The death of open tracking? The biggest change in email marketing for 20 years
The Apple announcement that they will from iOS 15 onwards prevent open tracking in Apple Mail has caused a stir in the email marketing industry. Not only will open rates be impacted but there are a whole host of knock-on effects.
What is happening?
Apple Mail, probably around 40% of all opens is going to start downloading all remote content in the email at the point of delivery.
This includes all images including the little 1×1 pixel you use for tracking if someone has opened the email. It also includes any CSS and fonts.
The impact goes way beyond the issue of open tracking and is possibly the biggest transformational issue to hit email marketing. Whole companies entire platforms designed to optimise your email marketing could be rendered useless by this change.
What will be the impact?
All emails delivered to Apple Mail will now appear to have been opened as the open tracking image will be downloaded at receipt. In effect this makes open rate 100% inaccurate. If your ESP doesn’t add a workaround in, then this will make it impossible to judge open rate on those email clients not affected.
Time based content
All images will be cached at the point of receipt rather than fetched at the moment of open. Any content designed to refresh based upon the latest time such as countdown timers will not function as intended in Apple Mail as the targeting time will be very close to the point of send, not open. However, let’s also remember that typically over 50% of opens are in the first hour of delivery, and this change is only impacting Apple Mail. It will be possible to swap content for Apple Mail still and the rest of your list continue to get the time based content.
IP based targeting
Using the IP address of where the recipient has opened the email to serve location or weather targeted content has been a no go for years in Gmail, Yahoo and Hotmail but the change will also prevent it working in Apple Mail. Personally, we are surprised so many real-time email vendors don’t allow a more accurate way of targeting than this anyway. Frankly, if they have been claiming IP is a reliable location targeting method they have been mis-leading at best.
Custom Fonts & CSS
It is expected custom fonts will no longer work thanks to external content being fetched. We don’t know for certain as our own tests on the iOS 15 beta we haven’t been able to see this pre-loading behaviour just yet on any of our test devices. We think it’s because we are based in the UK and only a few international markets are enabled for pre-loading right now.
Open rate by domain has always been a useful signal to determine if you have any potential deliverability issues with campaigns junking. If you can’t determine accurate open rates by domain it will make it harder to quickly see if you have challenges in this area.
Additionally, we all know that suppressing those that don’t open emails in the last year can help solve any deliverability issues. Removing that ability will make it harder still to resolve deliverability problems.
Resend to non-openers
An old tactic in the email world was to re-send the campaign to those that didn’t open. I’ve never been a huge fan, not because of re-sends but by excluding those who opened or even clicked and didn’t buy suggests a lack of interest. In fact if you were to resend to all, and look at where response comes from by their interaction with the first email you’ll see the highest conversion rates for past clickers, followed by openers and then the non-openers.
If you are going to do re-sends this is how I’ve always advocated it, and now email marketers won’t have a choice anyway!
A/B Testing & Send Time Optimisation
Both of these tactics usually require opens data so are under threat. Aren’t clicks superior anyway to make these decisions?
A click is more valuable than an open, and shows greater intent. The problem with clicks is there are far fewer of them. Therefore, if you are running A/B testing, or trying to predict the best time to send an email you’ll have far fewer events to pick from, and therefore it will be difficult to achieve reliable results with statistically significant results.
Let’s do some simple maths. A campaign to 100,000 people might currently get 18% open rate, so 18,000 unique opens. But perhaps it will only achieve 2% clicks (2,000). Trying to get statistical significance of testing or predictions from a tiny fraction of the original data.
Based upon the above numbers to get statistical significance you’ll need to achieve close to a 10% increase in click-thru rate to be sure your test has worked, but from opens just 3%. A/B testing and a lot of the predictive tools will only be in the reach of those with very large volumes of click data.
Yet lets not also forget that this will only impact around 40% of current opens. The other 60% will be OK if you can isolate the data effectively in your ESP or reporting tool so all is not lost, but it does mean a far bit of hassle.
Opens are also a more stable measure as they don’t tend to move by much from send to send, but click-rates can vary enormously based upon the content of the email. Simply getting a heads up of are we trending in the right direction will become more challenging.
What impact will this have on Reignite?
At Reignite we sometimes badge ourselves as a real-time email content platform. We do this because people want a label to understand who we are similar to. And indeed some of the functionality we offer benefits from the current ability to update content at the moment of open.
Yet we really consider ourselves an email personalisation company first and foremost. The main value we offer is around the ease of implementing 1-2-1 content in any email sending platform, the ability to automate content production, and the ability to create visually compelling personalised experiences.
The real-time element is always an added bonus for what we do.
Yet I keep saying this is only going to impact 40% of emails sent, the remaining 60% will function as before. To support that we are lucky we already have the tools and functionality to ensure content is delivered differently based upon the device opened – its how we have built workarounds for Gmail cache and others, and also simply a core part of our targeting suite.
Where real-time is still possible, it will be possible in Reignite.
A secondary element to the Apple announcement is the inability to reference IP address of the recipient opening the email, and infer the location of the user and their weather. As previously stated this is not new news and has been a poor strategy for some time anyway as more than 50% of email clients don’t support it. With Reignite we have always recommended you use first party location data and allow you to specify this in any number of ways from post code all the way through to longitude & latitude or even airport code or their favourite retail store.
What are the falsehoods I should be careful about listening to?
Oh blimey some of the nonsense we have read over the last few weeks on this is toe curling. The same people who were recommending strategies based upon resend to non-openers or subject line testing to improve opens the week before are now claiming you shouldn’t have been using opens to measure anyway!
There seems to be a bit of false bravado going on with so called experts trying to prove they won’t be negatively impacted. News flash – they will.
A lot of the tools that enable optimisation, problem solving and analysis have been taken away or reduced in effectiveness. Email marketing will become a lot harder, require more effort and hands-on skills in data and analysis will be required.
A little less bravado and a bit more creative thinking around how the email marketing industry can adapt to support brands is what is needed. I’d personally love ESPs to spend time coming up with a new standard open metric that takes out any iOS pre-loads from the calculations.
Will other email providers follow the Apple lead?
A rumour has started that Gmail will follow suit. But I’m going to put my neck on the line and predict they won’t. The reason for this is they already do most of what Apple do, except at the point of open and they respect caching (If you tell them not to cache an image they won’t).
There isn’t a huge consumer demand for the Apple change. The real reasons for the change are probably around ensuring emails load quickly in Apple Mail, and as a business which doesn’t have any reliance on media a way of sticking 2 fingers up at the likes of Google just like their change to 3rd party tracking in apps.
The ability to see if someone has opened an email is fairly low down on the privacy stakes compared to the old world where this could be used for cross-device identification of people by 3rd parties – and Gmail have that covered, as they also do on speed of email open with their current cache.
The cost to pre-fetch all content before it is delivered to the users inbox is huge. Even taking into account the fact that they’ll only cache the same image once for all users is it a technical challenge Google can be bothered with? What do they have to gain when consumers aren’t actively bothered by pixel tracking and Google is a business that relies on such tracking methods across the internet?
Assuming Gmail and others don’t follow Apple then the whole argument that open tracking is dead doesn’t hold true – it’s just severely hampered.
As stated before the key will be how the email marketing industry reacts, especially the ESPs who send and track your email campaigns. If they put some effort into changing their reporting by arriving at a new open rate calculation, you’ll not be able to do open targeting at individual level, but trend reporting, A/B testing, deliverability testing and predictions such as send time optimisation will all still be possible.
The fear is though that a trend I’ve observed for a while of ESPs ignoring this and making it the customers problem will continue. See some of the nonsense around deliverability where they blame the customers emails and data collection when in reality it is their crummy IP reputation thanks to selling their services to dodgy senders.
As a result, I believe a new wave of 3rd party reporting tools will become available to the market where they’ll address the laziness of ESPs and support brands by providing the reporting that they need going forward.
The overall view and your next steps
This is the biggest change to email marketing I can think of and I’ve been in this world since the start of the millennium.
By no means is email marketing dead – in fact for once the annual ‘Email is dead’ crowd have yet to surface. But it certainly is a game changer where email marketing has taken a knock.
Lobby your ESP:
What are their plans for ensuring your open reporting isn’t polluted?
Do you have an internal data team that power dashboards of email performance? Now is the time to start thinking about how different data might need to be pulled so you can still see open rates. If you can pull the raw data from your ESPs API on delivered and opens, complete with any user agent data for the open (which tells you if it was iOS 15) you’ll be able to arrive at something to still show open rate trends on non Apple data
Where do you use opens in your tactics? Is it maybe in segmentation of inactive users? Is it used in split testing decisions? Some of these are so fundamental to your day-to-day thinking that you do them on auto pilot. We recommend auditing where you use open data in your decisions so you can start to think about alternatives while you have time. This won’t simply be as simple as switching to clicks and conversions as you probably won’t have the volume of these.
What is the current state of play?
A strongly recommended test is to run a hold out cell from a typical email campaign. Then measure of those who didn’t get the email, how many visited the website or purchased in the next few days, comparing to those that did receive the email. The difference between the 2 groups is how much incremental uplift the email generated, and if you can also see how many that didn’t open, opened or clicked purchased you’ll get an idea of how important the open is. What I’m sure you’ll find is you’ll still see a lot of those that receive but don’t open the email as converting, and therefore help you understand the importance of simply having your brand in the inbox anyway.
Take a leaf out of direct mail:
Postal marketers have never had the luxury of open rates so how have they coped over the years? Firstly, they are able to predict response by analysing past mailings using Recency, Frequency and Monetary metrics (RFM) to see response rate per segment. This usually leads them to cut old data that hasn’t purchased in some time to keep mailing costs down as these are least likely to respond. Email marketers have never really had to worry about cost, and opens is really used to aid deliverability rates. However, deploying the same techniques around RFM and similar segmentation models allows you understand in advance what percentage of possible conversions you can achieve by including or excluding various RFM segments. This is likely to be a close proxy to open rates to achieve the same goals
Go further with personalisation
A step on from RFM is thinking about a single generic RFM set of segmentation, but altering that on a campaign by campaign basis based upon the content. Just because someone hasn’t bought in a year doesn’t mean they’ll have the same expected response to a promotion on shoes as they would beauty products. Overlaying predictive interest data makes this more complex, but ensures you’ll reach the optimum audience, and allows greater volume of emails to be sent without the same list fatigue issues.
The final word – for now
The iOS 15 release is still in beta and not due for rollout until September 2021. At this moment of time there are still inconsistencies with how this is being seen by people across the world. As we get closer to September we expect some changes and the functionality to be enabled everywhere.
We will aim to update this article as we get conclusive answers on certain areas, but the announcement from Apple is unlikely to be reversed – open tracking here and the associated consequences aren’t likely to be reversed or workarounds found.