For many retailers, optimization is seen as an in-channel action, taught at a basic level to new hires or done by an external agency who are tasked with optimizing a couple of channels that make up a fraction of the digital media mix. Senior staff are then responsible for campaign planning, using seasonality or historical data to optimize when and how much to spend, often unaware of the optimizations performed by channel specialists. The strings are playing Beethoven while the brass play Bach.
Without blowing my own trumpet too much, I earn a living optimizing digital channels for retailers across the globe, from Goliaths with more budget than a small country to Davids looking to dip their toes into the world of online marketing. No matter the size of the retailer, when it comes to optimization there is always a process I go through to ensure success, a lot of which starts before a penny is even spent.
Optimizing for retail can be harder than other categories; retailers understand the need for brand activity, but their bread and butter will always be performance marketing, where digital campaigns are dictated by in-store events, supplier-funded activity must be ran in a certain way, and metrics like reach or engagement will always play second fiddle to revenue numbers.
This article is by no means a definitive checklist, as all retailers and channels are different, but for those looking to know more about optimizing for retail read on.
Step 1: Understand your objectives
As much as the overall objective for any retailer is to make as much profit as possible, if you were to take the same approach when optimizing your digital channels, chances are you would pump your budget into shopping and search, hit a point of diminishing returns and never be able to grow your business beyond that point. Which is why, in order to optimize effectively, you must first understand your objectives at campaign, channel and business-wide levels.
This is often where the siloed nature of big retailers’ marketing teams can hold them back. The team responsible for paid social know how to optimize paid social, and your programmatic team knows programmatic. However, your paid social team might not know that you are achieving a lower cost per new user via programmatic or vice versa. Which is why it is crucial for a retailer to continuously share performance across channels and to have someone responsible for taking a channel agnostic view to optimization.
Campaign objectives for retailers can vary wildly, from generating new first-time customers to increasing foot traffic or building a userbase on a new social channel. Digital channels offer retailers the opportunity to go beyond just sales or awareness objectives, you just need to be ‘SMART’ (specific, measurable, achievable, relevant and timely).
Once you understand the objectives of the campaigns you want to optimize, you can start to group them. What activity is there to drive loyalty? Which channels bring in new customers? Or build brand awareness? Campaigns can have more than one objective and can fit into multiple groups, but it is paramount that all activity has a clear objective, otherwise optimization becomes pointless.
Step 2: Ensure the correct measures of success
Your measure of success should align with your objectives. If you want to drive new customers to purchase, then measuring the success of your activity on cost per video view is of no use. Simply saying ‘sales’ as a measure of success is one of my biggest bugbears. I could spend a hundred pounds getting a £5 sale or relentlessly re-target existing shoppers to make sales. Setting a return on investment or cost per new customer target will quantify your optimization decisions and allow you to better understand the success of your channels.
Ensuring the correct measures of success isn’t just about setting quantifiable measurements, it’s also about ensuring your campaigns are set up to measure correctly. I have previously spoken about the need to use third-party tracking and attribution providers in order to correctly track the success of paid media campaigns. This becomes even more important when looking to optimize across different channels and platforms. Having a ‘true north’ for all activity means retailers can optimize holistically rather than relying on individual platforms, who will likely inflate the importance of their platform in the customer journey.
Step 3: Strive for budget fluidity and cross channel consistency
A lot of people dislike the idea of keeping budgets fluid between campaigns and channels. These people tend to have a vested interest in a channel or platform, and this can sometimes be detrimental to the overall performance of your digital activity. Before any activity is even launched, I like to make it clear to my clients that budget splits are not set in stone, and the budget should be spent in the campaigns or channels that best achieve the objectives previously set out.
Sometimes this isn’t possible. Supplier-funded activity is often ringfenced into a particular channel at the request of the supplier, but as a retailer it is important to the optimization process that you can pull and push spend based on the data.
Consistency across channels does not just mean making sure all your images use the same color palette. To optimize, you need to be able to compare apples with apples, and that becomes a lot harder when one channel reports weekly at a top-level using platform data whilst another channel reports back fortnightly by audience using Google Analytics. As a retailer, you should have consistent views across your digital marketing mix.
Step 4: Decide on the level of control you need
Brands looking to forgo the heavy lifting that often comes with campaign creation and optimization see features like smart campaigns, automatic optimization features or tech that dictates budget mixes as a godsend. Why spend time and effort setting up campaigns a certain way when you can just give the platform your assets and it will do the rest? In some cases, this works well and will in fact lead to better results than manual optimization. You just need to be clear as to when and why to hand over some of your control.
Facebook’s Budget Optimizer (a handy little tick box that allows the campaign to put budget in the ad sets with the best performance) saves a marketer time splitting out daily ad set budgets and adjusting them based on previous performance. This is awesome when you don’t favour one audience over another and you are optimizing based on a Facebook metric, such as Thruplays.
However, if your measure for success is ROI and that is being measured outside of Facebook (via GA or Adobe, for example) then Facebook will spend the budget on the ad sets it thinks best using Facebook data, which might not necessarily align with what you’re seeing with your tracking provider. Budget Optimizer also means that you can’t easily move budget to a poorer performing ad set if an optimization requires it, for example a retargeting campaign might heavily favor spend to a basket abandoner ad set over a recent converters ad set.
So if you have an offer on for existing shoppers you would need to remove Budget Optimizer and manually increase the recent converters ad set budget, sending the campaign back into the learning phase.
Step 5: Research and test
Prior to launching a campaign, retailers can optimize their activity with good old-fashioned research. Historical sales data gives a clear indication of what tactics have or have not worked in the past, while demand and seasonality data tells you when to launch and how much to spend. For the big campaigns, survey and focus group data is a goldmine of information, allowing you to get a better understanding of what motivates a potential customer and thus feeding into the audiences you go after and the message you use.
From your objectives and your prior research, you can outline a few different hypotheses you want to test. Optimization becomes a lot harder when you have nothing to test. No matter the size or the complexity of your digital marketing mix, testing gives you more optimization triggers to play with and lessons you can take forward or share across channels.
More complex tests, like running brand uplift studies, will require more forethought – especially around the objectives and the measures of success. But testing doesn’t always need to be complicated. A/B testing creatives and copy can be done with relative ease. Testing audiences only requires that you are granular in your set-up by splitting out audiences rather than lumping them all together.
Step 6: Give it some time
There are some optimizations that can be done almost straight after a campaign has launched, such as excluding certain placements not already in your block list, or minor adjustments to CPC bids. However, too many changes straight away will strangle your performance. Think of your digital activity as a living, breathing organism. In its early days, your activity is still trying to find its feet, so changes like swapping out creative or introducing lots of new audiences will only elongate the time it takes to get up and running.
The likes of Google and Facebook have always recommended you wait until a campaign is out of its learning period before making big changes, and historically this was an arbitrary ‘wait 7/14 days and see’ kind of recommendation. But in the last two years or so, both platforms have started to alert users when a campaign is in the learning phase and when it has come out – a very useful feature for understanding when to start optimizing.
If your campaigns are struggling to get out of the learning phase you are probably spreading your budget too thin. Look to consolidate where you can and pull budget forward in order to get out of the learning phase.
Don’t just set your campaigns live and forget about them until they are out of the learning period; use the time to ensure everything is being tracked as it should be, that spend is flowing at a steady rate (you’ve not spent the daily budget by 2am) and your audience are beginning to complete the desired outcome of your campaign, whether that’s a click to site, purchase or store visit. By the time your activity has come out of its learning period, you should have a statistically relevant level of data available.
Step 7: Analyze the early data
Start by analyzing the data available for the campaign, both in the platform itself and through your tracking. This will likely vary by the channels you are using and the amount of tracking available. I always try to look at the data for several key optimization triggers before making any changes to the campaign, as knee-jerk optimizations can damage a campaign.
Personally, I like to look at the audiences we are targeting first; are they all behaving in a similar way, or are there particular audiences bringing up or down the average performance for the campaign? Are the audiences too big so we’re spending money too quickly without any results? Or too small so they are struggling to serve?
Then I look at creative and messaging, which is why a proper testing plan is so important. If you only have one ad, how will you know which message or visuals drive the most action from your audience? How are users engaging with creative A v creative B? Which version of the copy is getting more users to convert? How does this compare to copy on other channels?
Once I have a good idea of the role both audiences and creatives are playing in performance, I will start to delve deeper into the more granular optimization triggers such as time of day, placements, device or demographic breakdowns. By the end of step seven you should have a list of all the optimization triggers that you would like to change.
Step 8: Prioritize
Only then, when I’ve seen the data that tells me what is and isn’t working across the whole campaign, will I prioritize the changes I want to make. With too few changes money is being wasted, but too many too quickly and the campaign struggles to understand which direction you are pushing it in, leading to a longer learning period.
Prioritize getting smaller changes done quickly, a couple at a time. If an optimization will force your campaign to undergo its learning period again, use this opportunity to make one or two other big changes. Completely changing a campaign (for example, swapping out all creatives or audiences) will be like going back to the start of the campaign, so keep that in mind when you are analyzing your results.
Step 9: Replace, don’t reduce
In the early stages of a campaign, optimization should be more about replacing than reducing. For example, a week into your programmatic campaign you find that your ‘30% off select ranges’ creative is driving a weaker ROI than your ‘Sale now on’ creative. As you are still early into your campaign, switching off the ‘30% off’ messaging means you’d only have one creative version left live. Instead you should take learnings from previous campaigns or other channels and find an alternative for the ‘30% off’ creative. Keep all three creatives live until they have gone beyond their learning period, and only then optimize down to the two best performers.
Reducing too much too soon is especially damaging when optimizing audiences. Pausing all but the best-performing audiences will massively reduce the reach of your activity. There are a finite number of users available to reach and an even smaller amount who will convert, so reducing your activity down to only the best too early into a campaign will dry up the converter pool, leading to weaker performance later down the line.
Step 9.5: Funnel budget
Running alongside your replacement of weaker performing optimization triggers, you should also be looking to funnel your campaign-level budget into the stronger performing triggers. Put higher bids or more budget against the strong converters whilst still feeding your tests enough budget to see if your replacements have worked. Circling back to step four, if you have given too much control over to the platform, funnelling budget becomes much harder.
Step 10: Trim the fat
Replacing weaker optimization triggers can only be done for so long – at some point you will have to start switching things off. By far the easiest optimization to make is moving budget from a weak-performing audience/campaign/channel into one that is already performing well. This point comes back round to the need for sharing results between teams, channel agnosticism and budget fluidity during the optimization process.
Step 11: Repeat steps 7 to 10
Continue to analyze the data, prioritize your optimizations and replace your weaker performing optimization triggers. Once you’re sure that no replacements will beat your best performing optimization triggers, trim the fat. Depending on the length of the activity this step can be done once or a hundred times. The important thing is that it is regularly done.
Step 12: Wash up
Often overlooked once one campaign finishes and another begins, ‘washing up’ a campaign’s performance will mean you can easily replicate the successes of the campaign. The first question when washing up should be “did this activity achieve its objective?” and from there you can analyze performance by optimization triggers and come up with key takeaways that should be shared across the digital channels.
To summarize, optimizing digital channels isn’t just switching off any activity that doesn’t make lots of sales. Retailers must put the leg work in before campaigns even go live to ensure they properly understand what their campaigns are trying to achieve, what good looks like and how they are going to consistently improve performance as time goes on. Optimization should be based on data and quantifiable results, which means it is not always a quick fix – instead relying on testing, waiting, analyzing and testing again.
If you want to discuss how best to optimize across your digital channels, get in touch with the experts here at Summit today on [email protected].