I recently sat down to audit a Google Ads account for an ecommerce store that sells aftermarket car head units, those DIY stereo upgrade kits you fit yourself with a few basic tools. Lovely business: around 55-60% gross margins, an average order value of $800 to $1,000, four years of trading, and roughly $40,000 AUD a month going into ads. Healthy numbers. But under the bonnet there were a handful of problems quietly capping how much this account could sell, and they are the exact same problems I see on a huge number of accounts at this stage. So let me walk you through what I found and exactly what I told them to do about it.
The Search vs Shopping Imbalance
The first thing that jumped out was the revenue split. Search was pulling in around $150k while Shopping sat at just $50k. For most ecommerce accounts that is back to front, as Shopping usually does the heavy lifting. So I dug in, and the answer was simple: roughly $113k of that $150k in search revenue came from branded terms. People typing the brand name directly.
That tells you two things. The brand awareness is genuinely strong, which is brilliant. But it also means the account is leaning hard on people who already know who they are, and there is a whole world of non-branded demand being left on the table.
Impression Share Is Your Market Share
Here is the metric that made me sit up. The account's overall search impression share was around 15%. Some individual campaigns were as low as 13.68%.
I want you to internalise this: impression share is your market share.
A 15% impression share means you are showing up for roughly 15% of the searches you could be appearing in, and missing the other 85%. For context, I treat 30 to 40% as healthy, and 50 to 60% as exceptional. So this account was leaving an enormous amount of available market on the floor.
The next question is always the same: why are you losing it? Google splits this into two buckets, and the fix is completely different depending on which one is hurting you.
Lost to Budget vs Lost to Rank
When I broke down the impression share loss, only a small slice (somewhere around 5 to 11%) was lost due to budget. The majority was lost due to rank.
This distinction is the whole game, so let me give you the playbook.
Lost due to budget is the easy one. This is the quick win I love. If you are losing, say, 11.85% of impression share to budget, you simply raise your daily budget by roughly that same percentage. But before you touch anything, check two things. One, your bid strategy is Target ROAS. Two, your current ROAS is meeting or beating your target. If both of those are true, then raising the budget just buys you more of the same efficient volume. More sales at the same ROAS. It really is that clean.
Lost due to rank is the trade-off. This one means your bids are too low to compete; you are not paying enough to win the auction. Fixing it means raising your CPC bids, and raising bids will pull your ROAS down. So I only recommend this when your ROAS is sitting well above target and you have room to spend it. One campaign here was running at a 13x ROAS, which is a luxury problem; it can easily afford to chase a lower ROAS target in exchange for grabbing far more market share. If you are on Target ROAS, you lower the target to let Google bid higher. If you are on manual CPC, you raise the bids directly.
Sort Out Your Bid Strategies
Speaking of bidding, one campaign was still running on manual CPC, and I am not a fan of that for ongoing campaigns. Here is the approach I gave them, and it is the approach I would give you for any new campaign.
Start new campaigns on Maximise Clicks. Let them gather data until you hit somewhere between 30 and 50 conversions. Only then switch over to Target ROAS. That warm-up matters because smart bidding needs conversion data to learn from. Throw it on too early and it is flying blind. And once a campaign is established, leave manual CPC behind.
Performance Max Lacks Transparency
This client had already scaled back their Performance Max spend, and honestly, I understood why. They found it was quietly handing budget to placements that did not work, and the killer issue is that you cannot see the split between prospecting and retargeting inside PMax. You are essentially trusting a black box with your money and hoping it behaves.
So rather than fight the black box, I pulled the Channel Performance Report to see which placements were actually earning their keep. The results were really interesting:
Google Discover: 5.8x ROAS, a solid performer
Gmail: 8.5x ROAS, the standout
YouTube (non-product ads): 5.4x ROAS, very viable
Display: 1x ROAS, dragging the average down
The detail I found most telling was this: the ads that were NOT using the product feed were outperforming the product-feed ads in Discover and Gmail. The plain image and text creative beat the Shopping-style product ads in those placements.
Rebuild It as Demand Gen for Control
My recommendation was to step out of Performance Max for these placements and rebuild them as dedicated Demand Gen campaigns, where you actually hold the reins. This was my favourite part of the whole audit, because it turns a black box into something you can steer.
Here is the structure I laid out:
Demand Gen Image, Prospecting (for Gmail and Discover)
Demand Gen Image, Retargeting
Demand Gen Video, Prospecting (for YouTube)
Demand Gen Video, Retargeting
And the critical settings that make or break it: disable the Display network inside Demand Gen, because Display was the 1x dog. Exclude your retargeting audiences from your prospecting campaigns so you are not paying prospecting prices to reach people who already know you. And use square image and video assets, since those tend to perform best in Discover and Gmail.
The beauty of this is you finally separate prospecting from retargeting, so you can budget each one properly and actually measure what is working.
Conversion Delay Is Skewing Your Data
There was one more trap on this account, and it is one that catches out almost every high-ticket store. Only 33% of conversions happened on the same day as the click. The other 67% were time-delayed, anywhere from 1 to 12-plus days, with 11.7% taking more than 12 days to come through.
That makes complete sense for an $800 to $1,000 considered purchase: people mull it over. But it wrecks your data analysis if you are not careful. If you look at yesterday or the last 7 days, you are seeing an account that under-reports by roughly 30%, because a third of the sales simply have not landed yet. You will think a campaign has fallen off a cliff when really the conversions are still in the post.
So a few rules. Make sure you have at least 50 conversions before you trust ROAS data on anything. Account for the lag by avoiding very recent date ranges, or at least consciously excluding the most recent week. And use the bid strategy report to spot conversions that are coming but not yet reported.
Keep Bidding on Your Brand
Finally, I told them to keep bidding on their branded terms despite that big branded number. People sometimes assume they would win that traffic organically anyway, but organic brand click-through rate is typically only 30 to 50%, not 100%. And if you stop bidding on your own brand, your competitors will happily step in and bid on it for you. Brand traffic earns its place even if a slice of it would have converted for free.
A Word on Market Expansion
They also asked about expanding into the USA. My honest take: yes, the US is a far bigger addressable market, but it is also far more competitive. Do not assume your home ROAS will translate. The UK and Australia tend to sit at similar competition levels, but the US is its own beast. Some businesses thrive there from day one and others struggle at first, so go in with realistic expectations and a testing budget, not the whole farm.
Conclusion
This audit boiled down to a handful of fixable problems that show up on accounts everywhere. Impression share is market share, and 15% means you are missing 85% of your opportunities. Diagnose whether you are losing it to budget (an easy win, just raise the budget when ROAS is hitting target) or to rank (a deliberate trade-off, only worth it when ROAS is well above target). Warm new campaigns up on Maximise Clicks before switching to Target ROAS, and leave manual CPC behind. Performance Max hides your prospecting and retargeting split, so use the Channel Performance Report to find your winners (Gmail and Discover shone, Display flopped, and non-feed ads beat feed ads), then rebuild them as separate Demand Gen Image and Video campaigns for proper control. Respect conversion lag on high-ticket items by using wider date ranges and a 50-conversion minimum. And keep bidding on your brand, because organic only captures 30 to 50% of those clicks and competitors will take the rest.