One of the main things that I enjoy about working in paid media is that the work is never done. There’s always a problem to solve or an opportunity to take.
I’ll give you an example. When I first started using YouTube and Facebook ads many years ago, I had no idea what I was doing and there wasn’t the abundance of online resources we have now to learn from as both platforms were in their infancies.
I knew I could reach my target audience though because I could see where the organic engagement was. I failed a lot. More than I succeeded. But lessons were learned quickly and often.
The approach that I started to form – and saw success with – was to start from the desired output and work backwards until I found the problem or opportunity. With this, I would make amendments, see a change and then start again. This enabled me to compete with very little spend, not through getting it right the first time, but through taking risks and being responsive to pivot one way or another.
I’ve found over the past few years that troubleshooting performance issues is very often the best way to learn about how campaigns work, how to build them better in the future and how to get more out of them when they are already working well.
Troubleshooting helps you to understand what levers you have to pull and which ones affect which outputs – all of which can then go straight back into your strategy and planning.
“Victory comes from finding opportunities in problems.”
I actually think this frames performance in marketing nicely.
Problems can range in weighting from:
How do we get more sales when we’re already hitting KPI?
(not so much of a problem)
How do we get this channel to convert?
(more of a problem)
The proposition of this as a problem leads us to the starting point of ‘how do we solve it?’ rather than ‘this channel, the campaign (whatever it might be) doesn’t work’.
Mindset is crucial in problem solving. You need to be objective, which can be harder than it sounds when ‘you’ are tied to what you do and the numbers being delivered. You need to be curious and creative in your thinking to figure why often complex problems have a first principle cause.
There are an infinite number of reasons why your campaigns may not be meeting expectations. In this post I’m looking to show my methodology for finding and resolving the problems, illustrated with some examples of how I apply this as a strategic approach. I can’t write for every eventuality but hopefully this is enough of a framework for you to improve your results.
This is in-depth content and it will be a long read so bounce now if you’re looking for quick tips and tricks.
Setting the scene
Not everything you do will work.
This is very important to understand. It’s true for everyone and it shouldn’t come with the a priori connotation that this is a bad thing.
The differentiator is the ability to understand correctly why something didn’t work, move on and test again (bringing forward learnings until you’re unlocking opportunities in the marketplace).
This might take a few days, weeks or months depending on the budget and amount of data you have to work with but it’s the most effective way to work.
Not taking risks isn’t an option. Everything is a risk to different degrees – it’s more of a risk to not do anything when it comes to business growth than taking calculated risks… ironically.
To avoid growth opportunities dying slowly by committee on every arbitrary detail of your campaign, the ability to confidently go live quickly, learn and revise is the paid media superpower.
In my post focused on mindset, I talk about reducing fear of failure. The key starting point to troubleshooting is to flip the narrative, take any and all issues as problems to solve and therefore opportunities for growth. This will help you be more objective in your analysis and more scientific in your approach.
Volume or Cost-Efficiency?
Because there are so many variables involved in performance the best starting point is to cut down analysis and simplify the problem to: is this a volume or cost-efficiency problem?
The way we continue to investigate each of these types of problems is very different from one another.
Once we’ve established if we have a volume or cost-efficiency problem, we’re then ready to assess where our problem lies. If you’re saying both then pick which one is the higher priority to start with. This approach is looking at strategic ways to improve performance rather than very specific technical issues.
There is always an answer to the question, things don’t just stop working for no reason, we just need to put on our detective caps and get to work on our investigation.
So let’s start with scenario 1:
We’re getting conversions but we’re not getting enough.
To work out where the in-efficiency is in the campaign we can use the framework that I use in the measuring success blog post, where we break down the campaign into 3 sections: pre-click, on-click and post-click.
An easier way of thinking about this once you’re past the technical is as my buddy AJ Willcox says AMO (Audience, Message and Offer).
Starting from pre-click, we’re going to be applying Occam’s Razor:
“The simplest explanation is usually the right one”
We can bucket pre-click problems into two sub-categories:
- Not able to show (technical issues)
- Not able to show enough (audience size / statistical significance)
If you’re getting no sales or leads at all, it’s first time to check technical issues. Are you actually able to serve ads? If you’re not serving impressions then there could be a technical problem such as:
- Ads not being approved
- Credit card has been declined
- Audience misconfiguration
- Conversion tracking firing incorrectly
- Campaign being penalised (quality score etc)
- Human error
- Bidding strategy mis-configured (unrealistic cost-efficiency target)
Checking the change history for something as simple as an ad group or a particular high volume keyword being paused or accidentally removed (but the campaign being on still) can solve a lot of problems. It’s stating the obvious but it’s worth checking first, as soon as you can discount this from the process, you’re closer to the solution.
If you’re seeing no conversions but you are driving traffic then it’s worth checking that the conversion tracking is actually firing correctly. You can’t make any assumptions. Websites change frequently and conversion tracking can be knocked out very easily.
Quickly checking conversion tracking will avoid you spending hours trying to reinvent the wheel on something like ad copy and making unnecessary changes to the campaign – whilst it could be working very well.
You need to have enough understanding of conversion tracking to be able to take ownership of it. Relying on other teams and people for such an integral part of campaign performance is sub-optimal. With machine learning and automation, your bidding is only as good as the data you’re feeding in so you should be the one doing the feeding.
If the conversion tracking is working effectively, it’s worth looking at other marketing channels in a tool like Google Analytics. This can give you an indication if this is a wider issue. If the website isn’t converting broadly speaking then you likely have a CRO issue more than a pre-click issue. Skip ahead to post-click if this is the case.
Moving on to audience size issues. If you aren’t achieving the sales or lead volume you need, it’s worth forecasting out as if the audience is small or potentially too small (statistically speaking), then it’s perhaps correct that you haven’t seen a conversion as yet.
If the audience is too statistically small, you can model out what you would need your conversion rate to be to achieve the goal you have and then think what would need to go into making that conversion rate realistic.
If the conversion rate would need to be unrealistically high, this is where you can start looking at expanding targeting, whether that’s in or out of this campaign (providing that this campaign is profitable and just low volume).
This is often an issue, for example, with retargeting. Retargeting audiences are finite as they are built from more often than not, website sessions. They have low CPA and high ROAS as the intent level is there but you can’t scale them due the reliance on other sources to populate the audience in the first place.
The same can be true for search too. If there’s only a certain number of searches per month on a particular query, at a certain point you will need to move to other keywords or perhaps targeting methods or even channels.
If you’re using Smart Bidding, you may want to also check your configuration here. If you’re using a target strategy but have strict targets, your audience may be viable but you’re not able to enter enough auctions to drive a meaningful enough amount of data for a lead or sale to occur.
So, providing that volume in your audience is adequate to achieve targets, and there are no technical issues, we can move on to on-click. Essentially, what we’re looking for here is when your ads do show, does the message you’re displaying match the audience that you’re targeting?
Considerations here are where the audience you have selected is in terms of intent level.
Imagine you have a high street store, someone comes in and looks for a particular product. They’re actively searching for the product so have qualified intent to purchase it.
You see someone going into your competitors store that sells similar products, that person has predictive intent as it’s more than likely they would be interested in your products, however, you don’t know that for sure as yet.
You see a random person walk by who meets some of the common demographics your customers have, age, gender, location. This is deductive intent, they may be interested, they may not be but you have some work to do before you would know that.
You would talk to person 1 differently to close the sale than you would person 3, right?
Would you tell the person in your hypothetical store that you sell the product they’re already looking at? No. They know that you sell it and they already want it to some degree. You would say something to incentivise them to choose you as the place for their purchase.
Would you tell the random person walking by why they should buy from you (skipping what you sell)? No. They would probably look at you wondering what’s going on, especially if they don’t necessarily even want/need what you sell.
Ensuring that your copy meets the intent of the audience that you’re targeting means that you’re being efficient in converting from the impression to the click. I’m being careful here not to indicate that high CTR (in isolation) is a measurement of success as it isn’t.
You want to have the most efficient engagement rate you can for maximum volume of conversions (sales or leads) from your audience. Thinking about your copy matching your target audience is important as if you have adequate audience size but aren’t getting enough interactions statistically to get the conversions that you need then this could be your problem.
This doesn’t then necessarily extend then to ‘the more interactions you get the more conversions you will get’ (many optimisations get taken out of context like this).
Review your messaging themes in your campaign and ensure that it matches what type of intent your audience has. Now is also a good time to make sure that you have a clear proposition and differentiation.
Look at your auction competitors and ensure that it’s clear why you are different or better. More often than not businesses are using very similar, generic features / benefits and messaging on their websites, which means that those same generic ideas get cut and pasted into their ad campaigns.
In the machine learning world of paid media, this is where human input is not only necessary, it’s also a key area where you can compete and stand out.
Of course, ad copy testing is an on-going process, but ensuring that you have consistency and alignment here should help to ensure you remove any inefficiency in the on-click part of your campaign.
We’re now into the final place where we can increase our lead or sales volume – post-click. This is probably the place where most of the issues lie and where the biggest levers are to pull to create a meaningful and positive change.
Providing that you’ve worked through the prior two steps (so that we know tracking is accurate), audiences are viable, and messaging is correct we need to look at what happens after the ad clicks.
There are a huge number of factors here to consider and review, I’ve broken them out into 2 sub-categories:
- Low propensity to convert (alignment of offer to audience)
- Friction in conversion process (CRO)
Going back to our hypothetical store example, showing someone in your store who’s looking at products, products to buy, aligns very much to their intent level.
Shoving a random product in the face of someone who is just walking past your shop may work 1 in x times, but it’s highly unlikely to be consistently effective.
Matching your post-click offer to the intent of the users ensures that you’re simply not asking for too much and creating a very inefficient paid media campaign.
Providing that your offer is aligned to the intent of your target audience, the next more technical set of factors, as previously mentioned, is all about making it as easy as possible for your potential customer to buy.
There are a number of world experts on CRO and it’s an in-depth, valuable topic in its own right, however, here is a quick set of factors to consider that significantly affect your campaign side performance which is worth reviewing and optimising.
- Product/service propositioning on the website
- Mobile load speed
- Depth of feature/benefits
- Social proof
- Time sensitivity aspects (next day delivery etc)
A change made to any factor post-click is likely to significantly improve the volume of leads or sales in your paid media campaign.
Moving on to scenario 2, looking to improve cost-efficiency. In this section we’re looking at ways to improve CPA, CPL and ROAS. This is a scenario where you are seeing a reasonable volume of your desired goal but you are just not meeting the profitability target.
We can take the same approach to work through our campaign in a methodical order until we find the problem that we need to solve.
There are fewer factors to consider here, but multiple factors nevertheless. The three most common issues that I see here are:
- Economics of the channel / auction (base costs in relation to your targets)
- Bidding strategy settings (using maximise strategies)
- Audience intent (audience too high up in the buyer’s journey)
An ad platform such as Linkedin is an extremely costly ad network to use, when compared to others such as Facebook or YouTube. It may be that the economics of the platform (or auction) make the cost-efficiency target that you have unrealistic or unachievable, even with a very high conversion rate and well oiled machine of a campaign.
For example, if the avg CPC for the auctions you are entering is £20, achieving a CPL of £50 is going to be tough from the get go.
Forecasting comes in here once more. You can work out what you would need the conversion rate to be for this network to be viable and see how reasonable that is based on current performance to achieve.
Business goals and the bigger picture also need to come into play in decision making here. CPL is a secondary metric on it’s own without the wider context of lead-to-customer rate and average deal size. Your CPL may be high, but if it’s driving qualified leads into the sales team then it may well be profitable or even lucrative for the wider business.
If you need to move to a more economic targeting method or channel, that is worth exploring – providing that you’ve done due diligence on what happens after the lead is generated.
The same example can also apply to ecommerce. With particular auctions in search, the base cost of a given auction may mean that a particular keyword or targeting method isn’t cost-efficient for your business, based on your product margins and requirements. The factor to bear in mind here is customer lifetime value as the direct sale from your campaign may not be the only return they bring.
Machine learning based bidding strategies are a factor here once again, if you are using a maximise strategy, you have dictated your preference as aggressive growth over cost-efficiency. Moving to a target strategy, may help you to improve your cost-efficiency, depending on the economics of the auction and also your post-click experience.
Intent of the target audience is likely a given but it works in a similar way to that of the economics of the channel. Your target audience will be somewhere along the buyer’s journey, and where they are exactly will have a bearing on your ROAS or CPA. This is because the closer the audience is to being ready to buy, more than likely the higher conversion rate you will have and as such your cost-efficiency will be better – providing competition levels as mentioned above in economics of the channel aren’t pricing you out.
The challenge with intent is that search often gives us a really good indicator of the user being in-market for the product or service. Advertisers and businesses know this as it’s been around for a long time so those auctions are now overly competitive and not viable in many instances – similar to trying to reach decision makers on Linkedin, if your average deal size is low or your CPL target is too restrictive.
These problems are complex but this is why creative thinking is now such an important part of multi-channel paid media strategy. We have the tools to reach anyone anywhere, but doing so in-line with business goals often requires being able to think past the status quo.
There’s not a huge amount to impact here from a cost-efficiency perspective. The main area to focus on would be if your messaging isn’t aligning to your post-click experience or the intent of the audience you are targeting.
You could be accidentally (or intentionally) engagement-baiting people into interacting, who then bounce when they realise what your actual expectation was for them. This is a hygiene factor check rather than something that you can incrementally tweak to see a needle shifting change in cost-efficiency.
You can check metrics such as ad relevance in platforms to see how you’re faring. However, there are more effective ways to improve your cost-efficiency – providing that your creative be that image or copy is propositioning the offer that follows effectively.
This is the area where the biggest improvements can be made in terms of cost-efficiency. Increasing conversion rate will reduce your cost per acquisition, improve your ROAS and increase your volume simultaneously.
I’ve already addressed a number of ways to do this in the previous section. Thinking about reducing user friction in being able to convert and also ensuring that you have an offer that is competitive in the market place and also relevant to the audience that you have targeted in your campaign are methods that will allow you to improve efficiency.
Your brand can also have a bearing here. It’s plausible that the only factor between your user and their purchase is being familiar enough with your brand to take the final step. You can mitigate this with social proof, reviews and various other aspects previously mentioned. Something to consider here is that if you feel this is a factor, it may be worth looking at focusing some separate activity on brand building to affiliate yourself better with the product or service category that you are in.
Paid social, in particular, can be a highly effective way to do this. Just remember to measure that activity on it’s goal – to improve brand awareness, not to generate a direct return.
Another factor to consider in improving cost-efficiency post-click is to increase average order value (AOV). If you can increase AOV, the currently potentially non-viable ad network or acquisition source could become viable through offering upsells during the checkout process or a free delivery incentive over a certain threshold to increase the average revenue per sale, therefore increasing your average ROAS.
All roads lead to conversion rate
To come full-circle back to the beginning of the post, the main reason why I generally tend to work backwards is that conversion rate is more often than not where your opportunities or issues lie. If you’re seeing performance issues or looking to get more, it’s probably the best place to start.
Improving conversion rate can solve both volume and efficiency problems – it’s a Pareto distribution at work.
CRO and paid media shouldn’t be thought of as mutually exclusive disciplines especially with automation taking over control of the technical day-to-day in campaigns, where you can add huge value and impact business growth comes after the click. It doesn’t just work one way though.
The added benefit here is having to think about how a product or service is propositioned in the market, competitors pricing, business goals, what problems or opportunities your target customers are having and more is all extremely valuable insight to bring back into campaign strategy, ad copy and so on.
The volume vs cost-efficiency paradox
Whilst improving conversion rate can impact both volume and cost-efficiency, it’s important to highlight that optimising for one aggressively in other areas, often leads to a reduction in the other.
Refining your target audience to reduce CPA, for example, will often lead to a lower conversion volume. Adjusting your bidding strategy to increase ROAS, will lead to you ducking out of auctions where it is deemed unlikely to be possible to achieve your ROAS goal – resulting in a lower sales volume.
Conversely, opening up your requirements for CPA, CPL and ROAS can give you more freedom, not just in bidding strategies but also in broader strategy to take more risks and test new channels, campaigns and audience types. This leads to a higher volume of leads and sales.
It’s a balancing act and you’ll more than likely overcompensate in one direction and then need to course correct back the other way. Having this awareness though, is rare, and it aligns you much closer to business impact, rather than just marketing activity.
A lot of this can be mitigated before launch by ensuring that you understand the business goals that you are trying to achieve, meaning that you can then align strategy.
Review your active campaigns to ensure that you have alignment between your targeting, messaging, and post-click experience.
What is the problem that you are trying to solve?
Do you need to increase volume or improve cost-efficiency?
Work through the components of pre-, on- and post-click to workout where the improvements can be made, make the changes and see what happens. This is more than likely to take you on a road that is out of the paid media landscape but it’s a road that you should and need to begin to travel on.
The impacts are big and the value is worthwhile.