PPCChat | Automation – The Good, The Bad & The Frustrating | 29 January, 2019

Posted by & filed under PPCChat.

Hosted by Julie  Bacchini, this week’s PPCChat session discussed about automation, where it has worked and not worked, the areas of improvement, what automation is being used for and more.

PPCChat

Here is the screencap of the discussion

 

Q1: What, if anything, are you currently using automation for in your PPC work?

 

We primarily use automation for bidding, alerts, enabling/pausing entities based on budgets, assistance in IDing key focus points of an account for us to investigate. – 

currently – Quality score tracking, doing negative management for shopping campaign, negative conflict and finding account anomalies. our reporting is also automated with web queries and plug into Data Studio –

We use it for bidding strategies in ads, we have scripts that run for bids, scripts for other things as well, and then we use automation to identify errors like ads not serving or landing pages down. I guess the majority of reporting as well, we write custom parts of our reports, but the metrics and calculations are automated. My favourite is a script that changes bids for a finance client based on what currencies are trading at – I geek out – 

Budget control, time-based bid increases, ML-based bid management – 

All our reporting is automated. We use auto bidding when it makes sense for a client. We also have a few scrips running in the background of ecom accounts to look at negative keywords, competitors and other bits around each client – 

I only use it for littler stuff – alerts if something is amiss and some bidding. Unless you count all the stuff that FB automatically does for you… – 

Scripts for bidding, problem alerts, budget management, and search query mining. Recently started using scripts in combo with Google sheets for bid changes across multiple accounts at a time – 

Bidding, some scripts, budget. – 

Scripts and I use automated reporting so I can build out the reports and schedule. – 

Bids and budgets (both platform & 3rd party tools). The data gathering part of reporting. – 

We currently only use it for smaller things. We’d love to hear some success stories with bidding strategies. We had a bad experience on a couple accounts and have never went back down that road. Has it improved? – 

We use scripts to identify accounts that need a bit of help, find links that may be broken, budget control etc. Also, some bidding strategies but it depends on the client. – 

multiple bid strategies via our bid management platform, and some UI strategies – 

Primarily using automated bid rules, a few scripts, Smart Bidding (where’s it’s been tested and won), and @Supermetrics for automated reporting. – 

We use automation for keeping track of quality scores, bid optimisation where it fits, to keep track of which ads/ad copy is resonating with the audience and so on. – 

We currently automate our bidding strategies, and that’s about it. – 

 

 

Q2: Where and how is automation working really well for you?

 

Reporting 100% with @Supermetrics is working as is scripts. Bidding can be hit or miss with clients but that is expected. – 

definitely our reporting. It is pretty much an auto update system. Tables, graphs & every other aspect. – 

A “non-actionable” answer: working the best when it ASSISTS us in our mngmt. So when it solves an easily repeatable task to allow us to focus better on strategy and bigger pic management. I.e., assisting us with automatically enabling/pausing ads for testing, etc – 

we have a lot of time of day, conditional asset deployment, saves us all kinds of time – 

I mean, it keeps Google happy if we use their automated bidding (and after testing some of it has actually worked quite well). With scripts it allows us to manage large inventories or complex changes easily – 

I really like the script that identifies the accounts that are in a bit of trouble. It’s nice to know where to zero in vs. spending the time trying to figure it out on my own. – 

I’ve got a script that stops Google from overspending my daily budgets by 2X. It’s so nice to have. – 

Automation frees up a lot of the time historically spent micro-managing. Micro-management does produce long term gains, but if that’s all you spend time on then you’re not growing as much as you could – 

Bidding and Reporting. Allows my team to focus on strategy and content. Also allows for deeper analysis that is not fully automated. – 

Pretty much all of my FB campaigns are automated. – 

Basically an echo of A1 (we use it where it’s working well). More generally, automated bidding for search works well when there is good data and/or trends that can be identified. FB automation (although there isn’t much alternative) is also pretty reliable for me. – 

We find that automation works very well to help keep up to date with all the account, especially when reporting. – 

As mentioned in A1, the automated bidding is working well and one less thing to worry about with managing a client with 6 accounts. – 

 

 

Q3: Where is automation NOT working really well for you? Why do you think it is not working as well as it could/should? (The frustrations question is coming next!)

 

It doesn’t work well where you need nuanced changes to happen at an overview/strategy/long-term level or in industries where what is typically “good performance” is not good and actually jeopardizes the client relationship. – 

Bidding of course. a lot of one size fits all data doesn’t work across all accounts. Like smart goals in GA….nice idea in theory. Not reality. – 

I’m still unconvinced about pure automation in bidding (yet). My reasoning being, the data we use is the issue… not the algorithms. If we can’t give accurate attributed data to the machine, then we can’t expect it to automate well. And attribution is still a mess. – 

this is a big frustration for sure –  

The automated bidding has not been doing better than manual yet within GA. I keep testing it and the Google reps keep pushing it, but still just not even close. – 

it’s not working well with limited volume accounts. Algos need data to perform well. – 

Actually pretty happy with it. Most of the repetitive tasks I’ve been able to automate, even negative fencing analysis. Don’t use Google for automated bidding, use another tool (#noplug) and have been happy with results. – 

Doesn’t work well when traffic/engagement fluctuates frequently or when traffic/conversion volume is low. Insufficient or low quality data is basically my #1 nemesis when it comes to automation – 

Bid management. We’ve split tested autobid (tCPA, specifically) vs. manual bids over and over w/ same results each time: manual outperforms, and it’s generally not close. – 

Automation doesn’t always do things the I would have done them. It’s presented as hands-off, but you have to reinvest some time savings in oversight and skepticism. (and in some cases, just accept that there are multiple ways to achieve your goals) – 

despite the push by Google, automation in grants is a bit worthless there. In addition, automation in GDN gets me traffic, but throws quality out the window –

Echoing all the Bidding responses. Learned in 2018 that One PPCers Automated Bidding Success might not translate in to success for you. Must test and adjust. Account structure can affect automated bidding as well. – 

We tried some automated Facebook campaigns last year but were frustrated with the day/time-parting set-up – 

 

 

Q4: What is your biggest frustration with automation in PPC?

 

Vendors who take a one size fits all approach or have super bias data being used to figure bits out. – 

the in ability to look at bigger/macro factors (ie being in the news) and adjusting in real time for it – 

PPCers not taking advantage of it. 🙂 On the flip side, I don’t like ad platforms forcing it on us. – 

Low volume situations, it doesn’t work or make sense – 

The “we know best” attitude shown by platforms when it comes to automation or applying machine learning. If you don’t have all the data I do, then you cannot possibly “know best”. Smaller gripe – but @GoogleAds “rotate evenly” option DOES NO SUCH THING. – 

Automation tends to be a “My way or the highway” situation more often than not. Or “All or nothing”. Either way, campaigns, AGs, keywords, locations, etc, are all nuanced. Automation needs to be adjusted for each influence, not all in batched form. –

Honestly, probably Google pushing the AI like it is already the best thing ever. I believe in testing constantly and I do believe it will get there, but look at the data what what you want to see – 

Bid automation (tCPA) has consistently under-performed vs. manual. We’ve tested over and over, same results so far. Would love to save time for other tasks by letting the system automate bids, but the data so far says no. – 

For the system I use, I would love some alerts for anomaly detection. Have anomaly detection already, but not an alert that goes out via email. Have to log in find it manually. – 

Lack of transparency into what’s working and what isn’t. (FB in particular is disappointing here. I LOVE that xyz combination of audiences/targeting/creative works well, but it’s nearly impossible to know WHY it’s working) –

Apart from our A3 answer, not much comes to mind currently. – 

 

 

Q5: In light of Google Ads new initiative to start doing things in accounts “behind the scenes” automatically (unless you opt out) – what would you like them to tell them as PPC professionals?

 

What changes Google is making and why would be wonderful. We recently spoke with a Google Ads rep for one of our PPC clients and discussed changes to bid strategy and attribution, which we’ve moved to enact. That was fine. Google doing it for us – nope. – 

Don’t touch my accounts – 

STAY IN YOUR LANE, GOOGLE – 

The biggest part of this latest communication is that Google will offer adjustments if their recommendations don’t improve results. How do you compete with that. Google is cutting out the expense to hire an expert and providing the expertise themselves with a money back guarantee. That means more money for Google. Just my opinion on what is coming. – 

First: Hands Off! 2nd: (Since #1 is unlikely) Provide controls for each area for accounts to opt-in and opt-out without it being an all-or-nothing scenario. – 

Speaking of, Google just called me and its not a member of my agency team – 

“Please opt us out.” – 

NO NO NO NO NO…just No! Consult me before making any change & actually allowing me to say no to changes i don’t want. – 

I assert that “best practices” are normally distributed. Just because something “tends” to work across 1000s+ of accounts doesn’t mean it should be universally applied. TALK TO THE ACCOUNT OWNER to understand their goals and past learnings first. – 

Forgot the hashtag #ppcchat on the last tweet I posted I’m so aggravated by this idea. – 

Hard to word, but stay away from my accounts. Also they are showing they do not “really” care about agency partners, especially if they can take things over and make more $$ (not for the client, but for Google). – 

This should absolutely be an OPT-IN choice. You’re talking about changing account settings (anything but budget as I understand it) and also completely absolving yourselves of any resulting negative outcomes. Can’t have it both ways. – 

hat your team knows more than people who know nothing about search, but for the most part, your team really doesn’t know what they are doing other than to encourage spend  – 

agencies aren’t the enemy, they’re your biggest advocate. agencies work to make clients Happy. Full stop. You can freaking live without the little bit of margin that we do make, and it likely makes you better too. and until you can tell a dog from a muffin, and the different intent between chocolate milk and milk chocolate, you might be rushing the ai thing a bit – 

How many people approved the opt-out angle? Because every one of them needs to be reprimanded (not fired). Automatically opting in a user to such an intrusive “service” is a terrible customer experience. –

 

 

Q6: What areas to do you wish that platforms would do better with automation? What would be most helpful to you?

 

make it less intimidating. It’s either black box, or code required. trust is that we know our brands and make the tools reflect that see blockly for an example of easier coding – 

Be more transparent about how the automation is working. Automation’s goal should not be to obfuscate but to illuminate. – 

I know there is always a secret sauce, but if they could be more transparent somehow. The agency community could even help with making it better. If AI bidding gets better we can do more for our clients with strategy and creative. – 

Bid management, for sure. Majority of split testing indicates the systems favor driving up the price point to capture more volume (and spend more)….even if you layer in efficiency targets. Same issue for publishers’ platforms & 3rd parties alike. –

Showing what the platform has been learning from the automation and providing some guidance could be useful. – 

Haven’t fully fleshed out this thought, but some kind of “quality score” for conversion data would be cool. Not sure how that would play out, but some kind of warning if an advertiser tries to test automation but your data isn’t good. more hybrid bidding options (like Enhanced CPC, but you can choose what Smart Bidding strategy/objective the automated 50% is using). Basically, an easy way to do this that doesn’t require ongoing Experiments. – 

Transparency in how automation actually works would be great. I’d love to learn more and maybe be help to improve it. – 

lay out all the scenarios that could happen. Like yes, you’ll show in #1, but CPC will rise and click volume will reduce – 

 

 

Q7: What areas to do you wish that platforms would stay away from when it comes to automation?

 

the subtle implying of things can be solely automated and not need human oversight –

ad writing of any form. the best copy isn’t apparent. AI would never invent “just do it” – 

Can “rotate evenly” just rotate evenly in perpetuity, please? I need even splits to justify test results w/ stakeholders – 

Anything related to creative and copy generation! Please do not scrape images (often old) or auto-generate text that is potentially out of compliance or false. Also wanted to toss in automation of assigning search partner traffic…. not all search partners are the same, nor do they showcase the same intent or position in the customer journey. – 

 

PPCChat Participants:

 

Related Links:

Stop wasted ad spend with Karooya

Stop the wasted ad spend. Get more conversions from the same ad budget.

Our customers save over $16 Million per year on Google and Amazon Ads.

Start free trial Schedule a demo

Leave a Reply

Your email address will not be published.