(not provided) – The Good, The Bad & The Ugly

This week’s #EcomChat was on “(not provided)”.

As ever, it was a one-hour live chat session on Twitter. There were 3 questions designed to get to the heart of what people are doing in light of the demise of Google organic keyword data. Here they were:

1. How have you changed your approach to SEO since the dawn of (not provided) and how has it impacted your business?
2. How is SEO likely to evolve with 100% (not provided) keyword data?
3. What should ecommerce teams be doing to plan for a keyword blind future & what tools/techniques should they be using?

Below is the full writeup by @jacksaville1:

There were three recurring topics throughout the hour.  These were:

• The rationale from Google for implementing (not provided) data.
• The role of keywords in light of (not provided) data.
• Practical consequences of (not provided), such as the re-education of clients.

There were suggestions that the rationale from Google for implementing (not provided) data was not about privacy but about their own financial benefit.

• @therustybear: @JamesGurd – amazingly its moved focus away from pure keywords to the content on the pages … its like they did it on purpose
• @JamesGurd: @therustybear and made using adwords even more beneficial, who’d have thought?
• @Magique83: @ReferralCandy @seoeditors @AndrewGirdwood It’s difficult to see past the ecomonic benefit to Google.
• @AndrewGirdwood: @Magique83 @ReferralCandy @seoeditors I don’t buy the privacy line. I suspect it’ll be part of the alternative to cookie progams

So the obvious next issue is the role of keywords in light of the (not provided) data.

• @dmandalia: @JamesGurd @carmenmardiros maybe not visual design, but content design based on KWs is important
• @danbarker:@dmandalia fully agreed. for the most part, searchers are still searching by keyword. @JamesGurd @carmenmardiros
• @danbarker:@dmandalia the hierarchy of most sites is based round keywords, which is how must users think about sites.
• @danbarker:@dmandalia card sorting, IA, etc, most of those old UX tools are essentially keyword-based. @JamesGurd @carmenmardiros
• @JamesGurd:@danbarker @carmenmardiros Where it did really help is with keywords that revealed searc h intent such as “how does….”
• @JamesGurd :@danbarker @carmenmardiros but keyword searches like “blue jeans” don’t tell you anything really useful/actionable
• @danbarker :@JamesGurd @carmenmardiros but if you have a ‘jeans’ page & see ‘blue/red/black’ searches, suggests trying another hierarchy level
  

It was then suggested that other methods of measurement would seize the reigns.

• @carmenmardiros: I think behavioural segmentation will become more important in analytics. Using onsite patterns of behaviour to determine intent
• @AndrewGirdwood: Clients still (rightly) care about keywords. We’ve had to change from soley analytics to a hybrid with Google’s WMC
• @Magique83: @JamesGurd My approach evolved and hasn’t been keyword focussed for a few years now. Focus is on global gains from all sources.
• SEONorth: @JamesGurd Relegated keywords in strategies. They’re researched for strategic input and placed in titles etc but that’s about it.
• @JamesGurd: @SEONorth Yep makes sense. Other moves like over optimised anchor text also point to keyword level less important than page
• @AndrewGirdwood: @JamesGurd Like many, we’ve been urging clients to care less about rankings for a while (they’re all relative, these days).
• @SEONorth: Do people on brief offsite copy (of any kind) to include keywords, out of interest?
• @JamesGurd:@SEONorth I still think it’s important to research keyword trends and inform off-site copy, don’t see why you wouldn’t
• @aattias:@SEONorth used to but not any  more   .
• @MontseCano: @JamesGurd @SEONorth That’s what I’m hoping to do before the end of the month. Using customised reports to try and get round it.
• @AndrewGirdwood: Clients should care about sales, leads & brand boosts they can measure. The %100 NP challenge is mapping those to actions/tactics
• @Magique831h
• @JamesGurd I analyse and report at different levels depending on the source. But yes I will look at page level for Organic traffic
• @Magique83: Grouping page levels together e.g. product, category, sub cat, summary, dept etc gives a better picture than keywords IMO
• @MontseCano: @JamesGurd Agreed that we should check various things and not just focus on one thing.
• @SEONorth: @JamesGurd @aattias I research keyword trends for topics but consider that enough. No need to request writers use keywords.
• @seoeditors: @Magique83 using landing pages at different levels is ‘scaling’ – useful insights are possible but there are more clues to use
• @Magique83: @seoeditors There are indeed but ultimately they are becoming just clues. There is no full picture anymore

There was debate over the usefulness of a landing page approach.

• @JamesGurd: A1) I think page level performance supersedes keyword level focus – are landing pages working well?
• @carmenmardiros :@JamesGurd Agreed, but asessing audience mix is crucial to optimising landing pages. Who are we optimising that page for?
• @JamesGurd: @carmenmardiros designing a LP based on a keyword isn’t and never has been a sensible approach IMO given the data i’ve seen
• @JamesGurd :@danbarker@carmenmardiros but keyword searches like “blue jeans” don’t tell you anything really useful/actionable
• @AndrewBaker72 :@JamesGurd @danbarker @carmenmardiros except to ensure you have a cat/LP of blue jeans (with granular faceted nav options) ;-)) #ecomchat
• @dmandalia :Small tip: map internal search terms (if you have it) to LPs to hint at external KWs searched and to show LP intent #ecomchat

Could data from the landing pages be used to supplement webmaster tools data?

• @JamesGurd: @JamesGurd And is anyone making more use of the WMT keyword data even with its limitations?
• @AndrewGirdwood: @JamesGurd Yes, WMT is a central now. But it’s only an estimate. We try and show the error delta and suppliment w/ extra data
• @ismepete: @JamesGurd we’re looking at traffic to landing pages a lot in tandem with GWT data. The data is fuzzy compared with keyword data

Should the data be viewed in tandem with the customer journey?

• @carmenmardiros : Keyword analysis worked best when you group into concept buckets & map to stage of customer journey. Forget single keywords.
• @danbarker :@carmenmardiros go to the level of granularity that suits audience intent, not to the Nth degree just because you can. #ecomchat

Is it worth using any methods to try and get around the (not provided) situation.

Hacks & Toolbars:

• @danbarker : that’s one of the reasons I like the ‘not provided hack’ landing pages show up as KW in attribution reports.

• … – if you use that NP hack, you get the attribution one as standard.

• @danbarker :@AndrewGirdwood remember cookie dropping toolbars? wouldn’t be surprised if people launch keyword watching toolbars.
• @AndrewGirdwood :@danbarker Raises questions about whether agencies should buy keyword data… ethics.
• @reprovided : @AndrewGirdwood @danbarker Also potentially sampling bias – less savvy users have more toolbars, more savvy users may use diff kws #ecomchat

[DB: As a sidenote here, ComScore seem to essentially be doing this – gathering keyword data via their toolbar]

The final topic that was discussed was the practical consequences of the implementation of (not provided) data.

The first issue was that the change may have made operating in an agency more difficult.

• @Magique83:When working in house you should have enough understanding of your customers to work without keywords. Agency side is difficult #EcomChat
• @AndrewGirdwood:@ACChaudre @Magique83 The largest agencies do get to pull on other data sources though and vertical trends
• @Magique83:When working in house you should have enough understanding of your customers to work without keywords. Agency side is difficult
• @SEONorth:@Magique83 Interesting point. Would you add competitor knowledge as in house advantage/requirement over agency too?
• @Magique83:@SEONorth To some extent yes. Some agencies have a better idea of what competitors are doing on occasion though #EcomChat

Another large issue has been the re-education of clients in light of (not provided)

• @priteshpatel9 :@danbarker – i’m struggling w/ clients not grasping (change mindset) from keywords to page performance. Just not tangible enough
• @dmandalia : @priteshpatel9 @danbarker agree – most clients just about understanding the basics of SEO which revolves around KWs/rank.The game’s changed

• @priteshpatel9 : @dmandalia @danbarker – so what do you do? Start the process off with a 2-3 short training courses? Re-education is a big thing.

• @dmandalia  :@priteshpatel9 @danbarker It’s something I’m tackling right now – re-writing our internal SEO guide followed by webinars

• @JamesGurd : @priteshpatel9 What are you doing Pritesh with your Clients? Is B2B world taking this seriously or not on the radar?

• @priteshpatel9 :@JamesGurd – clients paid SEO’s to tell them which keywords bring in traffic et al. Now we can’t tell them. Now what? Charge less?

• @pferron : @AndrewGirdwood @JamesGurd @priteshpatel9 it’s a really good point non digital native MD’s already regard seo cynically harder now

• @dmandalia : @MontseCano @priteshpatel9 @JamesGurd In-house or agency side SEO has become much harder – but for everyone. Level playing field

• @ChinaShopBull :@AndrewGirdwood Are you suggesting cutomers focus more on outcomes/leads etc rather than SERP ranking?
• @AndrewGirdwood :@ChinaShopBull Yup! Focus on results and the actions necessary. Be aware there are steps in between, though.
  

One way of re-educating clients is by offering other data to help explain rankings.

• @SEONorth :@danbarker @priteshpatel9 We’re reporting on many more metrics than just rankings & SE traffic. Contextualises value of rankings.
• @JamesGurd :@SEONorth How do you correlate rankings with other metrics like per visit value/conversion etc? #EcomChat

• @SEONorth :@JamesGurd At page level for ranking term. But our also work refers traffic, generates social discussion, brand search volume…

 

Finally James Gurd moved the chat onto the topic of a keyword blind future

• @SEONorth :The sooner we’re used to (not provided) the better. Once semantic search and its user habits become the norm, kwds will be minor.
• @danbarker :@SEONorth not sure I totally buy. semantic search mostly uses inferred search terms. if G were 100% user centric they’d pass that

• @seonorth: @danbarker Sure but even if they do, the variance of terms would be huge. Much more long tail and head terms less representative.
• @dmandalia  : though PPC can help, many sites rely on SEO KW data gained over time for traffic/conv which being NP’d is shifting business plans
  
• @deedeelea : @JamesGurd referrals? Eg related blog content
• @AndrewGirdwood :@JamesGurd it’s not a keyword blind future. We have to build off WMC estimates. The future is algos to help map keyword knowledge
• @AndrewBaker72 : A3 If you use PPC you won’t be KW blind, so if you really care for KWs include PPC in your strategy
• @AndrewGirdwood : @AndrewBaker72 Combine PPC data, with WMC with other search engines – algorithm some sense into it and volia great service
• @AndrewBaker72 :@AndrewGirdwood yes, we’ve used this type of approach ever since NP arrived on the scene…
• @Magique83 :Q3) Use search volume to direct onsite SEO, industry knowledge to direct content creation, PPC for specific keyword targeting
• @SEONorth :@danbarker Focus on trends, topics, pages and visitors. It’s about themes not specific phrases.
• @danbarker :GA: There’s a good argument for storing the keywords you’re aiming for with pages, and tracking those as an event/custom dimension
• @JamesGurd :A3) Use the LP hack to find out which are key LPs that have np traffic – then work back to piece together then keyword puzzle
• @dmandalia :Q3) Focus on LPs and conversion/revenue from SEO together with rank tracking (which is hanging on by a thread!) + PPC test/learn
• @JamesGurd :A3) Seek out keyword data from other search engines – see what you can learn, then work out how to apply to Google
• @reprovided :Q3) Map LP traffic to KW rankings, use kws from history, PPC, GWT + Search suggest. Use as inspiration for strategy not hard facts
• @carmenmardiros :Travel example: do visitors start availability search by country (head) or country & city (specific)?
• @reprovided :@carmenmardiros Often ‘Deals’ IME 🙂 But depends on destination. EU generally city, long haul country
• @carmenmardiros :@JamesGurd My approach:what does this interaction/set of interactions tell me about customer segment & intent & shopping attitudes

In conclusion, there was a divergence of opinions over whether (not provided) should be:

• Treated as an obstacle that could be circumvented with hacks
• Treated as something to ignore, and to instead complement any remaining keyword data with other types of data.

But factors such as the re-education of clients seemed unavoidable.

Thanks again to @jacksaville1 for the writeup.

Join us at 1pm UK time on Monday, and do feel free to share this post either on Twitter or via email with anyone you think would be interested.

Dan & James.

ps. If you’re wondering “what is this all about?” there is an about page to explain everything.