The NAO calculates the contract with Concentrix delivered estimated savings of £193m against a payment of £32.5m. Estimated savings are assessed as £223m net of opportunity costs of £30m relating to the diversion of HMRC staff to complete Concentrix cases.

The payments to Concentrix included £23.1m in commission and £6.9m that related to mandatory reconsiderations where decisions were overturned and HMRC agreed not to adjust payments to Concentrix, along with amounts for partly worked cases and sub-contractor costs following termination, and additional IT-related costs. Concentrix told the NAO that it made a loss of £20.5m on the contract.

The three-year contract, signed in May 2014, was terminated early in November 2016 amid growing criticism over levels of customer service and Concentrix’s performance in handling claimants’ queries. The NAO found that at one point in July 2015 it was answering an average of 4.8% of calls within five minutes, compared to a target of 90%.

At the time the deal was signed, HMRC estimated that Concentrix would provide additional capacity to investigate up to a further 1.5m awards per year. Savings would come from stopping incorrect claims, reducing overpayments, and the recovery of money already paid out. HMRC expected to pay Concentrix between £55m and £75m over the life of the contract.

By March 2016, HMRC had more than halved its forecast of expected savings from £1bn to £405m. According to HMRC, this was down to a two month delay to the contract start date because of delays in developing the IT infrastructure to transmit and manage cases, and Concentrix working fewer cases than HMRC originally expected. Concentrix, however, said this was as a result of less fraud and error in the system and changes in the mix of cases it was given to work.

Performance targets

The NAO found that between November 2014 and September 2015 Concentrix consistently failed to achieve over half of its performance targets, meeting only 104 of a total 242 applicable monthly performance indicators.

Its performance was worst during the peak renewals period in mid-2015 when in July it answered an average of 4.8% of calls within five minutes (target 90%).

The poor performance meant HMRC reduced Concentrix’s commission payments by a total of £3.5m over the life of the contract, and in October 2015, introduced a further penalty that reduced the commission paid to Concentrix when it failed to meet customer service targets for handling calls and post.

In October 2015, HMRC and Concentrix agreed to vary the contract, introducing a revision to the performance management arrangements and an increase in the level of commission payments. Concentrix’s level of commission increased to 11%, compared with 3.9% (with a possibility to earn 6.9% if savings reached particular thresholds) in the initial contract.

After some improvement, the performance of Concentrix fell again during the 2016 renewals process, partly because of lack of resources in call centres to deal with the volume of inquiries. It had estimated weekly call volumes at around 8,000 during August 2016 but volumes reached six times this level. For example, in the week commencing 15 August, Concentrix received a peak of 48,000 calls, of which 19,000 were unanswered.

In addition, a routine technical update to Concentrix’s systems on 11 August 2016 prevented its staff from accessing or updating claimant details for a total period of 26 hours. This lack of access led to higher call volumes from 12 August onwards. The NAO found evidence that some claimants had to call multiple times to get in contact with Concentrix, or were unable to discuss their award.

The NAO inquiry found that in 2016, the number of provisional awards terminated as part of the high risk renewals process conducted by Concentrix reached 45,000, double the 21,800 anticipated in its plan. The audit watchdog said these higher than expected terminations would have been lower if Concentrix had processed more cases prior to 31 July.

Backlog

By 20 September when the high-risk renewals process was scheduled to complete, there was a backlog of 181,000 open cases. At this point HMRC reallocated a weekly average of 670 (full time equivalent) staff to work on clearing cases and to answer calls. In November 2016 HMRC and Concentrix agreed to terminate the contract with immediate effect. HMRC has said it will not replace Concentrix with another third-party provider and has transferred 243 staff from Concentrix under TUPE regulations to work on tax credit error and fraud interventions.

The NAO found that in total, Concentrix stopped or amended tax credit awards in around 12% of cases investigated, of which 32% of these decisions were overturned following a mandatory reconsideration.

Concentrix has stated that the average length of time for which claimants had their tax credits stopped and then subsequently reinstated was between six and eight weeks. Between November 2014 and mid-December 2016, HMRC had paid a total of £86,815 in compensation for complaints relating to cases handled by Concentrix.

Commenting on the NAO’s findings Frank Field, chair of the work and pensions select committee, said: ‘The NAO lays bare a contract that promised much but delivered very little, often at the expense of vulnerable people. I am concerned to learn that Concentrix was paid £6.9m for decisions to stop benefits that were subsequently overturned on appeal. These cases were not fraud and were not error.

The committee welcomed HMRC’s announcement that it will not again outsource benefit decisions in this way. The key now is that it learns the lessons of Concentrix for its in-house work on fraud and error.’

NAO’s report Investigation into HMRC’s contract with Concentrix is here.