> ## Content Index > Fetch the complete content index at: https://www.livain.com/llms.txt > Use this file to discover other available public pages before exploring further. # A counted conversion with a made-up value is worse than no tracking: the flat €10 add-to-cart that flattered four markets' CPA - URL: https://www.livain.com/blog/counted-conversion-made-up-value-flat-add-to-cart-placeholder/ - Published: 2026-10-12T07:00:00.000Z - Updated: 2026-10-12T06:59:59.000Z - Description: A soft add-to-cart event carrying a flat €10 value, set as a primary conversion action, had been training Smart Bidding for weeks. Fake tracking is more dangerous than no tracking — here is the order in which to fix it. - Author: Remco Livain - Tags: Google Ads, Conversion Tracking, Attribution, Marketing Tech, Ecommerce In four of the international Search campaigns I inherited on a home & living account, the thing being optimised for wasn't a sale. It was an add-to-cart event carrying a flat €10 per fire, sitting in the account as a **primary** conversion action, right next to the real purchase event. The purchase event barely fired in those campaigns at all. The add-to-cart one fired constantly. So the Conversions column was full, cost-per-conversion looked respectable, and the campaigns had been quietly bidding toward a number somebody typed into a box months earlier. That is a worse position than having no conversion tracking at all. With no tracking, nobody trusts the column, and you go and look at the order book. With a counted conversion carrying a made-up value, everyone trusts the column, the bidding algorithm trusts it harder, and the account spends weeks getting better and better at producing the wrong event. ### Why the placeholder is so convincing Google is explicit about what the two tiers of conversion action do. Primary actions are the ones [reported in the Conversions column and used for bidding](https://support.google.com/google-ads/answer/11461796?ref=livain.com); secondary actions are for observation only and don't influence bidding. There is no middle setting where an event counts for reporting but not for Smart Bidding. If it's primary, it's in the column and it's in the bid. Which means a soft action promoted to primary doesn't just add noise. It dilutes. A purchase worth a few hundred euros and a cart-add worth €10 land in the same column, get averaged into the same cost-per-conversion, and the campaign that produces thirty cheap cart-adds outranks the one that produced an actual order. The CPA "improvement" is a counting artefact. Nothing got cheaper; the denominator got easier. Google's own [guidance on conversion values](https://support.google.com/google-ads/answer/14791574?ref=livain.com) is more careful than most accounts are. A static value per action is fine as a way of telling actions apart — it suggests measuring at least two distinct values across at least two actions, neither of them zero. What it does not suggest is giving a soft action a round number and promoting it to primary alongside revenue. At that point you aren't differentiating anything. You've told the system a cart-add is worth €10 and asked it to go find more of them. > No tracking makes people sceptical. Fake tracking makes them confident. Confidence is the expensive one. ### The two findings you must not merge This is where I've seen the diagnosis go wrong, including in my own first read. Discovering the measurement defect feels like discovering the answer, and it isn't. There are two separate findings sitting on top of each other, and they need separate verdicts. One: the bidding has been mis-trained, so every reported figure from those campaigns is unreadable and should never have been compared to a revenue target. Two: the underlying performance may still be bad. On that account it was. Purchase tracking *worked* on the same country shops — the brand campaigns in those markets recorded real revenue perfectly well. So the near-zero purchase count on international generic traffic wasn't a measurement artefact at all. It was the truth, wearing a measurement artefact as a disguise. If you merge the two, you get the comfortable conclusion: "the numbers were broken, fix the tracking, performance is probably fine." I've watched that conclusion buy a campaign another two months of budget. The honest version is that the tracking defect means you *don't know yet*, and the one market where tracking demonstrably works is the control group that tells you whether the zeros elsewhere are real. I wrote about that distinction when a [ROAS figure doubled without a single extra sale](https://www.livain.com/blog/our-roas-doubled-we-hadnt-sold-more/), and it's the same shape here — the metric moved, the business didn't. ### What I actually change, in order Demote the soft action to secondary first. Not delete — it's a genuinely useful diagnostic, and cart-adds are the earliest readable signal on a long consideration cycle. Just get it out of the column that drives bids. Then make the purchase action primary per market, and pass the real order value instead of the placeholder. A long-consideration product with a wide price ladder cannot be optimised against one flat number; the system needs to know that some conversions are worth fifteen times others, because that's the only way it can learn which queries are worth paying for. Then — and this is the part people skip — restate every historical figure. Any ROAS or CPA you quoted from those campaigns while the placeholder was primary is not a revenue figure and never was. If a deck, a board update or a weekly report compared it to a revenue target, that comparison has to be withdrawn, not quietly dropped. Half of the credibility of a marketing function is being the person who says "the number I gave you last month was measuring something else." And only then do you judge the campaigns. Which is the right order, because the pauses and budget shifts you make afterwards are defensible — they rest on an event that corresponds to money. I've made the opposite mistake, [scaling budget while attribution was broken](https://www.livain.com/blog/scaling-google-ads-budget-while-attribution-is-broken/), and the only thing that scales cleanly in that state is the error. ### The pattern underneath Harvard Business Review has a good name for this: [surrogation](https://hbr.org/2019/09/dont-let-metrics-undermine-your-business?ref=livain.com) — the way a metric quietly replaces the strategy it was meant to represent. Nobody in that account decided to optimise for cart-adds. Somebody set up a conversion action during an implementation, needed a value, and typed a round number. Then Smart Bidding did exactly what it was told, for months, with enthusiasm. That's the operator's job in an AI-heavy stack, and it hasn't changed in twenty years of doing this: the machine will optimise whatever you point it at, far faster than you can notice you pointed it at the wrong thing. The leverage isn't in the bidding strategy. It's in knowing which column is lying — and the same discipline applies when the number is a [dashboard counting enquiry value as revenue](https://www.livain.com/blog/client-performance-dashboard-enquiry-value-not-revenue/). A tracking setup that reports nothing is a gap. A tracking setup that reports something plausible is a liability. ### Sources & further reading **External** [About primary and secondary conversion actions](https://support.google.com/google-ads/answer/11461796?ref=livain.com) — Google Ads Help [Conversion values best practices](https://support.google.com/google-ads/answer/14791574?ref=livain.com) — Google Ads Help [Don't Let Metrics Undermine Your Business](https://hbr.org/2019/09/dont-let-metrics-undermine-your-business?ref=livain.com) — Michael Harris & Bill Tayler, Harvard Business Review **Related posts** [Our ROAS doubled. We hadn't sold more.](https://www.livain.com/blog/our-roas-doubled-we-hadnt-sold-more/) [Scaling Google Ads budget while attribution is broken](https://www.livain.com/blog/scaling-google-ads-budget-while-attribution-is-broken/) [A client dashboard that counted enquiry value as revenue](https://www.livain.com/blog/client-performance-dashboard-enquiry-value-not-revenue/) [A price-list download is not a lead](https://www.livain.com/blog/price-list-download-is-not-a-lead-long-lead-time-attribution/)