Part 2 – paid search in the moment
Search in my opinion is one of the purest forms of intent a marketer can harness. A user comes to a search box and tells Bing or Google what they want, at the moment that they want it. It's been recently coined as micro-moment marketing.
Along with everything in the digital ecosystem, the search landscape is changing as a result of this moment trend. And this change is also forcing us to take a new perspective on PPC optimisation, a ‘moment-marketing’ perspective if you like.
Fragmentation and choice for a finite demand (in economy terms, "invariable") equates to increased costs, tough competition, pressures, innovation and some suppliers actually exiting the market.
The changing search landscape
In February this year, Google restructured its desktop SERPS, the way search results are displayed - in an obvious trend to unify the search experience across devices.
Where mobiles only have two ad slots at the top, and two at the bottom, with no side search, desktop has now almost caught up. Right hand side ads have been removed, and depending on search queries, there are three-four top slots and four bottom slots. Shopping ads can still show on the right hand side for now.
This is a classic first example of finite demand increasing supply conflict. The first people that have been hit are SMEs with limited budgets. Some markets derive most of the click volume from top slots, so competing at the bottom is not going to make sense from a time invested vs. return perspective.
Top slots will become increasingly expensive. A limited budget in this situation will actually lower the number of clicks you can buy.
Click costs are increasing
Next to be hit will be the bigger companies, as mentioned before (and anecdotally a few years back by Google themselves), the cost of bread is going up – CPCs are going up. Don’t believe otherwise. Everyone needs to find smarter ways of competing, throwing money at it recklessly will stop working even for the big guys.
Fragmentation and choice in search is also manifested in another way, more of an emerging trend right now some would say, but one that is growing so quickly it’s not to be ignored. It’s growing, because it just makes sense.
The rise of conversational search
Voice search right now seems like an out-of-reach novelty to search marketers. But think about it, Google launched the hummingbird algorithm update and powers Google Now (recently also Google assistant); meanwhile Bing powers Cortana, Xbox search and Apple’s Siri – the sheer volume of data these engines are collecting right now can put us in front of predictive search sooner than we expect.
This means we are moving from searches to searchers, from keyword semantics to searcher intent, searcher past behaviour, contextual signals, machine learning and real time optimisation decisions.
This moves us into micro-moments, where the searcher wants immediacy and have their demands met before they’ve actually voiced it, let alone type it.
All these things point to one conclusion – being smart, with data and technology, in capitalising on those moments that matter, is more essential today in search than ever before. There will not be enough money to outbid every single competitor. But there will always be enough innovation. Innovation is the only path to success in such a market.
So what can you do?
During the Attention Summit I explored a few areas where we can all be smarter with paid search, going beyond the basics of PPC.
I’m sure this is just the start, and in fact at the Adwords Performance Summit this week, Google has already announced more features in this vein.
So let’s look at the possibilities of micro-moment marketing translated into paid search optimisation.
The basic idea is that all clicks are not equal
Many companies/agencies obsess over visibility and I have seen many PPC accounts where the primary objective was to get clicks, with no discrimination as to where these clicks actually come from and if they are valuable to the bottom line or not.
If for example I was to search for what is ppc vs. how to choose a good PPC agency vs. contact Adido – you can already see how I’m moving along a continuum of commercial intent, and how clicks from these queries are of different value to us as an agency if we were to bid on them.
So let’s just go through them and clarify:
- Navigational queries – this will be your Adido website or Lloyds Bank login
People know what they want, they just want to navigate to the destination.
These have low commercial value, but can actually be used in paid search – and you can see them in action in brand and competitor bidding; brand will harvest your other activities and ensure competitors are pushed down in the search results.
It’s cheap, it’s effective.
Competitor bidding can come with quality score penalties on cost, but if done right can be very lucrative, not to mention you’re piggybacking on their search volume.
This is an opportunity because Google studies have found that 1 in 3 smartphone users have purchased from a company or brand other than the one they intended to, because of information provided in the moment they needed it.
So navigational queries represent a micro-moment you can use to defend yourself against competitor hijacking, to snatch people from your competitors, or become another brand in the searchers’ portfolio (credit cards).
- Informational queries - can be either completely useless to you as a marketer wanting results (like how old is Barrack Obama?) or can signal the start of the research journey for searchers (like credit cards).
This is where the bulk of search volume lies, but also where the most ambiguity and hence the highest competition and highest CPCs reside.
Most people are aware of the need but not the solution.
Some of these queries, if they’re more of the 'how to' type, can be tackled via organic search, but others in the upper stages of research deserve a carefully optimised paid search approach.
It’s your opportunity to enter the consideration set. Your final conversion objective might be unlikely, but you could be smart at alternative conversions such as newsletter signups, app downloads or whitepaper downloads, it’s a non-committal start of a relationship.
- Commercial-investigation queries are closer to transaction, they are more in the realm of testing a final consideration set, seeking reassurance or comparing.
Paid search is highly valuable here, as well as shopping for ecommerce.
Volume is still quite high but starts to drop as queries get more precise. You can start to potentially gain conversions or higher value alternative conversions (quote but not sale/ prospective callback).
- Transactional queries are the ones that accomplish your final goal or the highest value online goal you have.
Paid search and shopping are excellent for this to harvest this ‘I want to buy’ moment– costs can still be high, but some of these queries could become very product specific and therefore be very cheap but have very low volume.
Understand what people seek rather than focus on what you sell
One thing people always forget in paid search is that what you bid on is not always what people search for, match type dependent.
If I bid on ‘ppc agency’ broad, this could generate a few types of queries:
- ‘how to choose a ppc agency’,
- ‘best ppc agencies Bournemouth’
- ‘adido ppc agency contact’
One keyword with ambiguous, probably informational intent more likely, but three queries with highly different intents and values
So it’s important to understand this distinction, because it’s not only about the semantics of the keywords you bid on, but also the semantics of queries people actually search for, and it’s their search moment that you have to understand, not yours.
A first application of semantics in PPC management
Very granular account structures, where campaigns are prioritised by key search intent areas:
- generic (informational), and
- product/category (investigational & transactional).
Budgets are split accordingly, and sometimes a particular generic query could be split into its own campaign, since it could command so much budget on its own that it could cannibalise spend from any other term.
Then follows even more granular ad group segmentation, splitting within a type of query intent by further attributes: buy vs cheap vs. best etc...
This is so you can further split your big intent categories into smaller micro-intent chunks. The keyword selection then becomes closer to the actual search queries (rather than the ambiguities we spoke about in the Adido case) and therefore the most relevant ad messaging and landing page can be served and you can fully take advantage of that search moment.
Not only are there significant quality score implications hence CPC, position, and impression share benefits, but also the user feels that your ad responds the best to their query, and the promise is kept on the landing page – therefore significant CTR, conversion rate and ultimately ROI benefits.
If you think you’re going too granular – you’re not. This approach is one of your lifelines in ensuring you can still compete in the new search landscape – and if you can’t afford the very generic queries, you know where to prioritise your budgets and not risk sunk costs.
To maintain this structure, 3 things are important:
- Running regular SQRs - to exclude irrelevant queries and add new potential avenues
- Regular ad copy testing - to tweak the ad for relevancy but also for action – catch the right moment, speak to the searcher’s needs
- Conversion rate optimisation – e.g. landing page testing. Remember these queries are either highly commercial so you will want to make it really easy for the searcher to convert – remove all barriers; or they are higher up the funnel, so you will want to provide the content that best addresses their intent and guides them on the path you want them to take.
In time, and depending on the market, conversion rate optimisation efforts will be your best bet in negating the CPC increase impact.
Low intent queries
But what do you do with all the ‘low intent’ queries, especially the ones that are growing so much lately because of the smartphone; the search assistant?
Searches related to "how to" on YouTube are growing 70% year over year, of smartphone users, 91% turn to their devices for ideas while completing a task.
These queries are relevant, and as we saw before, could potentially turn into commercial queries, but there is also a lot of wastage.
Then there are also the informational, very broad queries like ‘guttering’. This can be highly commercial – or highly generic – but they have loads of search volume. What do you do?
You can do 3 things:
- Test them in paid search - in our tests, 'how to' queries have generally failed to deliver ROI on their own – better on organic or content. For generic ones like ‘guttering’ or ‘credit cards’, the spend needs to be controlled strictly, however with clever optimisation which we’ll touch upon later, they can become highly lucrative.
- Assign alternative conversions - for example a video view or a resource download hidden by data capture. These are micro conversions that marketers usually ignore because they’re not the final effort. However, think about the potential of this data. Once you have someone’s email address, 1-1 communication becomes a possibility, and there are few channels that can beat email conversion rates. Someone subscribed to your youtube channel will turn to it again and again for help and advice, and when they need to buy something, you’re top of mind.
- Attribution modelling – or at least understanding and optimising with attribution in mind – ‘guttering’ is a query that on its own delivers weaker results than the target ROI. However, being so generic and high up the conversion funnel, it starts 2X more conversions that it actually wins. Usually, these sales couldn’t have happened if you didn’t enter the consideration set at an early stage. Unless maybe you’re the cheapest and you snatch the sale at the last moment – difficult to maintain this business strategy though.
So don’t ignore your generic queries, they can be more powerful than you think.
Know your context
The other type of intent & moment analysis is context – where, when and how (or what device).
Google search interest in "near me" has increased 34X since 2011 and nearly doubled since last year. The vast majority come from mobile, 80% in Q4 2014.
So what can you do to capitalise on these moments?
1. Ad formats
- Local inventory ads for shopping signals close by - have the item people need in stock, and at the best price/ offer
- Location extensions - offer directions and opening hrs information
- Shorter, snappier ads for mobile devices with call extensions or even call only ads that allow you to take immediate action
Advanced location targeting can be really powerful, if understood correctly. This is where understanding the search query, but also in combination with IP targeting, comes into play.
Do we want to target people who are in Bournemouth, looking for a ppc agency; or people who are in Bournemouth looking for Bournemouth ppc agency; or people who are in Portsmouth looking for Bournemouth ppc agency; or people in London looking for ppc agency?
You can see these queries are potentially very different, and could mean different outcomes to us –
- Do we want to limit ourselves to the Bournemouth area?
- Do we want to be national?
- Are we prepared to travel to Portsmouth or London for clients?
But it also has implications over CPCs and volume. PPC agency as a query will be more competitive than Bournemouth ppc agency, and competing on it in Bournemouth vs. London will be wildly different.
Who we compete with will also be different – so maybe our messaging should be customised. One search, many considerations to contend with.
But wait there's more...
- Smart bidding – all clicks aren’t equal, so all bidding shouldn’t be equal. If you’re still blanket bidding a term until your budget for the day is depleted, think again. You’re potentially missing the important moments later on, or you’re not visible enough for the critical moments (or at all), you may be wasting money on non-valuable moments. Paid search is becoming more about maximising share of intent rather than having the highest share of voice for every single query in the market. You need to make sure you are there for the queries that matter, at the moments that matter, and outweigh your key competitors’ share.
- Location bid adjustment – bid higher or lower by different areas (you can go quite granular in larger cities) depending on performance/closeness to your physical presence
- Combine with time of day and day of week bidding - I’ve recently been introduced to the world of football betting, a perfect example of context and moment. Super Sunday includes loads of games, but people place their bets on a Saturday. Pre game vs. in play bets. Google only allows 6 time-day modifiers, but scripts have been created that allow every hour of the day by day week modifiers, and you could even layer pay day (day of month) on top of that
- Device bid adjustment – especially if you are competing with the big boys who can afford 100% presence in #1 all day. For some of our accounts we have gone so granular that every ad group contains only one keyword, so we can modify the mobile bid for each query, and serve different ads by query and device.
Know thy searcher
We’ve talked a lot about the search intent, semantics and context but we haven’t talked about the searcher yet.
Do we have any signals in paid search that we can utilise to better manage our activity and go for the right audience?
Social and display, they started our audience off, and have loads of refinements in how granular you can search for the perfect user. But if the user comes to you, like in search, what do you know about them?
Past behaviour is the key here – key to adding an advanced level of intelligence to your bidding and campaigns, key to understanding more about those generic ambiguous queries, key to tapping into predictive search.
Simply because there will be no ‘generalised’ user intent
Instead, there will be personalised intent, your own web, tailored to you, with your permission.
Predictive search will know, based on your previous behaviour, what you want, maybe even before you know it!
Past behaviour in search is currently revolving around interaction with your site/app.
- Are you a new or returning visitor, are you an existing customer, and if so of what value?
- What pages have you visited? How many? Have you dropped off the shopping basket or an app form step 3? Have you visited 3 times before or just once? Have you visited in the last 7 days, or haven’t we seen you since October last year?
- Have you visited via mobile or desktop? Which channels & terms did you use to visit us?
You got it – these are all signals that give clues as to how a particular searcher is, what they’re looking for, and with what intent? Like in a dating scenario, providing the 'how serious are you about me' kind of signals.
- Google Analytics is your friend, but so is your own data.
- RLSAs (or Remarketing Lists for Search Ads) – something that has been around for a while, but only now truly pushed and important.
This is not remarketing in display – you are not followed with banners all around the web if you visited a site. This is when you search for a query that I bid on, and I know you have visited my site before, therefore I might want to give you privileged treatment. We’ll discuss that in a minute.
- Customer match – this is new. Uploading your own data in Adwords – similar to Facebook/Twitter custom audiences, this allows offline conversions to be taken into account
So how can you be smart with these features?
The most common use is to bid on more generic terms
Terms that are usually expensive but have a lot of volume and potential, but only for the most qualified users.
Generic terms can be expensive and have a lot of low commercial intent which you can’t afford. DSA (Dynamic Search Ads) campaigns will scrape your site and show dynamically created ads + the most relevant landing page to searches you might not have explored before but people still do.
Note: Tackling these areas without RLSA can be very dangerous, especially DSA – you could suddenly spend all your budget in a few hours, and appear on ‘adido login’, ‘how do I park at adido’ or ‘train times from London to adido’ – things like these, when you care about new customer acquisition (and even if not) are massive spend wasters.
Qualify the searchers by previous behaviour, screen them if you like, and only bid on the ones that seem serious; bid higher based on how serious they seem. You don’t have the money to buy ads all the time on ‘credit cards’ like the big boys, just buy them when you feel the connection is right and you could get a customer in.
Competitor name bidding can sometimes get great results if it’s done strategically.
Using RLSAs you can create a campaign where you bid on your competitors’ brand names, but the ads only show when the user searching is already on your audience list of users who have previously visited your site.
This means you know your brand is possibly in the running for their purchase decision and you can ensure you stay in their mind when they are comparison shopping with other brands, while not wasting ad spend on every user who might be searching for your competitors but isn’t considering your brand.
Obviously you can do competitor bidding without this, but depending on the competitor this can be expensive (quality score penalties will make it even more costly) and there can be more search volume that you can handle.
Brand bidding to new users only
With RLSAs/customer match you could create a brand campaign and ad group that targets only users who have not previously visited your website, or who have never bought from you.
This is ideal if your campaign is focused particularly on driving new visitors, as you can isolate the brand spend to just new users. This is if you don’t believe the incremental value of brand, but still want to be there in case competitors attack you.
Exclude converted users from your campaign
For some businesses, a user is no longer valuable once they have completed the desired action on the website. Bidding on them won’t drive any additional value to you, so it’s wasted spend.
Increase/ decrease bids where users are of particular value to you
Tailor your bids where a potential searcher is more/less valuable to you.
This is where analysing the quality of a user really comes into play. Have they bought before, how valuable are they and are they likely to convert again?
If they’ve never converted, do the signals they left on site show intention or just initial browsing? Bidding more or less in real time gives you access to the right audience in the right moment.
Tailor messages & landing pages to upsell, cross sell or re-engage
With these features, you can either apply segments to your campaigns, or you can create new campaigns that only target specific users lists – the latter way also allows you to customise the messaging and landing pages.
- Up sell, cross sell, or just re-engage searchers and remind them of you.
- You know they already bought a bike, do you try to sell them a bike rack, some lights and a better saddle as well?
- Or you know they bought car insurance in May last year and renewal time is round the corner?
- Were they close to purchasing but didn’t – do you offer a discount on that bike whose page they visited three times last month?
The impact on bottom line from RLSA is significant, especially in markets where multiple browsing and longer conversion journeys are the norm (increasing for most industries)
- CPAs can be 3X lower on previous visitors
- List size will be impacted by the level of segmentation
- Map your site journeys first and identify the best potential cohorts
Beyond Google and bing
What we’ve talked about so far is generally achievable through Adwords and Google Analytics. However, there is a lot of data out there that you can use.
Adwords scripts as well as certain 3rd party companies like TVTY (which also work on plugging data into scripts and making automated decisions) is the next level of data integration and smart bid layering so you can optimise for the right moment, serve the right message and in the right context – and squeeze some extra intent out of those queries.
TV data is particularly important because of the dual screening phenomenon. Millward Brown previously found that 32% of time on devices is spent dual screening while watching TV, with nearly half (48%) doing so during ad breaks
Weather conditions can really affect a retailer like Argos – in fact last year they won a search award for this specific strategy.
You can imagine how important live football matches, and the TV ads of big boys like Bet365, Ladbrokes, Will Hill and BetFair can impact a bidding strategy for a gambling company, especially with in play markets available
Inventory changes can help you automatically stop your ads when something is out of stock, so you don’t waste the click, or customising your message with a sense of urgency and countdown
The possibilities are endless, it’s just about taking the plunge into understanding scripts and utilising them to the moments that matter most during your audiences’ journey.