure is not verifiable, two options remain. You can keep quoting it, or you can calculate without it.
We calculate without it and apply five percent, disclosed as an assumption and below every figure we collected. The calculation is based on indirectly sourceable spend, meaning energy plus purchased services and consulting. The two other inputs, return on sales and cost structure of mid-sized companies, come from the KfW SME Panel 2025, a freely accessible and vendor-independent source.
The result turns out sharper than the initial question suggests. A five percent price reduction on this part of costs improves pre-tax profit by 21.5 percent in retail, by 13.7 percent in manufacturing and by 9.3 percent in construction.
The thinner the margin, the greater the relative effect of the same absolute saving. Retail is the sector with the thinnest margin. For retail buying groups that is not a footnote.
It does not prove that the figure of eight to fifteen percent is wrong. Such proof would require a study that we do not have.
It proves that the publications do not allow you to determine what the figure refers to, and that it is therefore verifiable neither as correct nor as incorrect. The difference between those two statements is considerable.
And it applies to our own figure in exactly the same way. The five percent profit effect stated on our homepage stood there without a derivation until this study was published. It therefore meets the same criterion as the other eight sources and appears as the ninth entry in the same overview. The study is the derivation.
Does this mean sourcing software saves nothing? No. It means the widespread figure is not a sound basis for a decision. The effect can be explained, and that is exactly what chapters 3 to 6 of the study do, without adopting a single percentage from the industry.
Why don't you name the sources you examined? Not in the blog, but in the study. All nine sources appear there verbatim, with the retrieval date, so that every figure can be checked. This is about method, not about individual firms.
We have been planning with a savings rate for years. What now? Check which price your own rate is calculated against. If it is the price last actually paid, the rate is sound. If it is the first bid of the eventual winner, it describes something else.
The full study runs to nine chapters plus an appendix containing every assumption with its effect, all calculation paths, the nine sources verbatim and a list of what we discarded and why. It is built so that every figure can be recalculated and repeated with different assumptions.
→ To the white paper
For a central office proposing a shared platform to its members, that is a problem. Member companies are independent entrepreneurs. They ask questions.
In August 2026 we collected eight publicly available publications that state such a figure. Two procurement consultancies, four software vendors, two trade media. The result in one sentence. The range is quotable, but it is not verifiable.
What the materials say, and what they do not
None of the eight publications states which reference price the calculation was made against. Not even implicitly.
None gives a sample size. Every figure is phrased as “up to” or as “on average”. How many projects sit behind an average is not apparent from any of the sources.
The most frequently repeated single figure says that organizations using electronic tendering achieve average savings of eight percent. It is attributed to an internationally active consulting and benchmarking house, without a year and without a reference. A targeted search for the underlying publication produced no primary document. What can be found is the attribution, not the study.
Across all eight sources the stated values range between two and sixty percent. That is a factor of thirty.
Why this is more than a formal flaw
One could take this for carelessness in communication. It is not, and the reason lies in the figure itself.
A saving is a difference. It presupposes a price to calculate against. Which price that is does not change the result by decimals, it changes it by a multiple. The same procedure can be presented as seven percent or as thirty-four percent savings, without either figure being untrue.
A percentage without a reference basis is therefore not merely incomplete. It has no informational value. It can neither be confirmed nor refuted, and it cannot be transferred to your own case, because what was measured remains unknown.
For a buying group this has a practical consequence. A number that gets questioned in the advisory board or the members' meeting and then has no derivation will not carry a decision. It costs trust, precisely among the member companies whose participation the project depends on.
What we do instead
If the widespread figure is not verifiable, two options remain. You can keep quoting it, or you can calculate without it.
We calculate without it and apply five percent, disclosed as an assumption and below every figure we collected. The calculation is based on indirectly sourceable spend, meaning energy plus purchased services and consulting. The two other inputs, return on sales and cost structure of mid-sized companies, come from the KfW SME Panel 2025, a freely accessible and vendor-independent source.
The result turns out sharper than the initial question suggests. A five percent price reduction on this part of costs improves pre-tax profit by 21.5 percent in retail, by 13.7 percent in manufacturing and by 9.3 percent in construction.
The thinner the margin, the greater the relative effect of the same absolute saving. Retail is the sector with the thinnest margin. For retail buying groups that is not a footnote.
What this recalculation does not claim
It does not prove that the figure of eight to fifteen percent is wrong. Such proof would require a study that we do not have.
It proves that the publications do not allow you to determine what the figure refers to, and that it is therefore verifiable neither as correct nor as incorrect. The difference between those two statements is considerable.
And it applies to our own figure in exactly the same way. The five percent profit effect stated on our homepage stood there without a derivation until this study was published. It therefore meets the same criterion as the other eight sources and appears as the ninth entry in the same overview. The study is the derivation.
Frequently asked questions
Does this mean sourcing software saves nothing? No. It means the widespread figure is not a sound basis for a decision. The effect can be explained, and that is exactly what chapters 3 to 6 of the study do, without adopting a single percentage from the industry.
Why don't you name the sources you examined? Not in the blog, but in the study. All nine sources appear there verbatim, with the retrieval date, so that every figure can be checked. This is about method, not about individual firms.
We have been planning with a savings rate for years. What now? Check which price your own rate is calculated against. If it is the price last actually paid, the rate is sound. If it is the first bid of the eventual winner, it describes something else.
The next step
The full study runs to nine chapters plus an appendix containing every assumption with its effect, all calculation paths, the nine sources verbatim and a list of what we discarded and why. It is built so that every figure can be recalculated and repeated with different assumptions.
→ To the white paper