Taking Google Shopping to the Next Level

wishlistOver the year I’ve witnessed a lot of frustration about the lack of keywords in shopping campaigns. Oftentimes I wanted to to say, “Hey, actually…”. But it was too soon. I had to keep the secret.

A few months back, Kirk Williams published a 2014 Google Shopping Wishlist with keyword targeting being the number one wish. As it turns out, with Christmas shopping in full swing, Google still hasn’t fulfilled this wish and has given no indication that they ever will.

But now the time to reveal the secret has come. Let me show you how it’s done.

Marketing FestivalIn November I had the incredible honor to speak at Marketing Festival. This international conference with over a thousand passionate online marketers had only one track and only let a few hand picked experts speak. It was the perfect venue to reveal not only the tactic that enables us to bid on individual keywords, but also a complete strategy that utilizes this tactic on a larger scale.

All of the presentations were filmed in awesome quality. They sell them all for 40 Euros (about 50 US Dollars), but I was able to convince them to let me show mine here. So – enjoy this handsome German explaining how to take Google Shopping to the next level.

Now I’d ask for your comments but if you want to get to work right away, I understand 😉

After countless of comments, discussions and consulting hours about this strategy, I’ve written a guide about dealing with common pitfalls: Troubleshooting the Google Shopping Search Query Segmentation

About Martin Roettgerding
Martin Roettgerding is the head of SEM at SEO/SEM agency Bloofusion Germany. On Twitter he goes by the name @bloomarty, and you can find him regularly on #ppcchat.

  • Jordan McClements

    Marty, thanks very much for this.

    I have noticed this myself with Google Shopping but was not 100% sure of he best way to get around the problem.

    It’s not often I watch a presentation and and think that it was time very well spent! 🙂

  • Calin Sandici

    Thank you for sharing this presentation, Martin, it sure is a smart way of controlling shopping campaigns.

    Being somewhat OCD about the search queries that shopping campaigns attract, I used a rather different approach.

    I try to keep search terms product-oriented as much as possible and for the more general terms (asics / running shoes) I either use them as negatives and leave them be or use them as negatives in the shopping campaign and as keywords for text ads that lead to brand / product category pages as they are a) more relevant for the query and b) showcase a larger selection which the visitor might not see for websites that do not have state of the art “related products” or “people who bought this also looked at” pages.

    I’m also using a bit of traffic shaping methods, meaning that for a search term such as “running shoes” I add it as a negative in all ad groups except 1-2-3 that are best sellers in the running shoes category.

    But all this means a rather complicated structure and quite a bit of work. Your strategy, on the other hand, looks like it would take less time to deploy and maintain.

  • David Bailey

    Hi Martin,

    Really enjoyed the video. Unfortunately I don’t think that approach would work for us as brand names or product info aren’t frequently searched for in relation to our products. Have you done any work on similar businesses to us in terms of campaign structure?



    • Martin Röttgerding

      Hi David,
      We mainly use this structure for brand and product related keywords, but other segmentations can be done the same way. The key is finding the right patterns. One such pattern could be technical terms: Maybe people who know a bit about a product don’t use layman terms anymore. Technical terms could then be worth more. Or maybe there are other patterns for more valuable terms.. people looking for specific quantities, sizes, or colors are often further down the funnel.
      Or you could still segment by brand. Even if it’s not often that people look for it – when they do eventually you could make sure you’re there.

      • Yeah, I’ve found success with this even in non-grouped terms. Literally, identifying those converting terms and pulling them into another campaign, even if no rhyme or reason to them other than that they are profitable terms. As time goes on, continue to update it with negatives for traffic funneling.

        Shopping – Widgets – Low-Converting – US
        Priority: High
        Lower Bids.

        Shopping – Widgets – High-Converting – US
        Priority: Medium
        Higher Bids.

  • I’ve been implementing and chewing and implementing and chewing ever since I saw this. One test of a strong strategy is time. The more time I spend on this strategy and the more data that comes from it, the more I realize how truly brilliant it is. Well done, Martin. Someday I want to shake your hand.

  • Jordan McClements

    As an aside… I have noticed after using this method a for a fair while now that occasionally non brand searches will appear in my brand campaign even though it has a lower priority than the non brand campaign. It doesn’t happen very often but it is puzzling if the priority is meant to ensure that the brand campaign bids are only taken into account when negatives prevent the non brand campaign from being used..

    • You are using a shared budget on those 2 Brand/Non-Brand campaigns?

      • Jordan McClements

        aaarrgh – I can’t remember off top of my head. Would this make a difference?

        • It doesn’t make a big diff unless 1 hits a budget cap.

          If, let’s say your NB campaign hits a cap and drops off, then the whole strategy goes out the window because all of those non-brand queries shift into the brand/SKU campaigns, and those are usually higher bids so CPCs go high on poor converting, general terms.

          • Also another suggestion is if you’re using every variant of the negative (misspellings, plural, tense, etc).

          • Jordan McClements

            No, that’s not it either.

          • Jordan McClements

            Ah, yes I see what you mean. No, I use shared shopping budgets.

          • Another idea here is If/When you add keywords as negatives into the NB campaign, you also have to add those now into the BR campaign (every time!) or you will end up just pushing them into the BR.

            As a sidenote, I don’t think you’re going crazy. I haven’t done any exhaustive research, but I believe I’m seeing the same thing in my accounts. I have to keep a close eye on SQR it seems, especially in my SKU campaigns.

          • Martin Röttgerding

            Don’t bother finding a good reason for this behavior – there is none. As mentioned on Twitter, we’ve been seeing this as well. It has to do with budget constraints. If you narrow down the dates when this phenomenon occured, you’ll should find that you came close to your budget cap on those days. We’ve been on this for some time now…

            I’ll share details in a blog post soon, maybe over the weekend.

          • Jordan McClements

            Though you may have hit on something. Maybe the shared budget is not working 100% as it should?

          • Do you have it set to accelerated delivery? Perhaps this is somehow part of the equation. Perhaps when not set to accelerated, even with a shared budget, Google determines to show certain campaigns for certain queries. This could explain why sometimes those NB terms still slip through.

  • Grzegorz BogdaÅ„ski

    Very informative, thank you!

  • aidin

    Great presentation.

    What match type should be used for negative keywords? I’m assuming phrase.


    • Martin Röttgerding

      Hey aidin,
      Phrase or broad will do.

  • Snows White

    Hey Martin. That was quite enlightening.

    So basically, we can make separate three campaigns for each category or subcategory with a shared budget using the keywords.

    We dont have to use custom lables as such? If yes, then how?

    In category ELECTRICAL, I have LED Lights which are further segmented into LED bulbs, LED Strp Lights, LED Flood Lights

    So my structure would be:

    1. LED Bulbs – Product

    2. Buy Syska LED Bulbs Online-Brand

    3. Rest

    Priority low, med & high in that order.

    Where and how do I use custom labels?

    Thanks in advance

  • Mike

    I really enjoyed your video.

    I came up with a similar strategy early last year right around when shopping campaigns were released. I was managing PLAs for a watch company. Similarly I had the 3 separate campaigns and 3 different bidding levels, but I kind of had the reverse way of employing the negative keywords. I had the luxury of a good amount of history of search terms and metrics from the already running PLA campaigns.

    For the original campaign, I had designated that for model number searches. I added negative keywords for any search terms that have ever spent over $10 or 5 clicks or something like that, that did not contain a number. The 2nd campaign had lower bids and negative keywords from any search terms that did not contain brand names or specific names of the watch. The third campaign had the lowest bids with very few negative keywords.

    I didn’t mess with the priority levels as I would assume the bidding levels would control the hierarchy of the campaigns. I guess I also didn’t fully trust Google’s priority level settings. It’s actually good to know that works. All of these campaigns had basically unlimited budgets. If a model number search slipped into another campaign, who cares I thought. Obviously I had to constantly review search terms and keep adding in negative keywords to the first 2 campaigns.

    Although a little different in implementation i think it achieved pretty much the same results and the performance of the PLAs had improved dramatically.

    Curious to know what you think.

  • Chris


    Excellent video, fantastic.

    Had a question about bidding for each campaign. If for example (to keep it simple) the existing Shopping campaign has a $1.00 bid, what do you recommend for the bids for each of the campaigns you recommend: Brands, Products, Everything Else. Bid less than $1.00 for the Everything Else & Brands, or bump up the Products campaign to say $3.00 click, have the Brands campaign at less than that ($2.00) and the Everything Else campaign at for example $1.00?

    In other words is your base for bidding higher or lower the original bids you had before implementing your strategy, going higher than that, or is it taking the original bid as the highest and the other two campaigns bid lower? Thanks!

    • Martin Röttgerding

      Hi Chris,
      Your original shopping campaign is probably tuned to the average value of a click. Since its an average, the separate parts have to be both above and below.

      If you split it up into products, brands and rest, the value of product related queries will be above the average, brands will probably above as well, and the rest will be below. Your new bids could be $2, $1.50, and $0.50, for example.

      What preciselyI the right values will be cannot tell you as it depends on your individual case. You could just split up your traffic, use the old bids (the $1) in the duplicated campaigns and then adjust them according to their results.

      • Chris

        Excellent, thank you!

  • Nice share..

    Thanks for sharing such an informative post with us. Marketing trends are changing very soon and its important to go with the flow in order to taste success.

    Your ideas are great and the way you have express them is quite fascinating
    Keep up the good work..


  • Stu

    We’ve set up some test campaigns with the three tiers but finding that the product campaign is receiving no traffic. The campaign works as we switched it to medium for a few days and it stole all the traffic from the brand tier, so we’re left scratching our heads.

    • Martin Röttgerding

      Hi Stu,

      I’ve heard about this twice recently. In both cases there was no good reason, but when they switch their negatives to broad instead of phrase match, it worked.

      • Stu

        Thanks Martin, we’ll give this a go and let you know if it works

  • Julian

    Hi Martin, I’ve been using your strategy since I discovered your post a year ago. Recently I read an article that it’s related to our campaigns performance: http://www.ppchero.com/googles-not-selling-your-products-like-it-used-to/

    I would appreciate if you could share your thoughts and if there’s is any update to improve our campaign’s results.

    Thanks a lot for your great job.

    • Martin Röttgerding

      Hi Julian,
      Sorry for not getting back to you.

      I tried to reproduce those results at a small scale and I found that I could neither prove nor disprove the theory. I cannot say if a product title closer to the one displayed makes a product rank higher.

      I doubt it, though. The products are associated through GTIN’s, which should make them all equally relevant. Ranking should be determined by bids and the visible parts that affect click-thorugh rate: price, shipping, and the merchant’s name.

      As far as this strategy goes, the added price transparency makes product queries potentially less valuable if your price isn’t the lowest. Still, segmented campaigns are your best chance to spot such a development that would only affect product queries. This way you can react and lower your bids if products no longer perform for product specific queries – just like you normally would.

  • Orod Hatamifar

    Hi Martin!
    Thanks for your helpful informations. I have set up three campaigns exactly like your instructions, products (Low Prio), Brands (Medium Prio) and Rest (High Prio). I added negative queries in Brands and Rest campaigns as a broad matching. The budgets are the same amount for all but I have zero traffic in Products campaign.

    • Martin Röttgerding

      Hi Orod,
      It might be a stupid question, but are people searching for those product keywords? In my experience, a products campaign only works for few advertisers – those who can actually isolate product specific terms.

      For example, if you sell laptops like the HP ProBook 470 G3 Notebook you could use “470 g3” as a negative to isolate terms for such a product. The whole product name probably won’t do you much good as it’ll only affect queries where people use all five words in “HP ProBook 470 G3 Notebook”.

      If that’s not it, I’d look for search terms that I’d think should be in the products campaign but end up elsewhere. For example, is there a search term like “470 g3 notebook” ending up in the wrong campaign? Another alternative would be so search on Google for a term that should go into the products campaign. Then check if your ads show…

      Does this help?

      • Orod Hatamifar

        Thanks for your answer Martin.

        Definitely not a stupid question, we are dealing with professional cleaning and hygiene products to private and enterprise customers in the north of Europe. Some of the brands are famous internationally with the a lot of daily queries but some brands are not yet popular to buy from internet among B2B community.
        To be sure that I explained you right about the negative terms I put some real example here for you:

        1) Rest Campaign – “High Prio” + “Low bid”
        Negative broad match terms:
        semperguard – ( A Brand name )
        517802 – Vinylhandske Semperguard Pudrad str M ( A Brand Item No & product name )

        2) Brands Campaign – “Medium Prio” + “High bid”
        Negative broad match term:
        517802 – Vinylhandske Semperguard Pudrad str M ( A Brand Item No & product name )

        3) Products Campaign – “Low Prio” + “Very high bid”
        Negative broad match term:
        No negative terms

        The products campaign ad shows with this term: ” 517802 – VinylHandske Semperguard Pudrad str M ” but as you say nobody would search this long term. I’m not sure if a term ends up in the wrong campaign. The fact that the product campaign doesn’t have any traffic is not so important as long as the other two campaign’s set up are right and ads show effectively for the most regarded search query.

        • Martin Röttgerding

          Hi Orod, sorry for taking so long to answer. Yes, that sounds like the negatives are just too specific to catch any real searches as probably no one will search using all of these words at the same time.

          My recommendation: Check your queries and try to find out how people are looking for specific products. Sincesuch queries are usually quite valuable, you should find some of them if you just check your converting queries – no need to sift through all of them. Then check if you can isolate them with negatives.

          The problem is usually the scale: If you need to isolate terms like “Semperguard Pudrad str M” and the only part you have available in a structured way is the brand name, you probably won’t be able to separate those term at scale. However, if the numbers (like “517802”) are what set those queries apart, it might be easy since you might already have them in a structured way (maybe as MPN’s in your data feed) or because you can easily extract them.

          Anyway, in many cases, a products campaign makes little sense. If you don’t find a solution to handle product terms at scale, you’re still in good company.

          • Travis Sapolin

            Hi Martin, I would love to hire bloofusion to run our shopping campaign and implement a keyword level bidding strategy with priority settings and brand/style/product negative keywords. Our website is called New York Glass (www.nyglass.com) and we retail eyewear. Most of our customer acquisition comes from google PLA’s. Current monthly spend is $6-10k. We know we can do better with this strategy. Are you free to discuss?

  • Carol S

    Thanks for this post and video Martin.

    I’ve been working to implement this for my account however I’m running into 0 impressions for some of the campaigns.

    I have 5 different brands of products. And thus tried this method with 5 groups of 3 campaigns (non-brand, brand, and product/SKU), all within the same account.

    Only one group of the campaigns has all 3 running impressions. The others have only 1 campaign (the non-brand one) receiving traffic, though minimal. The brand and product/SKU campaigns of these other 4 brands have all received 0 impressions over the last 2 weeks.

    I was wondering perhaps my overall set up is incorrect. Should I only have 3 campaigns total, instead of 3 x 5? Does having multiple campaigns with each priority setting cause issues for Google?

    Thanks for your thoughts on this.

  • Mike

    I’m confused on the Priority of the Campaigns. Why would you want the “everything else” campaign to get the highest priority??

    • Martin Röttgerding

      The everything else campaign needs to attract all search queries EXCEPT those that are blocked.

      Priorities are used for segmentation here. For this sake, priority is not to be confused with importance, preference, or value.

  • Joe Wilde

    Hi Martin, great presentation! I’m very new to digital marketing (I’ve been in my first job <5 weeks) so this looks really useful to help me make a great impression!

    My question is do you recommend setting up this structure for each individual product or can you do it by loosely grouped products? We have over 1600 different items in our store so setting it up for each individual product would be a lengthy process!

    For example, we sell Christmas decorations as well as other home and garden products, and some of our biggest sellers are Christmas lights. Premier is the brand that performs best and under their brand name they have a multitude of different products like Supabrights, Treelights etc. Is it necessary to set up a different product campaign for each individual product or would it work simply by applying negative keywords for each of the products to the Branded and Rest campaigns and then having one grouped product campaign?

    • Martin Röttgerding

      Hey Joe,
      At the risk of being to late with my answer: There is no need to set this up multiple times. Basically you structure your products inside a (one) campaign and then duplicate this campaign in the way described to capture different queries.

      The way you describe in your example is exactly how it should be. Let’s say you have products like “Premier LV132062W Supabrights” and “Premier LV081168W Supabrights”. Both products would be in all three campaigns, with the keywords LV132062W and LV081168W excluded in Brand and Rest. You don’t have to worry that Google will show the wrong product for one of those terms.

      (note: check first, if people are actually looking for something like this AND if you can get these parts – if one of those isn’t true, skip the product campaign).


      • Joe Wilde

        Brilliant, thank you for the clarification! We set up Brand and Rest campaigns last week when i originally commented and they’ve been performing really well – thank you!

  • Petter Norlen

    Hi Martin,

    Thanks for sharing this strategy!
    I´ve been working with this one for a time and it works really good and nowadays i pay the right price for the right click which is really nice to see.
    On the Brand and everything else campaigns i take the product title as negatives and that helps me save a lot time and its working good, the problem when i do this is when i have customer with 10 000+ products because its limited to 10 000 negatives for a campaign. Is that any good workaround to this little problem?

    Thanks again!

    • Martin Röttgerding

      Hi Petter,
      Yes, this is a common problem. There are several things you can do. In ascending order of effort:
      – Make room for more negatives: Put them at adgroup level. There you’ve got a limit of 20,000 criteria. The downside is that this includes not only negatives, but also positive criteria, including product groups.
      – Make room for more negatives: Split negatives across campaign level, ad group level, and negativ keyword lists (5,000 negatives per list).
      – Make room for more negatives: Split your product groups across adgroups and only use the negatives associcated with each ad group.
      – Reduce the number of negatives: Use only negatives for products that actually receive impressions.
      – Reduce the number of negatives: Use only negatives if there are corresponding queries (i. e. don’t use “xy123” as a negative if there’s no query containing “xy123”).

      One of those should do it 🙂


      • Petter Norlen

        Thanks for your help! 🙂
        Is it any big changes in how you would do this strategy today compared to what you did 2 years ago and what you talk about in the presentation?

        • Martin Röttgerding

          I wouldn’t say big changes, no. The core still stands, but we’ve refined many things and dealt with many issues that come up when you do this at a larger scale – that’s where the elaborate answer to your question stems from.

          It’s worth mentioning that while the segmentation into product terms, brands and everything else is universal, there are different ways to use the segmentation technique. For example, if you only sell wooden furniture you might want to bid higher queries that contain a reference to wood.

          • Petter Norlen

            Ok, i see. Yes i have noticed that i have to change a little bit on the strategy when i work with some clients but it works good. Thanks for answer my questions, It surely is a great strategy to improve Google Shopping!


  • Andrea Lombardi

    Thank you very much for your presentation! I’ve been implementing shopping campaigns like suggested by you.
    I have createde:
    -rest campaign
    -brand campaign
    -brand campaign with discounts products

    So the 3rd campaign has got different products from the other 2 and it is dynamic due to commercial proposals form the client.
    How can i manage the priorities in this specific case? I gave HiGH to brand campaigns with discounts products, MEDIUM to brand campaign, LOW to rest campaign.

    I have also added brand terms as negative for rest campaign and generic terms as negative for brand and brand with discounts campaign.

    • Martin Röttgerding

      Hi Andrea,
      Sorry for taking so long. If the discount campaign has different products, it falls out of the query segmentation. You can only segment queries if the campaigns have the same products.

      As for the other two, the rest campaign needs to have the highest priority. Otherwise it won’t catch any queries.

      • Andrea Lombardi

        Hi Martin,
        thank you for the feedback.

        Yes the discount campaign has different products. In this case I use only specific product + brand names (generic terms as negatives).

        Actually the rest campaign is set with the lowest priority and catch queries because it doesnt’ trigger brands but only generic terms. I keep this campaign with a very low bid (0,05 for mobile for example) and it gave me good results in terms of CPA (40% lower than other shopping campaigns), however the conversion volume is lower than others.

  • Tilmann Zechlin

    Hi Martin,

    I was looking around for a script that allows me to define some POSITIVE BROAD MATCH keywords for a Shopping Campaign and found: none, at least none that works.

    Now watching your video, I realized that your system is doing exactly that!

    The negative Broad Match Keywords in the Higher Priority Campaigns function as positive BM Keywords in the Lower Priority Campaigns. Problem solved! Thank you again.