Concentrix’s “New Business” Metric: What Half the Revenue Does and Does Not Show

What Concentrix reported

Concentrix, a customer-experience outsourcer, published third-quarter fiscal 2026 results on 29 September 2026 for the quarter to 31 August. Revenue was $2,453.7m, down 1.2% as reported and 0.5% in constant currency¹. Non-GAAP operating income rose 1.3% to $309.0m, a margin of 12.6% against 12.3%, and non-GAAP diluted earnings per share were $2.92 against $2.78¹. Adjusted free cash flow was $218.3m against $178.8m a year earlier, which the company calls a record for a third quarter¹.

The statutory result is different. A non-cash goodwill impairment of $1,050.0m produced a GAAP operating loss of $910.3m and a net loss of $988.1m¹. The release attributes the charge "primarily" to the recent trading range of the company's share price and market capitalisation¹. The chief financial officer said the same on the call².

Guidance points the same way. For the fourth quarter the company expects revenue of $2.410bn to $2.460bn, a constant-currency decline of 3% to 5%, and non-GAAP operating income of $310m to $320m¹. By my arithmetic that implies a non-GAAP operating margin of roughly 12.6% to 13.3%: profit guided flat to higher on a falling revenue line.

What "new business" contains

The chief executive framed the quarter as an inflection point: "50% of our revenue is coming from business we have won and deployed within the last 3 years since the introduction of AI"¹. The results release does not break that figure down. The call does, in three parts²:

• $3bn from new and existing clients that have "gone through heavy transformation or has AI influencing the revenue".
• $1.3bn of "net revenue resulting from the compression of our traditional services" using the iX Suite AI platform.
• $0.7bn from new high-value services, such as risk and compliance.

The three parts sum to $5.0bn. Nobody states that total on the call; the company slides, as relayed by Investing.com, give "approximately $5 billion"³. That article carries a notice that it was generated with AI support and edited, and I did not view the slide images themselves.

On what was deployed, the chief executive said the quarter brought "61 opportunities, including more than 30,000 advisers live on our iX Suite this quarter"². The sentence does not make clear how the 61 and the 30,000 relate, and it gives no installed base. The slides separately cite more than 400,000 desktops running AI across more than 1,000 clients, a different measure³. Management also expects about $120m of software-licensing annual recurring revenue from iX Suite at the end of fiscal 2026².

My reading of the definition is that it measures revenue touched by AI-era change, not revenue that is incremental. Only the $0.7bn is a new service in the ordinary sense. The $1.3bn is what remains of existing work after compression, counted as new because AI changed how it is delivered. The $3bn is defined loosely enough that a transformed existing client qualifies.

The 30% growth rate, the retention advantage and the margin advantage of the "new business" are management's own figures. The revenue decline, the impairment and the debt are in the results release.

Builder claims, labelled

Four claims about the new business rest on management alone. Growth of 30% in fiscal 2026 and revenue retention four times higher than the traditional business come from the call². A 100-basis-point gross-margin improvement in new wins from fiscal 2025 to date appears in the relayed slides but not in the transcript³. Revenue from new services and technologies above $6bn in 2027 is stated on the call². I found no independent test of any of them.

The call offered little challenge. The three analysts who asked questions covered second-half 2027 momentum, a competitor's agentic product, offshoring and margin stability; none asked about the definition or its components².

What the Q4 decline cannot be assigned to

Management gave three causes for the weaker fourth quarter and did not apportion them². Two hyperscale clients are ending support for certain client sets; the original plan spread that impact into the second quarter of 2027, but the decision is now to finish by the end of Q4, producing "an oversized impact". Automations were deployed "much faster than what we originally expected". And a shift to offshore delivery was "about a 3% headwind this quarter", with the chief financial officer putting it at about three points of growth for the year.

So the evidence supports a narrower claim than the headline suggests. AI-driven automation is one of three stated drags, and the filings give no way to size it.

The vertical figures do not settle it either. Technology and consumer electronics fell 10%, healthcare 17% and communications and media 7%, while banking and financial services grew 12% and retail, travel and e-commerce 6%¹. The release does not link any of these to AI.

Applying the same scrutiny to the success side matters. The record third-quarter free cash flow sits inside a weaker nine months: adjusted free cash flow of $316.0m against $339.3m, and non-GAAP earnings per share of $8.16 against $8.27¹. Nine-month revenue grew 2% as reported and 0.7% in constant currency¹.

The sector, and the balance sheet

Concentrix is not alone. Teleperformance reported first-half like-for-like revenue down 1.7%, with a narrower second-quarter decline helped by "growth in AI-powered and back-office services", and recurring EBITA of €662m against €697m at an unchanged 13.6% margin⁴. That is a different company, period and definitions, so it is context and not a control. It does show that revenue weakness in the sector is not unique to this deployment. It also shows margin held flat at a peer where Concentrix reports 30 basis points of expansion.

The balance sheet limits how much room the strategy has. Goodwill fell to $2,636.7m from $3,671.7m¹. Debt due within a year was $450.0m, against $65.6m at the end of fiscal 2025, with long-term debt of $3,925.4m and cash of $244.7m; adding the two debt lines gives about $4.38bn¹. Management cites net debt of about $4.12bn and targets net leverage of about 2.6 times by year-end and 2.2 times by the end of fiscal 2027, with gross debt repayment above $550m in each year².

Three things are not disclosed: the build and run cost of iX Suite, the economics per adviser, and how compression savings are shared between Concentrix and its clients. The chief financial officer says newer revenue needs less capital expenditure, but gives no figure².

What this proves, and what it does not

Management expects growth in the new business to offset "much, if not all" of the automation losses in 2027, while giving no 2027 guidance².

The figures show an outsourcer protecting profit while its revenue line shrinks; they do not show why the revenue is shrinking. They do not show that the new business is more profitable, since the evidence is management's, with no baseline or replication. They do not show that AI cannibalisation is the main cause of the Q4 guide. They are one company's quarter, reported by the company, for a services model whose pricing mechanics are undisclosed.

What a decision-maker should conclude

Treat "new business" as a disclosure to be decomposed, not a result. Before relying on a vendor's AI-revenue share, ask for three splits: how much is incremental work and how much is compressed existing work; how the savings from compression are divided between vendor and client; and how much of any decline is automation, client loss or relocation of delivery.

For a buyer, the disclosure gives no basis to assume that AI savings in delivery will reach your contract. For an investor or portfolio owner, the sturdier evidence here is margin, cash and leverage, and these currently offset a shrinking revenue line, a $1.05bn impairment and debt that must fall by more than $550m a year.

The test arrives in 2027. If new-business growth does offset automation losses, as management expects, revenue should stabilise while margins hold. If the company reports the metric again without the three splits, the claim stays untested.