LBM is one of the largest outbound dialling outsourcers in the UK, making over 125 million calls per year. They have now turned off Answer Machine Detection and have actually seen an increase in sales conversion rates.
They are recommending that all outbound call centres should investigate turning off Answer Machine Detection.
While LBM were making lots of outbound calls they were finding an increasingly strained reaction from the people they were calling.
“There is a price to be paid for outbound calling and that is public dissatisfaction,” said David Walters, Chief Information and Innovation Officer of LBM Direct Marketing Ltd. “Complaints from the general public were becoming a real issue for us.”
The situation was compounded by the latest OFCOM regulations
Up until 2008, OFCOM had measured abandonment rate, but not included false positives in the calculation.
False positive is a term used to describe the situation when a dialler detects a live connect (human answer) as an answering machine, i.e. it falsely detects a positive connection, and drops the call.
Current guidelines are that no more than 3% of live connected calls can be dropped in a 24-hour period. In September 2008 OFCOM announced that they would be including Answer Machine Detection (AMD) false positives in calculating abandonment rates, impacting all companies using predictive diallers. When adding this statistic in, companies need to reassess their dialler strategy to ensure they are fully compliant.
OFCOM’s original estimate was that 20% of all calls dispositioned by the dialler as an answer machine were probably false positives. This would have meant that almost every dialler would breach almost every day and was clearly unrealistic.
Rather than stipulating a set percentage, OFCOM left it to the providers to arrive at a reasoned estimate of how many of the total number of calls dispositioned by the dialler are false positives.
LBM Contact Centre
Arriving at a reasoned estimate will not be easy for many contact centre providers. The consultation document states “Providers may wish to actually test their equipment in order to provide an actual false positives figure to OFCOM in the course of an investigation. Accuracy of AMD could be tested by comparing the differing connection rates when it is on and off or by making test calls to a range of numbers where the actual presence of an answering machine is known in advance. Providers could listen to a range of calls where AMD is being used. Calls where an answer machine is detected could also be passed to live operators for a limited period and this may help to quantify numbers of false positives.”
Taking on board the advice from the document, it would not be acceptable to simply state that the dialler provider claims their AMD is 99% accurate and therefore only 1% of calls dispositioned as AMD are false positives. It requires an explanation of how the reasoned figured is arrived at, which can, of course, be carried out with the support of the dialler provider.
Controlled Testing
To establish the real proportion of false positives detected by the predictive dialler, LBM performed a series of controlled tests . The testing was primarily to measure the real rate of AMD, but also to assess the impact on:
- Sales per head
- Wait time
- DMCs per head (DMC = Decision Maker Contact)
- Conversion
The predictive dialler used by LBM had three options available to use in terms of AMD
- Accurate Detect – this is the most stringent method of detection available
- Quick Detect – the least stringent method of detection available
- Off – no detection used.
LBM used each of these methods for 2/3 weeks to gain an adequate sample of results from which to calculate the percentage of false positive detects. It is important to appreciate that even with AMD on, a proportion of answering machines will get passed to agents – these are known as agent answering machines.
Their key findings were as follows:
- There was very little difference in the overall rate of answering machines detected regardless of which method of detection was used – the percentage remained the same, regardless of whether they were system or agent answering machines.
- Switching off answering machine detection showed approximately an 8% reduction in the total volume of answering machines seen
- This variance is seen as false positive detection
The Results
After each change in detection method, LBM saw an immediate short-term drop in performance. This dip in performance was expected, and as such pre-emptive measures had been taken to ensure that similar issues were not experienced. This work primarily involved workshops with managers and agents alike to manage expectation, and share the real positive results from the trial as detailed below.
- The number of DMCs needed per agent to hit target reduced. In real terms this meant that LBM’s campaigns were making the same amount of profit, but by using less data – a direct cost saving to the business.
- DMC talk time remained consistent – regardless of the number of opportunities received by each sales agent, they were still engaging each for a similar amount of time, indicating no lack of drive or performance from them.
- Wait time reduced significantly. Unproductive time between calls reduced, keeping agents focused and increasing the productive hours of each agent’s day
- Conversion increased – adding weight to the argument that LBM are now getting ‘more from less’ – better performance from less data – driving efficiency.
- Sales per head (SPH) – the ultimate measure of any campaign. The highest SPH result over the 8-week trial period was seen with AMD switched off.
Life without Answer Machine Detection
After exhaustive real-life campaign testing LBM concluded that it is not possible to remain within the OFCOM guidelines of 3% drop rate when factoring in AMD false positives. LBM is the UK’s largest outbound outsourcer and after running the tests over several months and multiple campaigns they have concluded that the actual false positive rate is 8%. Clearly, an 8% false positive rate is outside of the OFCOM instruction.
After consultations with a number of major predictive dialler suppliers and customers who operate their own call centres it was LBM’s view that the standard tests do not accurately measure the false positive rate and that it is not possible to remain within the OFCOM guideline. The LBM testing, however, gives a more secure methodology for determining true abandonment rates.
As a result of this research, LBM have switched off AMD from its campaigns and would recommend that all other companies using predictive diallers undertake a similar process of testing, which are likely to arrive at the same conclusion.
“Once agent behaviours were managed we discovered that the number of sales or positive outcomes has not been affected. We have been able to maintain sales performance through a rigid control and measurement of campaign KPIs and training. From our customer’s perspective, this is good news. We are completely compliant with OFCOM guidelines surrounding dropped calls and false positive factoring; complaints from the general public have been reduced further and sales performance for our clients has been unaffected,” said David Walters.
For further information, please contact David Walters at LBM Direct Marketing Ltd on 0161 616 0604 or via www.lbm.co.uk
