Metrics from all types of data are required — not just one — to produce accurate performance levels.

By September 15, when Apple’s iOS changes roll out for its Mail Privacy Protection (MPP) feature, Phrasee CEO Parry Malm expects more marketers will use the scoring system his company developed to provide an accurate view of performance.

“We threw out preconceptions about what different metrics mean — open clicks, unsubscribes, and conversions — and let the numbers do the talking,” Malm said. “We found that fusing different metrics, information points, and mixing in machine learning techniques called ensemble learning, we could get a more accurate metric to optimize.”

Open rates for email will soon become a less reliable performance metric with the launch of Apple’s MPP feature, he said.

While it presents challenges, the shift also provides opportunities for marketers to become more strategic in how they track email performance and optimize key performance indicators including clicks, conversions, and return on investments.

Phrasee last week introduced a KPI powered by artificial intelligence. It uses advanced statistical techniques to give brands a better picture of how their email performs by combining information from multiple sources, Phrasee Score solves the problem of what Malm calls unreliable “noisy KPIs.”

In June 2021, Apple announced the privacy updates to its email client that will affect parts of email marketing across all devices. MPP will give Apple Mail users the option to block open-tracking pixels, making open rates less reliable.

Rather than downloading images at read-time, Apple Mail will download images at time of receipt, giving a false 100% open rate at the same moment in time for all emails received by those with this feature switched on.

Open rates still matter, but the metric has become “noisier.”

“When a metric becomes ‘noisier,’ it’s still useful, but you need to control the noise through statistical techniques,” Malm said. “That’s what we’ve done.”

Metrics are subject to random noise, variance, and elements of bias, so marketers can combine all the metrics into one and produce what Malm calls a Phrasee score, which is more reliable than using one metric.

Previously, marketers would have said the campaign with the highest open rate or click rate is the best — but that’s not always the case, because each are noisy.

Apple Mail client users represent between 10% and 40% market share of all email opens, depending on the region, according to Phrasee report, which cites SparkPost data.