Two of the most useful Smart Bidding controls in Google Ads are also two of the least used: seasonality adjustments and data exclusions. Both live under the same Bid Strategies settings, both temporarily override how the algorithm reads your conversion data, and both are aimed at exactly the situations that wreck a low-volume B2B SaaS account — a launch spike the model cannot anticipate, and a tracking outage the model should never have learned from. Used correctly they protect your bidding from distortion. Used carelessly, or ignored, they let a single bad week set your bids wrong for a month.
This guide explains what each control does, the narrow situations each is built for, and why B2B SaaS accounts specifically — with their thin conversion volume and fragile, multi-system tracking stacks — need both in their toolkit. The short version: a seasonality adjustment is a forward-looking warning about a brief, predictable change in conversion rate, and a data exclusion is a retroactive instruction to forget a period when your conversion data was simply wrong. Getting them right starts with knowing which problem you actually have.
Seasonality adjustments: priming the model for a known spike
A seasonality adjustment tells Smart Bidding to expect a short-term change in your conversion rate that it has no way to predict from history. You specify a date range and the percentage conversion-rate change you anticipate, and the algorithm bids more aggressively into that window, then reverts cleanly once it closes without treating the spike as a new baseline. The classic use cases are a two-day product launch, a limited-time promotion, or a brief demand surge tied to an external event — moments where you know conversion rate will jump but the model, left alone, would react too slowly on the way up and over-correct on the way down.
The critical constraint is duration. Google intends seasonality adjustments for events shorter than about seven days, because Smart Bidding already models recurring seasonal patterns and longer trends using its own historical data. Apply a large adjustment across several weeks and you are fighting the algorithm rather than helping it, which usually produces over-bidding followed by a hangover of pulled-back bids. For B2B SaaS, the honest use cases are narrow: a launch day, an end-of-quarter push when buying committees rush to close, or a conference-driven spike. Outside those, resist the urge to micromanage — if you are unsure whether an event qualifies, it probably does not. For the strategy layer underneath this, see our guide to Google Ads bidding strategies for B2B SaaS.
Data exclusions: making the model forget bad data
A data exclusion is the opposite maneuver: it tells Smart Bidding to ignore conversion data from a past date range because that data does not reflect reality. The overwhelming reason to use one is a tracking failure. If your conversion tag stopped firing, a site deploy broke the event, your GTM container was misconfigured, or the site went down, then for those hours or days Google recorded far fewer conversions than actually happened. Left uncorrected, Smart Bidding reads that as a genuine collapse in performance and pulls bids down — often for weeks after the outage is fixed, because the algorithm is still digesting the bad stretch.
Applying a data exclusion over the affected dates removes those conversions from the bidding model so the outage stops poisoning future bids. The discipline here is strict: exclude only genuine tracking failures, never a period of real but disappointing performance. Excluding a legitimately bad week to flatter the numbers just teaches the model a fiction and produces worse bidding later. Apply the exclusion as soon as you confirm and fix the outage, cover exactly the affected range, and document why. Because SaaS conversion tracking runs through so many moving parts, outages are common and often silent — which is why you should be able to spot them fast. Our walkthrough on debugging conversion tracking with devtools and the broader conversion tracking for SaaS guide cover how to catch a break before it costs you a month of bidding.
Why thin volume makes both controls high-stakes
These controls matter more for B2B SaaS than for a high-volume e-commerce account for one structural reason: SaaS accounts run on thin conversion volume. When a campaign generates a few demos or SQLs a day rather than hundreds of purchases an hour, every data point carries outsized weight in the model. A two-day tracking outage that would be statistical noise in a retail account can meaningfully drop a SaaS campaign's learned conversion rate, and a launch spike can just as easily convince the model of a demand level that will not persist. Low volume amplifies both the distortion and the cost of not correcting it.
The second structural factor is fragility. A B2B SaaS measurement stack typically spans a marketing site, a tag manager, a product, a CRM, and offline conversion imports, and tracking tends to break at the seams — during a site redesign, a CRM migration, or a routine deploy that nobody flagged as touching analytics. Those breaks are frequently silent: the campaign keeps spending while conversions quietly stop importing. The combination of thin volume and a break-prone stack is exactly why data exclusions should be a routine part of account hygiene rather than an exotic tool, and why you need monitoring that surfaces a tracking gap in days, not at the end of the month. If your account already struggles with sparse signal, our guide to running Google Ads on thin data covers the wider playbook.
A practical workflow for using them
Start with detection, because both controls depend on you knowing what actually happened. Keep a simple monitor on daily conversion counts so an unexpected drop to zero, or a sudden implausible spike, triggers a review within a day or two. When conversions crater, first confirm whether it is a tracking failure or a real performance drop — check the tag, the recent deploys, and the CRM import — and only reach for a data exclusion once you have confirmed the data itself is wrong. Then apply the exclusion tightly over the affected dates and note the root cause so the same break does not recur unexamined.
For seasonality adjustments, work from your calendar of known events, not from reactions to daily noise. A few days before a launch or a time-boxed promotion, estimate the conversion-rate lift honestly — modest and evidence-based beats aspirational — set the adjustment for the exact window, and remove or let it expire immediately after. Avoid stacking adjustments, avoid using them for anything longer than a week, and never let either control substitute for the fundamentals. If conversion tracking is unreliable or the bid strategy is wrong for your funnel, no adjustment or exclusion will save the account; they are precision tools for a well-built stack, covered in more depth in our Google Ads optimization checklist.
Common mistakes to avoid
The most damaging mistake is using a data exclusion to hide poor performance. Excluding a genuinely weak period because you dislike the numbers does not fix anything — it feeds the model a lie and degrades every bid that follows. Exclusions are only valid for data that is factually wrong, which in practice means tracking failures. A related error is excluding too wide a range: if the outage lasted two days, do not exclude two weeks out of caution, because you will also discard good data the model needs. Match the exclusion to the actual failure window as precisely as you can establish it.
On the seasonality side, the recurring mistake is overuse. Applying adjustments to long periods, to routine week-to-week fluctuation, or to events you cannot actually predict does more harm than leaving Smart Bidding alone, because you are overriding an algorithm that already models seasonality from history. Reserve seasonality adjustments for genuine, short, predictable spikes, keep the magnitude honest, and treat both controls as occasional interventions rather than dials you touch every week. The accounts that get the most from these features are the ones that use them rarely and precisely, on top of tracking and bidding that are already sound. For how bidding targets themselves behave under change, see our note on the August 2026 bidding target update.