Welcome to Online Advertising Answers, where I share three of the most interesting questions I’ve been asked on Quora.

This month I take a look at the value of celebrity endorsements (on Twitter), how to work *with* the Facebook algorithm to get more conversions from your ads, and the saviour of Google Ads that never was – VCPM.

Let’s dig in.

To ask your own digital marketing questions, just go to our editor’s profile on Quora and ask away!

&

A: Working Out The CTR

To work out the CTR on Twitter from a celebrity, you need to get two stats from them (or their team) – Impressions and Link Clicks. They can get this by clicking on the little graph icon at the bottom of their tweets (while they are logged in).

NOTE: If no link clicks have occurred, then this stat will simply not appear in the list of metrics.

A tweet is supposed to become irrelevant after 2 hours usually, but for a celebrity, it will last a lot longer. I would, therefore, recommend collecting these stats 1 month after the tweet has gone out.

You can then use these stats to work out the CTR using this formula:

Click to enlarge

Or else you can use this CTR calculator to save time: CTR Calculator (Click-Through Rate)

CTR probably isn’t the best stat on its own, however, as with a celebrity you’re also paying for the branding value they bring to the table.

Engagement Rate

Engagement rate of influencer tweets also has value – seeing what percentage of people saw the tweet and did something (anything with it). Just cutting through the noise is a big step forward on Twitter.

To calculate Engagement Rate you just need to find out Total Engagements and Impressions then you can use this formula to calculate it:

Click to enlarge

Or again, you can use this engagement rate calculator for convenience: Engagement Rate Calculator

The higher the engagement rate, the more likely followers of that celebrity are interested in your site. Or at least, the more likely that celebrity can convince their followers to be interested in your site.

Fandom can be a funny thing though – a celebrity’s followers might love your site forever after reading one endorsement tweet, or they might retweet about it and never think about it again.

Therefore in these cases, it’s a good idea to track actual click-throughs, and not just the CTR to see what you are actually getting for your money.

Tracking Clicks

The best way to track a click from a celebrity endorsement is to make a specific link for them, going to a bespoke landing page for their fans. Don’t use this link for anyone/anything else, and it can provide great information.

By keeping the link unique, it means you can just monitor traffic to that page. If the traffic comes directly from Twitter, you can attribute it to the tweet(s). If it comes from other sources, you can see if the celebrities fans have spread the link on their own, which of course has it’s own value.

If you’re selling something, knowing that users have entered through that page makes it much easier to attribute sales to that influencer campaign.

Best of all, you don’t have to ask the celebrity for stats at all – you get them on your end and don’t have to take anything on trust.

On top of this, you can make the landing page which directly relates to the celebrities fans, which is much more likely to convert than a generic one.

Unique Tracking URLs

If you can’t make a bespoke landing page (which you really should) and just have to use a link that already exists, you should add UTM tracking. This doesn’t change where the links go, it just means Google Analytics can break out stats of anyone who clicks this link.

Calculating CPC

If you are ultimately looking to see how much bang for your buck you are getting from a celebrity endorsement, then you should probably do it on a cost per click basis.

How much did each visitor to your site cost you? You can work that out using this formula:

Click to enlarge

Or again, by using this calculator: CPC Calculator (Cost Per Click)

Summary

It can be hard to compare different advertising channels when they are different sizes (in this case, Celebrities with different amounts of followers). This usually would mean using a metric which monitors the rate at which something happens – eg CTR or Engagement Rate.

However, with influencer marketing, you should also always keep your eyes on the prize. If you are trying to get people to go to your site, then comparing CPC makes the most sense. If you’re trying to get people to buy things, then compare your CPA.

Don’t overthink it – decide what you want, and work towards that goal! Other metrics are informative but don’t be distracted by them.

I hope this helps.

[This originally appeared on Quora]

&

A: “Hi Priyanka,

The most effective way to optimise Facebook Ads for conversions is simply to get Facebook to do it for you. To do this you need to choose Conversions as your objective when setting up your campaign. You will also need to have set up a Facebook pixel on your site so that Facebook knows when a conversion happens.

However, that is not all you can do. While Facebook will try and get you conversions if you do both of these things, it will still also respect whatever settings you have used for your campaign.

Note: Facebook uses the term “Result” instead of Conversion, and displays “Cost Per Result” as standard, so you should, therefore, focus on lowering your Cost Per Result to maximise “Results”.

Work out your best targeting options

Before you start optimising your Facebook ads, you need to have some options to optimise. This can be thought of as working out “who” to target.

You can work out who it is you should be targeting by working through these three broad categories:

If you have any data you have already to collected (eg from site surveys, previous ad campaigns, Facebook Analytics, Google Analytics etc) then you should try and translate that into Facebook targeting options.

Actual data on your audience will provide the best baseline for who to target on Facebook.

A standard marketing method (that I actually really dislike) is to come up with personas for different types of people who might be interested in what you’re buying. This basically means a brainstorming session using who you *think* your product is good for.

While I’m not a fan of this type of marketing in general – when you don’t know who your audience is you have to start somewhere. Once you’ve got four or five different sets of people who you *imagine* might be interested in what you are selling, try to translate them into Facebook targeting options.

The results from these personas can vary wildly as it is *imagination* based. However, this brainstorming process will help you when you are actually setting up targeting in Facebook (using these personas as well as the data you already have which is mentioned above).

This is because finding the interests on Facebook that relate to what you’ve come up with does require a bit of creative thinking. For example – you know (or think) that your audience is also very into rock-climbing. This can mean targeting people who have the interest “Rock-Climbing” but it can also mean people who have studied rock-climbing, follow the pages of famous rock climbers, or famous rock climbing brands.

3) Custom Audiences – People who have interacted with your site or page

Everyone advertising on Facebook should use one of it’s most powerful (and simple) tools – Custom Audiences. Custom Audiences are simply a set of users on Facebook who hit a certain set of criteria. Used in conjunction with your Facebook Pixel this can mean people who have interacted with your site in specific ways.

People who already have expressed some amount of interest in your site or product are often much more likely to convert than the rest of the population.

For convenience, let’s refer to the conversion you want as a sale. With this in mind you should set up the following Custom Audiences:

• People who visited your website within the past 7 days.
• People who bought something from you within the last 90 days (if you can reasonably expect repeat sales).
• People who went to your sales page, but DID NOT make a purchase within the last 90 days. (Do this by making an audience from people who did go to your sales page, but didn’t go to your thank you page).
• If you have a large website, you could also make a 1% lookalike audience the people who bought from you within the last 90 days.WARNING: Only even consider doing this if you sold more than 1,000 items in the past 90 days. These lookalike audiences work best if they have a large set of people to “lookalike”, so if you only sold a few things then this lookalike audience will be basically meaningless.
• If you have a very popular Facebook page, you could also set up a Custom Audience of people who have visited your Facebook page in the last 7 days.

Once you’ve made these custom audiences, and personas, and looked through any data you previously had, you’ll have a lot of potential targeting options to test.

You can target any of these things individually, or combine them to make smaller (but better) options. You don’t need to use all of these audiences straight away. Instead, each time you run an Ad Campaign on Facebook you can try out a few of them and see what works best.

When you set up your campaign, instead of just having one ad set within it, you should create multiple, each with their own set of targeting. If you have run previous Ad Campaigns on Facebook, then choose targeting that has worked before. You can also add in some Ad Sets with the targeting options discussed above.

You can also split your targeting using options within Facebook – so if you had a great Ad Set in your last campaign, why not run it again but split in two – with one Ad Set targeting men and another targeting women?

By splitting your targeting like this, you can see which set of targeting performs the best, and either switch off the worst-performing Ad Sets (after they have got through their learning period) or add budget to the best-performing ad sets.

There are a few things to note when optimizing your Ad Sets like this:

• You should always have one ad set that is trying a new idea – different interest targeting or something. In this way, you will constantly expand the knowledge you are gaining about your audience from running ads.
• The amount you can break down your campaign into Ad Sets depends on your budget. I personally wouldn’t break down a budget any further than £10 (or \$15) on average per day per Ad Set. You need to be running enough of each set of targeting to get a real feel for what their performance is, and if you spend too little on them your results won’t really mean anything.
• Your Ad Sets targeting shouldn’t overlap. This means that you need to make sure that if someone is seeing one Ad Set because of your targeting, then you should make sure they don’t see any of the others. For Demographic targeting this is easy enough, but for platforms not so much. If you are using interest-based targeting (which you should) then make sure to exclude those interests in other Ad Sets.

• If you don’t have time to optimise the budget of your campaign manually, you can just use the Campaign Budget Optimisation option, which will share your budget between all your Ad Sets, and move it to where it will do the most good. You can find this on the settings page of your Ad Campaign.
• Make sure to name your Ad Sets clearly, so you can look at them later and understand exactly what they were doing without having to dig into them!

In general, you should have at least 3–5 ads per ad set. If you have an ad or ads which already works well, I would recommend generating simple variations in the text, image, or call-to-action for your best ad(s).

I would also always recommend trying one which is quite different from what you have run before. Just as with Ad Sets, this is so that you can use every ad campaign you run to learn something new about your audience. If the new ad works, great! If it doesn’t, then switch it off.

Setting up multiple ads can be more annoying to set up than multiple Ad Sets (as if you set up 5 Ad Sets and 5 ads for each, that is still only 5 ad sets – but 25 ads!). However, in terms of optimising it is much easier, and simply comes in two phases:

1. In the first half of the campaign (eg Up to Week 3 in a 6-week campaign), from a week after the learning phase is complete, go through each ad set and switch off any ads with a higher Cost Per Result than the overall average for that ad group. You should do this regularly, but how often depends on how much money you are spending – a reasonable rule of thumb is under £100 a week and you should check in weekly at most, over £100, and you should check in every 2–3 days. Doing this gives each Ad Set the best chance at performing well, as different targeting may respond differently depending on the ads running.
2. In the second half of the campaign, you should look at all the ads across the campaign (ignoring the Ad Sets). When you do this, again look for any ads with a Cost Per Result which is worse than the overall average for the campaign and switch them off. This will make sure that your Ad Sets had enough time to do their best, but will also make sure you are focussing your budget where it will perform best. If any ad sets run out of ads due to this process, simply switch them off.

There are a few things to note for this version of optimising too –

• As the Facebook post-click conversion window is 7 days, this means that a conversion could be attributed on Facebook up to a week after the related click happened. Due to this, you might switch off an ad, but then see the conversions creep up making it viable again. Don’t fight this! If an ads Cost Per Result improves again and puts it below average then switch it back on!

Things to keep in mind

• Many ad campaigns will get very very few results meaning that judging parts of your campaign based on limited conversion data will cause you to switch almost everything off! This CAN be a good idea as the majority of your ads and targeting might just have been bad. However, you need to judge all your campaigns based on their ROAS – if an ad campaign is costing more money than it makes, switch it off. Then give it a week (to gather the rest of its conversions) and use whatever info you can get from it to design a new ad campaign.

Click to enlarge

• Facebook audiences can get “worn out”. This means if you keep showing the same set of people the same set of ads, they will work worse over time. This means that you should not advertise to the same audiences all the time & not use the same ads all the time. If your Facebook advertising stops working altogether, stop doing it for a few months (3–6). When you come back, it will probably work better! Also (*Top Tip*) occasionally look back at your old campaigns that worked really well and repeat them – if you haven’t run a campaign in a long time then it might work well again!
• Facebook attributes conversions by only looking at things that Facebook has done. So if someone clicks on an ad on Facebook, then clicks on 50 other types of ads, and then converts, Facebook will still claim that conversion. This means you need to use a different source of information to really be able to judge the success of Facebook campaigns (and all campaigns) – Google Analytics is probably the best choice currently.
• Ad Campaigns shouldn’t run forever. Run them for up to 3 months at most and then review it, and start again. This is so that you can take at least a small break in between ad campaigns (and let your audience rest!), but also so that you will also be able to properly improve what you are working on. People will naturally forgive a long (but successful overall) campaign for a few bad weeks, but will not accept a new ad campaign that never gains traction. Use your own in-built prejudices to get more from your ads!

I hope this helps,

Justin”

[This originally appeared on Quora]

&

A: “Hi Amit,

The difference between VCPM and CPM in Google ads is whether or not you can be sure that someone actually saw an ad.

CPM stands for Cost Per Thousand (M is the roman numeral for 1,000), and refers to how much an advertiser pays to have their ad load 1,000 times. The key word there is ‘load’.

A webpage could have an ad slot at the very bottom of a page, and an ad *could* load there and be chargeable whether or not anyone actually sees it.

Enter VCPM – Here to save the day from overly expensive ads

The VCPM payment model aims to address the problem of ads costing money without being seen. This used to be a huge problem, with some studies saying that over half of ads were never actually seen by anyone.

For VCPM, instead of ad impressions being counted regardless of where they appear on a page they are only counted for ads which are at least 50% on screen for at least 1 continuous second. (That’s for Google anyway – other ad networks often have slightly different definitions.)

VCPM stands for Viewable Cost Per Thousand (sidenote: I personally prefer the original acronym CPvM – Cost Per Viewed Thousand).

Google AdSense calls these viewed (or viewable depending on how you look at it) impressions “Active View Viewable” because “Active View” is their tool which scans to see if an ad is actually onscreen.

Using Viewable Impressions instead of just impressions means that advertisers can be sure that they are actually getting what they paid for. It also gives websites a push to place their ads above the fold whenever possible and to only load ads once they appear onscreen.

Is VCPM better than CPM?

VCPM as a payment model is better than CPM in many ways such as:

• It increases trust between advertisers and websites
• It saves money for advertisers by reducing wastage
• The improved performance of ads, in turn, increases the value of ads and therefore profits for websites
• It reduces the number of ads at the bottom of pages (which perform badly anyway). This increases page loading speeds, which is good for website users (and good for SEO)

However, it’s not all upside. As the VCPM payment model rewards websites for actually showing ads, it also encourages having more annoying ad types. For example anchor ads on mobile phones which display at the bottom of webpages (in front of content).

It also encourages stuffing ads at the top of the page. Above the Fold ads do perform better than ads below the fold, however, there is a limit to what users will tolerate.

The very existence of VCPM also increases inequality amongst websites as it gives an advantage to websites who can afford to make ads which only load once they are on screen (or just before anyway).

In general, though VCPM is a much better payment model than CPM.

So is VCPM going to replace CPM?

You would have thought so wouldn’t you? Unfortunately, however, that doesn’t seem to be the case. It’s been around for years now and is still not that popular.

The reason why is VCPM is in a sort of chicken and egg situation. It won’t be popular enough to replace CPM until most ads are measured as being viewable, and websites won’t make sure all their ads are viewable until they have a reason to.

That’s not to say that most ads are still not seen by anyone. The internet is evolving and ads which are placed above the fold perform better, and so earn more money. This naturally results in sites who use them surviving longer than those who are still cramming ads at the bottom of every page so that no-one sees them.

This type of makes the VCPM model less valuable than it otherwise should be. If ads are generally being seen, why pay more to ensure it? And again, this takes us back to the chicken and egg conundrum of why VCPM advertising doesn’t take off.

Also, there is still no industry-wide agreement on what actually constitutes a viewable (or viewed) impression – or even if it is called VCPM or CPvM!

Sure – if you’re running a branding campaign. Otherwise, you should be running a CPA campaign anyway!

CPM campaigns are traditionally for big companies who want to improve their image. Unless you are that (or if you are great at optimising ads), stick with a payment model that actually gets you the results you want.

I hope this helps,

Justin

[This originally appeared on Quora]

I hope this all helps,

Justin